GC, 1st chamber extended composition, September 18, 2024, No T-671/19

THE GENERAL COURT (First Chamber, Extended Composition),

1 By its action under Article 263 TFEU, the applicant, Qualcomm Inc., seeks annulment of Commission Decision C(2019) 5361 final of 18 July 2019 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.39711 – Qualcomm (predation)) (‘the contested decision’) or, failing that, the cancellation or reduction of the fine imposed on it in that decision.

 Background to the dispute

 Context of the case

2 Qualcomm is a US company established in 1985 operating in the field of cellular and wireless technologies. It conducts its business primarily through Qualcomm CDMA Technologies (‘QCT’) and Qualcomm Technology Licensing. QCT is a developer and supplier of, amongst others, baseband chipsets, a type of semiconductor, and system software based on Code Division Multiple Access (CDMA), Orthogonal Frequency-Division Multiple Access (OFDMA) and other technologies primarily used in voice and data communications. Qualcomm’s chipsets are sold (and its system software is licensed) to companies which use them in mobile phones, tablets, laptops, data modules and other consumer electronics. Qualcomm Technology Licensing operates Qualcomm’s intellectual property licensing programme, licensing primarily mobile device suppliers.

3 Mobile devices such as mobile phones, tablets and other connected devices require mobile broadband connectivity to the internet through cellular mobile telecommunications networks.

4 The core component providing mobile connectivity in a device is the baseband processor, which provides the signal processing functionality according to communication protocols described by the cellular standards. Baseband processors can be embedded directly in mobile devices, such as smartphones, or in external modules which are in turn plugged into a device. Such a processor is made of semiconductor material (such as silicone die) and is packaged into a chip called a ‘baseband chip’.

5 In addition to the baseband processor, certain types of mobile devices require an application processor, used for running the operating system and applications (such as messaging, internet browsing, imaging and games). That application processor can be provided as a standalone product, packaged into a separate chip, or can be integrated with the baseband processor within the same chip. Baseband chips can, accordingly, be divided into:

– standalone baseband chips (also referred to as ‘slim baseband chipsets’ or ‘slim modems’), where no application processor is included;

– integrated baseband chips, where the baseband processor has been integrated with the application processor.

6 Regardless of the presence of an application processor, a baseband processor is typically paired with two additional components to complete its functionality, namely the Radio Frequency integrated circuit, also known as the ‘RF transceiver’, and the Power Management (PM) integrated circuit. All three functionalities (baseband processor, RF transceiver and PM circuitry) are necessary for mobile connectivity and are usually purchased from the same supplier, either as a bundle or separately.

7 Baseband chips are typically sold to original equipment manufacturers (OEMs), which incorporate them into devices using mobile connectivity, such as Apple, HTC Corporation, Huawei Technologies Co. Ltd (‘Huawei’), LG Corp., Nokia Corporation, Samsung Group (‘Samsung’) and ZTE Corporation.

8 During the period from 2009 to 2011, mobile devices incorporating baseband chipsets could be grouped into two broad categories. First, mobile phones (of different types, ranging from those providing only basic functionality, such as voice services, to smartphones) and, second, MBB devices, namely devices providing connectivity other than mobile phones, usually without voice services (for example, tablets, data cards such as USB sticks with cellular access, wireless routers working on MiFi or laptops).

9 Many MBB devices, such as data cards, have tended to use mainly slim modems, because they did not require any processing functionality, but only connectivity, and the market for those devices, in particular the market for devices compatible with the third generation (3G) communication standards based on Universal Mobile Telecommunications System (UMTS) technology, was considerably smaller than that of mobile phones.

10 As stated in paragraph 4 above, in order to provide connectivity, a baseband chip must use one of the communication standards. Originally, the first generation communication standards (1G), namely analogue communication standards, and the second generation (2G) communication standards which replaced them, namely digital communication standards, provided only voice communications. Over time, the standard developed for the 2G communication standards, namely the ‘Global System for Mobile Communication’ (GSM), was expanded to support higher speeds and packet data transport, via ‘General Packet Radio Services’ (GPRS) and ‘Enhanced Data rates for GSM Evolution’ (EDGE).

11 The 3G communication standards, based on UMTS technology, namely a wireless and mobile communication technology, enabled, in their infancy (around the year 2000), processing of data rates of up to 0.348 megabits per second (Mbps), which was insufficient to support typical broadband applications such as full internet browsing or video streaming. Subsequent developments have strengthened the data transmission capacity of those communication standards. ‘High Speed Packet Access’ (HSPA) technology made it possible to process data rates up to 14 Mbps, then ‘Evolved High Speed Packet Access’ technology (HSPA+) made it possible to process data rates up to 28 Mbps and even 42 Mbps.

12 Typically, chipsets compliant with UMTS technology (‘UMTS chipsets’) also provide support for the broad GSM/EDGE standard, which was initially optimised for voice telephony. That is because that broad standard continued to be indispensable for mobile telephones, since, for most mobile network operators, GSM continued to play an important role in terms of coverage and capacity. That broad standard may also be useful for MBB devices, although it does not provide broadband connectivity. Via its basic connectivity support, however, GSM could ensure continuity of service in case of gaps in UMTS network coverage.

13 From the end of 2008, ‘Long Term Evolution’ (LTE) technology started to emerge. The initial LTE chipsets were compliant exclusively with that technology, which limited their practical use due to the limited deployment of networks using LTE technology. Gradually, the major baseband suppliers developed chipsets supporting both UMTS and LTE technologies, with the first such products becoming commercially available in 2011 or 2012. UMTS and LTE technologies were developed in parallel to improve performance and interoperability.

 Administrative procedure

14 On 30 June 2009, Icera Inc. lodged a complaint with the European Commission against Qualcomm, subsequently replaced by a revised and updated version thereof dated 8 April 2010 (‘the complaint’), on the basis of which the Commission initiated its investigation.

15 In 2012, the intervener, Nvidia Corp., which had acquired Icera in May 2011, supplied further information, supplementing the complaint and making allegations of predatory pricing against Qualcomm.

16 Between June 2010 and July 2015, the Commission sent a number of requests for information to Qualcomm, Icera, Nvidia and other players in the baseband chip sector. In particular, it sent to Qualcomm (i) a request for information dated 7 June 2010, adopted pursuant to Article 18(1) and (2) of Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles [101 and 102 TFEU] (OJ 2003 L 1, p. 1), (ii) a request for information dated 3 November 2011, adopted pursuant to Article 18(3) of that regulation, (iii) a request for information dated 10 July 2013, adopted pursuant to Article 18(3) of that regulation, (iv) a request for information dated 13 February 2014, adopted pursuant to Article 18(1) and (2) of that regulation, (v) a request for information dated 13 October 2014, adopted pursuant to Article 18(3) of that regulation, and (vi) a joint request for information dated 14 January 2015, since it was also adopted in Case AT.40220 – Qualcomm (exclusivity payments). The final decision in that case was appealed before the General Court (judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments), T‑235/18, EU:T:2022:358).

17 On 16 July 2015, the Commission initiated, within the meaning of Article 2(1) of Commission Regulation (EC) No 773/2004 of 7 April 2004 relating to the conduct of proceedings by the Commission pursuant to Articles [101 and 102 TFEU] (OJ 2004 L 123, p. 18), proceedings against Qualcomm in Case AT.39711 – Qualcomm (predation). Those proceedings concerned alleged abuse of its dominant position by Qualcomm in the form of predatory pricing on the UMTS chipsets market.

18 On 8 December 2015, the Commission adopted, as provided for in Article 27(1) of Regulation No 1/2003 and Article 10(1) of Regulation No 773/2004, a statement of objections addressed to Qualcomm (‘the SO’).

19 Between December 2015 and July 2016, Qualcomm was granted access to the file.

20 In particular, by letter of 18 April 2016, Qualcomm applied to the Hearing Officer, in accordance with Article 3(7) and Article 7(1) of Decision 2011/695/EU of the President of the European Commission of 13 October 2011 on the function and terms of reference of the hearing officer in certain competition proceedings (OJ 2011 L 275, p. 29), for further access to certain parts of the file. Following a review by the Hearing Officer, more complete versions of the file were provided to Qualcomm.

21 On 15 August 2016, Qualcomm sent its response to the SO (‘SO response’), which disputed the Commission’s preliminary findings.

22 On 10 November 2016, an oral hearing took place at Qualcomm’s request.

23 Following that hearing, the Commission took further investigative steps and, between 2017 and 2019, sent further requests for information to Qualcomm, as well as to other players in the baseband chipset sector.

24 In particular, on 30 January 2017, the Commission sent Qualcomm a request for information under Article 18(1) and (2) of Regulation No 1/2003, to which Qualcomm did not reply. On 31 March 2017, the Commission sent Qualcomm a request for information by decision pursuant to Article 18(3) of that regulation.

25 On 13 June 2017, Qualcomm lodged an action for annulment before the General Court against the Commission’s decision of 31 March 2017. It also lodged an application under Articles 278 and 279 TFEU seeking, primarily, the suspension of that decision or, in the alternative, the adoption of interim measures in that regard. By order of 12 July 2017, Qualcomm and Qualcomm Europe v Commission (T‑371/17 R, not published, EU:T:2017:485), the President of the General Court dismissed the application for suspension and, by judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission (T‑371/17, not published, EU:T:2019:232), the General Court dismissed the application for annulment of the decision of 31 March 2017. Qualcomm’s appeal seeking the annulment of that judgment was dismissed in its entirety by the Court of Justice in its judgment of 28 January 2021, Qualcomm and Qualcomm Europe v Commission (C‑466/19 P, EU:C:2021:76).

26 On 10 November 2017, the Commission sent Qualcomm a new request for information pursuant to Article 18(1) and (2) of Regulation No 1/2003.

27 On 19 July 2018, the Commission adopted a Supplementary Statement of Objections (‘SSO’) addressed to Qualcomm, alleging the duration of the predation was more limited and using a revised methodology to compare Qualcomm’s prices and costs relative to the alleged predatory sales concerned.

28 Between 31 July and 28 September 2018, Qualcomm obtained access to the documents added to the Commission’s file after the adoption of the SO of 8 December 2015.

29 On 22 October 2018, Qualcomm sent a response to the SSO (‘the SSO response’), disputing the Commission’s preliminary findings set out in the SO and supplemented by the SSO.

30 On 10 January 2019, a second oral hearing took place at Qualcomm’s request.

31 On 5 February 2019, the Commission sent Qualcomm a request for information pursuant to Article 18(1) and (2) of Regulation No 1/2003.

32 On 22 February 2019, the Commission sent Qualcomm a letter of facts (‘LoF’) which, in its view, was intended to (i) provide Qualcomm with clarification regarding certain elements set out in the SSO with which Qualcomm had taken issue in the SSO response, (ii) inform Qualcomm of pre-existing evidence which was not expressly relied on in the SO and SSO, but which, on further analysis of the file, might have been relevant to support the preliminary conclusion reached in the SO as supplemented by the SSO, and (iii) bring to its attention certain limited updates to the price-cost test conducted in the SSO. The LoF contained, as attachments, other documents which had not previously been provided to Qualcomm.

33 On 24 March and 25 April 2019, Qualcomm submitted its comments on the LoF.

34 On 18 July 2019, the Commission adopted the contested decision.

 Content of the contested decision

 Products concerned

35 After disputing Qualcomm’s objections to alleged procedural irregularities affecting the administrative procedure, the Commission set out a detailed description of the context in which Qualcomm and its competitors, including Icera, were operating, in the area of technology and intellectual property. In particular, it stated that the products concerned by its investigation were UMTS chipsets, namely Qualcomm’s MDM8200, MDM6200 and MDM8200A chipsets, which, at the time of the alleged infringement, competed with Icera’s UMTS chips, in particular the ICE8040, ICE8042 and ICE8060 chipsets.

36 According to the Commission, all those products constituted standalone baseband chipsets supporting data connectivity at a maximum downlink data rate ranging from 7.2/14.4 Mbps up to 28 Mbps between 1 July 2009 and 30 June 2011 (‘the relevant period’). In particular, the MDM8200 chipset was Qualcomm’s first standalone baseband chip intended for MBB devices which supported HSPA+ technology, with downlink speeds of up to 28 Mbps. It was first marketed in May 2009 and moved to the End of Life commercial stage on 30 March 2011. Since 2010, it has been gradually replaced by the MDM8200A chipset, an improved version of the MDM8200 chip which, after the implementation of minor changes, could also support voice functionality. The MDM6200 chip, as for the MDM8200 and MDM8200A chips mentioned above, was mainly intended for data-only applications. It supported HSPA+ technology at downlink speeds of up to 14.4 Mbps and voice functionality without the need for modifications. Supplied in limited quantities from the second quarter of 2010, it was sold in higher quantities from 2011 and continued to be marketed at least until the end of 2017.

37 The Icera ICE8040, or Espresso-300, chip was also a standalone baseband chipset, launched in October 2008. Initially, it supported a maximum downlink speed of 10 Mbps and, on account of its characteristics, could easily benefit from performance upgrades and enhancements in the form of software updates, for example in order gradually to increase its downlink speed to 21 Mbps. The ICE8042 chipset, or Espresso-302, was an improved variant of the ICE8040 chipset, which was commercially launched in December 2009 with a downlink speed of up to 14.4 Mbps, which increased in March 2010 to 21 Mbps by means of software updates. A downgraded version of that chipset, namely the Espresso-302-1 chipset, able to deliver only a maximum downlink speed of 7.2 Mbps, was sold by Icera to ZTE. Finally, the ICE8060 or Espresso-400 chipset, announced in October 2010, benefited from Icera’s software-defined modem architecture and a downlink speed of up to 28 Mbps. A downgraded version of that chipset, namely the E-400-1 chipset, reaching a maximum downlink speed of 7.2 Mbps, was also offered.

 Relevant market

38 The Commission defined the relevant product market concerned as comprising the ‘free’ market for slim and integrated baseband chipsets, which are compliant with the UMTS standard (‘the UMTS chipset market’). It reached that conclusion by taking account, in particular, of the substitutability between UMTS chips and chips compliant with other technologies, as well as the absence of competitive constraints exerted by vertically integrated baseband chipset makers. From a geographic point of view, that market was defined as being worldwide.

 Dominant position

39 The Commission found, on the basis of the elements set out below, that Qualcomm held a dominant position on the worldwide market for UMTS chipsets, at least from 1 January 2009 to 31 December 2011.

40 First, Qualcomm had a market share of around 60% on the relevant market during the period from 1 January 2009 to 31 December 2011.

41 Second, there were a number of barriers to entry and expansion on the relevant market, such as the need to undertake significant initial investments in research and development (R&D) activities relating to the design of UMTS chipsets, as well as various barriers relating to Qualcomm’s intellectual property rights, and its grant-back network.

42 Third, the commercial strength of customers who purchased Qualcomm’s chipsets was not capable of affecting Qualcomm’s dominant position during the relevant period.

 Abuse of dominant position

43 The Commission concluded that Qualcomm had abused its dominant position by supplying, during the relevant period, certain quantities of three of its UMTS chipsets, namely the MDM8200, MDM6200 and MDM8200A chipsets, to two of its key customers, namely Huawei and ZTE, below cost prices, with the intention of eliminating Icera, its main competitor at the time in the leading-edge segment of the UMTS chipsets market.

44 According to the Commission, by limiting Icera’s growth in the leading-edge segment of the UMTS chipsets market, which was at the time made up almost exclusively of chipsets used in high-speed MBB devices, Qualcomm intended to prevent that undertaking, which was small and financially constrained, from gaining the reputation and scale necessary to challenge its dominant position on that market, in particular in view of the expected growth potential of that segment due to the global take-up of so-called ‘smart’ mobile devices. The Commission found that Qualcomm was thus seeking to deprive OEMs in that segment of an alternative source of chipsets for their mobile telephones, thereby reducing consumers’ choice.

45 The Commission identified the factors set out below as being key to its findings.

46 Qualcomm’s pricing practices took place in the context of Icera increasing its presence on the UMTS chipsets market as a viable supplier of UMTS chipsets, which posed a growing threat to Qualcomm’s business. In order to ensure that Icera’s business would not reach a critical size threatening its market position, Qualcomm took preventive measures in the form of pricing concessions targeted at two of its strategically important customers, namely Huawei and ZTE. It found that Icera’s growth prospects depended on its ability to establish commercial relations with those two undertakings. Qualcomm’s preventive actions were based on a ‘multi-chipset’ strategy covering its three chipsets which competed with Icera’s more advanced chipsets, which sought, in particular, to protect Qualcomm’s market position in the segment of high-speed chipsets intended for mobile phones, a market which Icera intended to enter after having secured its presence on the segment of chipsets intended for high-speed MBB devices.

47 According to the Commission, analysis of the prices charged by Qualcomm to Huawei and ZTE and of Qualcomm’s costs of manufacturing its chipsets demonstrated that Qualcomm had sold certain amounts of chipsets below its long-run average incremental costs (‘LRAIC’) and, in any event, below its average total costs (‘ATC’), as well as a limited amount of the MDM6200 chipsets at prices below its average variable costs (‘AVC’). The results of the price-cost test conducted were borne out by evidence in the form of internal and contemporaneous Qualcomm documents demonstrating Qualcomm’s intention to eliminate Icera.

– Lack of justification

48 The Commission concluded that Qualcomm had not provided any valid objective justification or proper defence in respect of its conduct.

– Single and continuous infringement

49 The Commission concluded that Qualcomm’s predatory sales to Huawei and ZTE, taken together, constituted a single and continuous infringement for the whole of the relevant period.

– The Commission’s jurisdiction

50 According to the Commission, it has jurisdiction to apply Article 102 TFEU and Article 54 of the Agreement on the European Economic Area (OJ 1994 L 1, p. 3; ‘the EEA Agreement’) to Qualcomm’s infringement, since that infringement was implemented and is liable to have substantial, immediate and foreseeable effects in the European Economic Area (EEA) and since it appreciably affected trade between Member States and between contracting parties to the EEA.

– Penalty

51 While Qualcomm’s infringement had come to an end at the time when the contested decision was adopted, the Commission nevertheless required Qualcomm to refrain from repeating the conduct described in that decision and from any act or conduct which would have the same or an equivalent object or effect as for that conduct.

52 The fine imposed on Qualcomm in respect of the infringement, calculated by the Commission on the basis of the principles set out in the Guidelines on the method of setting fines imposed pursuant to Article 23(2)(a) of Regulation No 1/2003 (OJ 2006, C 210/2, p. 2, ‘the 2006 Guidelines’), amounts to EUR 242 042 000.

 Procedure and forms of order sought

 Procedure

53 By application lodged at the Registry of the General Court on 1 October 2019, the applicant brought the present action.

54 The Commission made two applications for an extension of the time limit for lodging the defence, on 4 February and 24 March 2020 respectively, given the length of the application and the number of annexed documents. That extension was granted.

55 By document lodged at the Court Registry on 17 March 2020, Nvidia applied for leave to intervene in the present proceedings in support of the form of order sought by the Commission. That application to intervene was served on the main parties in accordance with Article 144(1) of the Rules of Procedure of the General Court. The main parties raised no objections to it.

56 On 11 June 2020, the Commission lodged the defence at the Court Registry.

57 On 6 July 2020, the applicant applied for an extension of the time limit for lodging the reply, given the length of the defence and the number of annexed documents. That extension was granted.

58 On 5 October 2020, the applicant lodged the reply at the Court Registry.

59 By document lodged at the Court Registry on 15 October 2020, after having obtained a number of extensions of the time limit, the applicant and the Commission requested, pursuant to Article 144(2) and (7) of the Rules of Procedure, that certain information in the application and in some of the annexes thereto be treated as confidential vis-à-vis Nvidia. Those parties lodged a common non-confidential version of those documents.

60 On 23 October 2020, the Commission applied for an extension of the time limit for lodging the rejoinder, given the length of the reply and the number of annexed documents. That extension was granted.

61 By document lodged at the Court Registry on 19 November 2020, the Commission requested that certain information in the defence be treated as confidential vis-à-vis Nvidia. On the same date, with the agreement of the applicant, the Commission lodged a common non-confidential version of the defence and its annexes.

62 By document lodged at the Court Registry on 20 November 2020, the applicant requested that certain information in the defence and its annexes and in the procedural documents sent to the Registry during the period between 30 June and 5 October 2020 be treated as confidential vis-à-vis Nvidia.

63 By order of 25 November 2020, Qualcomm v Commission (T‑671/19, not published), the President of the Fifth Chamber of the General Court ordered that Nvidia be granted leave to intervene and reserved the decision on costs.

64 By documents lodged at the Court Registry on 15 December 2020 and 18 January 2021, Nvidia raised objections to the requests for confidential treatment relating to the application, the defence and certain annexes.

65 By document lodged at the Court Registry on 16 December 2020, the Commission requested that certain information in the reply and in Annex C.8 to the reply be treated as confidential vis-à-vis Nvidia.

66 By document lodged at the Court Registry on 22 December 2020, the applicant requested that certain information in the reply and its annexes, and in the procedural documents sent to the Registry between 5 October and 9 December 2020, be treated as confidential vis-à-vis Nvidia. On the same date, the main parties lodged a common non-confidential version of the reply and its annexes.

67 On 21 January 2021 the Commission lodged the rejoinder at the Court Registry.

68 By document lodged at the Court Registry on 25 February 2021, after having obtained an extension of the time limit, the Commission requested that certain information in the rejoinder be treated as confidential vis-à-vis Nvidia. On the same date, with the agreement of the applicant, the Commission lodged a common non-confidential version of the rejoinder and its annex.

69 By document lodged at the Court Registry on 26 February 2021, after having obtained an extension of the time limit, the applicant requested that other information in the rejoinder and its annex, as well as in the procedural documents communicated to the Registry between 18 December 2020 and 10 February 2021, be treated as confidential vis-à-vis Nvidia.

70 By order of 22 July 2021, Qualcomm v Commission (T‑671/19, not published, EU:T:2021:502), the President of the Fifth Chamber granted the requests for confidential treatment relating to certain information in the application and in Annexes A.1 and A.29, as well as in the defence, and rejected similar requests relating to the other procedural documents. Therefore, the applicant and the Commission were given a time period within which to submit new non-confidential versions of certain documents in the file. On 16 and 17 September 2021, after having obtained an extension of the time limit, the main parties lodged a common non-confidential version of those documents.

71 By document lodged at the Court Registry on 19 October 2021, Nvidia lodged its statement in intervention.

72 By document lodged at the Court Registry on 19 November 2021, the Commission stated that it had no observations regarding the statement in intervention.

73 By document lodged at the Court Registry on 29 November 2021, the applicant submitted observations on the statement in intervention.

74 The written part of the procedure was closed on 29 November 2021.

75 On 8 February 2022, the applicant made a request for a hearing.

76 On a proposal from the Fifth Chamber, the Court decided, pursuant to Article 28 of the Rules of Procedure, to refer the case to the Fifth Chamber sitting in extended composition.

77 Following a change in the composition of the Chambers of the Court, pursuant to Article 27(5) of the Rules of Procedure, the Judge-Rapporteur was assigned to the Extended Composition of the First Chamber of the General Court, to which the present case was, therefore, allocated.

78 Acting upon a proposal of the Judge-Rapporteur, the General Court decided to open the oral part of the procedure.

79 On 2 December 2022, by way of measures of organisation of procedure, the Court requested the main parties to answer certain questions. The applicant answered those questions on 16 December 2022. The Commission, having obtained an extension of the time limit for submitting its answers, responded on 16 January 2023.

80 At the request of the main parties, the time limit for lodging requests for confidential treatment of their replies to the Court’s questions was extended. The time limit was ultimately set at 31 January 2023, the date on which common non-confidential versions of those answers were lodged.

81 By document lodged at the Court Registry on 16 February 2023, after having obtained an extension of the time limit, Nvidia raised objections to those requests for confidential treatment concerning the main parties’ answers to the questions put by the Court.

82 A report for the hearing was sent to the parties. The applicant and the Commission submitted observations on that document on 27 January 2023 and 16 February 2023 respectively. The Court took formal note of those observations.

83 By order of 8 March 2023, Qualcomm v Commission (T‑671/19, not published, EU:T:2023:125), the President of the First Chamber, Extended Composition, granted certain requests for confidential treatment relating to information in the main parties’ answers to the questions put by the Court, and rejected others. Therefore, the applicant and the Commission were given a time period within which to submit new non-confidential versions of those answers. On 10 March 2023, the main parties lodged a common non-confidential version of those documents.

84 As a member of the First Chamber, Extended Composition was unable to sit, the President of the First Chamber designated another Judge to complete the chamber.

85 The oral procedure was closed at the end of the hearing on 15 March 2023. Since a member of the Chamber was prevented from taking part in the judicial deliberations following the expiry of his mandate on 27 September 2023, the Court’s deliberations were continued by the three judges whose signatures the present judgment bears, in accordance with Article 22 and Article 24(1) of the Rules of Procedure.

 Forms of order sought

86 The applicant claims that the Court should:

– annul the contested decision;

– ‘annul, or in the alternative, reduce substantially the amount of the fine’;

– order measures of organisation of procedure or measures of inquiry to require the Commission to confirm that the redactions in certain documents in the file are indeed based on sound assertions of legal professional privilege and to inform Qualcomm accordingly or to obtain those documents in order to examine the validity of the intervener’s claims;

– order the Commission to pay the costs.

87 The Commission contends that the Court should:

– dismiss the action;

– order Qualcomm to pay the costs.

88 Nvidia claims that the Court should:

– dismiss the action;

– order Qualcomm to pay the costs.

 Law

89 As a preliminary point, as indicated in paragraphs 59, 61 and 62 above, the main parties requested that certain information in their pleadings and in other procedural documents be omitted vis-à-vis Nvidia and, therefore, vis-à-vis the public. Nvidia raised objections to the requests for confidential treatment of certain information. By orders of 22 July 2021, Qualcomm v Commission (T‑671/19, not published, EU:T:2021:502), and of 8 March 2023, Qualcomm v Commission (T‑671/19, not published, EU:T:2023:125), the Presidents of the Fifth Chamber and of the First Chamber, Extended Composition, respectively, granted certain requests for confidential treatment and rejected others.

90 When a party makes an application under Article 144(2) of the Rules of Procedure, the President is to give a decision solely on the documents and information the confidentiality of which is disputed (see, to that effect and by analogy, order of 26 January 2018, FV v Council, T‑750/16, not published, EU:T:2018:59, paragraph 14 and the case-law cited).

91 However, notwithstanding the fact that there is no challenge, the Court cannot be prevented from rejecting requests for confidential treatment in so far as they concern data the public nature of which is manifestly apparent from the information in the file or the confidential nature of which becomes manifestly obsolete as a result of the disclosure of other information in the file (order of 15 September 2016, Deutsche Telekom v Commission, T‑827/14, not published, EU:T:2016:545, paragraph 46). Accordingly, in certain circumstances, it may decide to rule on the non-contested aspects of a request for confidential treatment (see, to that effect, order of 11 April 2019, Google and Alphabet v Commission, T‑612/17, not published, EU:T:2019:250, paragraph 16).

92 It should also be recalled that, according to settled case-law, information which was confidential, but which is at least five years old, must for that reason be considered historical and be communicated to the other parties, unless, exceptionally, the party seeking to preserve that confidentiality shows that, in spite of its age, that information still constitutes essential secrets, in particular industrial or commercial secrets, disclosure of which would harm that party or the third party concerned (see order of 11 April 2019, Google and Alphabet v Commission, T‑612/17, not published, EU:T:2019:250, paragraph 19 and the case-law cited).

93 In that regard, the Court, in the context of the application of Article 66 of the Rules of Procedure, must reconcile the principle that court rulings are made public with the right to the protection of personal data and the right to the protection of professional secrecy, regard also being had to the public’s right of access, in accordance with the principles laid down in Article 15 TFEU, to court rulings (see, to that effect and by analogy, judgment of 5 October 2020, Broughton v Eurojust, T‑87/19, not published, EU:T:2020:464, paragraph 49).

94 In the present case, the Court has decided not to redact, in the non-confidential version of the judgment, certain information covered by the requests of the main parties, the confidentiality of which has not been disputed by Nvidia. Some of those data may be inferred from the content of other parts of the present judgment and are therefore in the public domain. Some data are historic data for which maintaining confidentiality more than a decade later has not been appropriately justified. Lastly, certain other data provide factual details relating to the conduct investigated by the Commission. To redact that information would affect the public’s understanding of the Court’s judgment (see, to that effect, judgment of 2 February 2022, Scania and Others v Commission (T‑799/17, EU:T:2022:48, paragraph 82).

95 In support of the application for annulment, the applicant puts forward 15 pleas in law:

– the first, alleging procedural irregularities;

– the second, alleging ‘manifest errors of assessment’, of fact and of law, and failure to state adequate reasons as regards the definition of the relevant market and the applicant’s dominant position during the relevant period;

– the third, alleging a failure ‘to apply the correct legal standard thereby committing manifest errors of law’;

– the fourth, alleging that the ‘“predation theory” is illogical and unsupported by evidence’;

– the fifth, alleging ‘manifest errors’ of assessment and failure to state reasons regarding the reconstruction of alleged ‘effectively paid’ prices;

– the sixth, alleging that the ‘allocation of non-recurring engineering expenses is manifestly incorrect’;

– the seventh, alleging failure to ‘establish an appropriate cost benchmark’;

– the eighth, alleging that the price-cost analysis is ‘manifestly incorrect’;

– the ninth, alleging ‘manifest errors of law and of appraisal’ as regards the finding that the prices charged by the applicant foreclosed Icera and caused consumer harm;

– the tenth, alleging manifest errors of assessment, of fact and of law, failure to state reasons and infringement of the right to be heard and of the principle of sound administration, in that the Commission concluded that the applicant’s pricing practices were the implementation of a plan to exclude Icera;

– the eleventh, alleging a ‘manifest error of appraisal’ of fact and of law, and failure to state reasons as regards the Commission’s rejection of the objective justification put forward by the applicant;

– the twelfth, alleging that the contested decision is not adequately reasoned;

– the thirteenth, alleging manifest errors of assessment, lack of foundation and failure to state reasons in the contested decision as regards the duration of the infringement;

– the fourteenth, alleging that the contested decision ‘manifestly errs’ in the imposition and calculation of the fine;

– the fifteenth, alleging ‘manifest errors of law, fact and assessment’, and failure to state reasons in the contested decision, in establishing the Commission’s jurisdiction and the effect on trade.

96 Those pleas will be examined below in the order followed by the applicant, with the exception of the third, fourth and eighth pleas, which are based on, or summarise, certain arguments addressed in particular in the context of the sixth, seventh and ninth to eleventh pleas and which will, therefore, be dealt with after the eleventh plea.

 The first plea: procedural irregularities

97 The first plea comprises two parts. The first alleges breach of the principle of sound administration in that the Commission failed to conduct a thorough, objective and diligent investigation. The second alleges breach of the rights of the defence and of the principle of equality of arms, in that the Commission failed to disclose to the applicant evidence relevant to its defence.

 The first part: breach of the principle of sound administration

98 The first part comprises three complaints. The first alleges that the duration of the investigation was excessive. The second alleges that the case file was not sufficiently complete and accurate. The third alleges that the investigation was biased.

– Preliminary observations

99 Article 41(1) of the Charter of Fundamental Rights of the European Union (‘the Charter’), enshrines the right to good administration and provides that every person has the right to have his or her affairs handled impartially, fairly and within a reasonable time by the institutions, bodies, offices and agencies of the European Union. The Explanations relating to the Charter, published in the Official Journal of the European Union of 14 December 2007 (OJ 2007 C 303, p. 17), state that that article is based on the existence of the European Union as subject to the rule of law whose characteristics were developed in the case-law which enshrined good administration as a general principle of law (judgment of 13 December 2018, Transavia Airlines v Commission, T‑591/15, EU:T:2018:946, paragraph 37 (not published)).

100 According to the case-law on the principle of sound administration, where, as in the present case, the institutions of the European Union have a power of appraisal, respect for rights guaranteed by the European Union legal order in administrative procedures is of even more fundamental importance. Those guarantees include, in particular, the duty of the competent institution to examine carefully and impartially all the relevant aspects of the individual case (judgments of 21 November 1991, Technische Universität München, C‑269/90, EU:C:1991:438, paragraph 14; of 13 December 2018, Transavia Airlines v Commission, T‑591/15, EU:T:2018:946, paragraph 38 (not published); and order of 17 January 2022, Car-Master 2 v Commission, T‑743/20, not published, EU:T:2022:33, paragraph 66).

– The first complaint: the duration of the investigation was excessive

101 The applicant submits that the 10-year duration of the investigation is excessive and demonstrates, in particular, a lack of diligence on the part of the Commission.

102 According to the applicant, owing to the duration of the investigation, it was not in a position to defend itself properly. In particular, it raises changes in the make-up of the Commission team responsible for the investigation, a request for information sent eight years after the complaint was lodged, requests for clarification concerning documents which the Commission had had in its possession for a number of years and the modification of the scope of the investigation at a very advanced stage. It also states that the passing of time has affected recollection of the facts, both on its part and on the part of Huawei and ZTE, which, for that reason, were not in a position to respond to certain, sometimes crucial, information requests by the Commission. Lastly, it denies that the complexity of the case could have justified that duration and adds that it has always fully cooperated with the Commission.

103 The Commission disputes the arguments put forward by the applicant.

104 In accordance with settled case-law, compliance with the reasonable time requirement in the conduct of administrative procedures relating to competition policy constitutes a general principle of EU law whose observance the EU Courts ensure (see judgment of 19 December 2012, Heineken Nederland and Heineken v Commission, C‑452/11 P, not published, EU:C:2012:829, paragraph 97 and the case-law cited).

105 Whether the time taken for each procedural stage is reasonable must be assessed in relation to the individual circumstances of each case, and in particular its context, the conduct of the parties during the procedure, what is at stake for the various undertakings concerned and its complexity (judgment of 20 April 1999, Limburgse Vinyl Maatschappij and Others v Commission, T‑305/94 to T‑307/94, T‑313/94 to T‑316/94, T‑318/94, T‑325/94, T‑328/94, T‑329/94 and T‑335/94, EU:T:1999:80, paragraph 126).

106 In addition, in competition policy matters, the administrative procedure before the Commission may involve an examination in two successive stages, each corresponding to its own internal logic. The first stage, covering the period up to notification of the statement of objections, begins on the date on which the Commission, exercising the powers conferred on it by the EU legislature, takes measures which imply an accusation of an infringement and must enable the Commission to adopt a position on the course which the procedure is to follow. The second stage covers the period from notification of the statement of objections to adoption of the final decision. It must enable the Commission to reach a final decision on the infringement concerned (judgment of 21 September 2006, Nederlandse Federatieve Vereniging voor de Groothandel op Elektrotechnisch Gebied v Commission, C‑105/04 P, EU:C:2006:592, paragraph 38).

107 Furthermore, where the breach of the principle that cases must be dealt with within a reasonable time has affected the outcome of the proceedings, such a breach may be such as to result in the contested decision being set aside (see, to that effect, judgment of 21 September 2006, Technische Unie v Commission, C‑113/04 P, EU:C:2006:593, paragraph 48 and the case-law cited).

108 It should however be stated that, for the purposes of the application of the competition rules, the fact that a reasonable time is exceeded can constitute a ground for annulment only in cases of decisions finding infringements where it has been proved that the breach of the principle that cases must be dealt with within a reasonable time has adversely affected the rights of defence of the undertakings concerned. Save in that specific case, failure to comply with the obligation to adopt a decision within a reasonable time cannot affect the validity of the administrative procedure under Regulation No 1/2003 (see, to that effect, judgment of 21 September 2006, Nederlandse Federatieve Vereniging voor de Groothandel op Elektrotechnisch Gebied v Commission, C‑105/04 P, EU:C:2006:592, paragraphs 42 and 43).

109 In the present case, it must be observed that a period of more than six years elapsed between the complaint being lodged and the SO being sent. However, as stated in paragraph 15 above, it was not until mid 2012, that is to say three years after the complaint was lodged, that the complainant made its first allegations of predatory pricing and that the Commission was able to initiate its investigation into the conduct at issue. It follows that the first phase of the administrative procedure lasted for more than six years since the complaint was lodged, but only a little over three years since the complainant’s first allegations of predation.

110 The second phase of the administrative procedure, lasting from receipt of the SO to the adoption of the contested decision on 18 July 2019, lasted approximately three and a half years.

111 Taken as a whole since the time of the first allegations of predation, the duration of the investigation, namely approximately seven years, is, however, not excessive in the light of the specific circumstances of the case, and, in particular, its complexity.

112 As the Court stated when ruling on the action for annulment brought against the Commission’s decision of 31 March 2017, the practice alleged in the present case required complex analyses of large amounts of data, most of which data could be accessed only by the applicant, in order to re-construct the cost-price structure for the purpose of ascertaining whether or not pricing was predatory. Such an exercise was, in addition, all the more complex because it related to composite products (judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission, T‑371/17, not published, EU:T:2019:232, paragraph 125).

113 First, the applicant submitted more than 31 000 documents to the Commission, as is apparent from the application, in response to eight requests for information, sometimes followed by subsequent requests for clarification. Numerous meetings and telephone calls were also organised by the Commission, with the applicant, with the complainant and with third parties. In addition, it is apparent from Section 3 of the contested decision, which sets out the various stages of the administrative procedure, that the Commission did not remain inactive at any point during the investigation. Lastly, account must also be taken of the fact that that decision contains an analysis of the conduct at issue which analysis is both complex and detailed, and that the Commission was at pains to respond to the numerous arguments raised by the applicant during that procedure, in compliance with the applicant’s rights of defence.

114 Second, the conduct of the parties during the administrative procedure had an impact on the duration of that procedure. In that regard, it should be noted that it was only three years after having lodged the complaint that, for the first time, the complainant raised allegations of predation. As for the applicant, first of all, it called on the Hearing Officer on nine occasions to resolve questions relating to access to the file, then requested a number of time limit extensions, the postponement of a hearing, and requested that an additional hearing be held. Lastly, by bringing an action for annulment of the Commission’s decision of 31 March 2017 and then an appeal against the judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission, T‑371/17, not published, EU:T:2019:232, which dismissed that action, the applicant could not have been unaware that that would necessarily slow down the investigation.

115 Consequently, since the duration of the investigation was not excessive, the present complaint must be rejected as being unfounded.

116 In any event, even if the duration of the investigation could be regarded as excessive, the applicant has not demonstrated how that could have had a negative impact on its ability to defend itself.

117 In the first place, the applicant in no way explains how simple changes within the Commission in the staff responsible for the investigation at all levels of the hierarchy could have affected the rigour, accuracy, stability and scope of the investigation or its rights of defence.

118 In the second place, as regards the document submitted in December 2013, in respect of which the Commission did not seek clarification until January 2017, it is sufficient to recall that, as the Court has held, account should also be taken of the general duty of care attaching to any undertaking or association of undertakings, by virtue of which they are required to ensure the proper maintenance, in their books or files, of information enabling details of their activities to be retrieved, in order, inter alia, to have the necessary evidence available in the event of legal or administrative proceedings. Accordingly, since the Commission had requested information from the applicant under Article 18(2) and (3) of Regulation No 1/2003, as from 7 June 2010 it was incumbent on the applicant, at least from that date, to act with greater diligence and to take all appropriate measures in order to preserve such evidence as might reasonably be available to it (see, to that effect, judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission, T‑371/17, not published, EU:T:2019:232, paragraph 136 and the case-law cited).

119 In the third place, the Commission has explained that, while the investigation had evolved over the course of the administrative procedure, that was so precisely in order to take account of the observations and arguments put forward by the applicant in response, in particular, to the SO and the SSO, which, contrary to the applicant’s allegations, simply demonstrates that the Commission fully observed its rights of defence.

120 In the fourth place, the applicant does not explain how the fact that Huawei and ZTE were unable to provide certain explanations requested by the Commission concerning the payment of non-recurring engineering expenses (‘the NRE payments’) was linked to the passage of time. In any event, it must be held that the Commission examined whether, by granting such payments to Huawei and ZTE, the applicant had intended to grant price reductions to those two customers, which is as a result a subjective element, and therefore unrelated to those two companies, which companies could not have adduced any significant exculpatory evidence in favour of the applicant. That is, moreover, confirmed by the fact that, as is apparent from the examination of the sixth plea, the Commission proved that element by relying on a body of consistent evidence which did not include testimony from those two undertakings.

121 In the light of the foregoing, even if it were to be well founded, the present complaint cannot result, in the present case, in a finding that the applicant’s rights of defence were infringed, resulting in the annulment of the contested decision.

– The second complaint: the case file was not sufficiently complete and accurate

122 The applicant submits that, by failing to gather certain potentially exculpatory information, the Commission failed to compile a complete and accurate file. In its view, that failure is illustrated by the three examples set out below. In the first place, the Commission did not ask any questions of Huawei and ZTE concerning a Qualcomm employee, despite the large number of documents authored by that employee on which the Commission relies. In the second place, Huawei and ZTE were unable to provide a meaningful answer to a request for information concerning the NRE payments, even though they were crucial. In the third place, the applicant maintains that the Commission granted it access to certain documents which had been substantially redacted, without having ascertained whether the reasons for the documents’ confidentiality, relied on by the complainant who had submitted them, were well founded, even though such documents might have contained potentially exculpatory evidence.

123 The Commission disputes the arguments put forward by the applicant.

124 In accordance with settled case-law, it is, in principle, for the Commission to decide whether a particular item of information is necessary in the context of an investigation into an infringement of the competition rules (see judgment of 15 July 2015, GEA Group v Commission, T‑45/10, not published, EU:T:2015:507, paragraph 311 and the case-law cited). In addition, in so far as, in criticising the Commission in the present part of the plea for having failed to seek to obtain information which was probably exculpatory, it also relies on the guarantee of the rights of the defence in the present part of the plea, such a guarantee does not require the Commission to carry out further investigations where it considers that the investigation of the case has been sufficient (judgments of 16 May 1984, Eisen und Metall v Commission, 9/83, EU:C:1984:177, paragraph 32, and of 11 March 1999, Thyssen Stahl v Commission, 141/94, EU:T:1999:48, paragraph 110).

125 In addition, as regards the existence of potentially exculpatory documents which the Commission has not sought to obtain, it is apparent from the case-law that the undertaking concerned must show that it would have been able to use those documents for its defence, in the sense that, had it been able to rely on them during the administrative procedure, it would have been able to put forward evidence which was not consistent with the inferences made at that stage by the Commission and therefore could have had an influence, in any way at all, on the assessments made by the Commission in its decision (see, to that effect, judgments of 1 July 2010, Knauf Gips v Commission, C‑407/08 P, EU:C:2010:389, paragraph 23 and the case-law cited, and of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 97).

126 It follows that the undertaking concerned must establish, first, that it did not have access to certain exculpatory evidence and, second, that it could have used that evidence for its defence (judgments of 1 July 2010, Knauf Gips v Commission, C‑407/08 P, EU:C:2010:389, paragraph 24, and of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 98).

127 In the present case, the applicant does not explain why the Commission erred in concluding that the investigation of the case had been sufficient and how the potentially exculpatory information which the Commission had failed to gather, including the three examples which it cites, could have been used for its defence in the sense that if it had been able to rely on it during the administrative procedure, it could in any way have influenced the assessments made by the Commission in the contested decision. In any event, as is apparent from the case-law cited in paragraph 124 above, the Commission is not required to continue the investigation in order to gather all potentially exculpatory evidence, where it considers that the investigation of a case has been sufficient.

128 In particular, as regards the employee who is the author of numerous Qualcomm documents on which the Commission relies in the contested decision, the applicant provides no explanation at all as to how the answers to any questions to third parties on that topic by the Commission could have been of any importance for the investigation of the case or could have been used in any way to ensure its defence or could have had any influence on the assessments made by the Commission. While the Commission relied on some of those documents, principally in order to demonstrate that there was a plan to eliminate Icera, it also relied on other decisive evidence, which it concluded was sufficient. Lastly, the applicant does not explain how a third party could have been better placed than it to provide evidence challenging the role of that employee or the Commission’s interpretation of the documents in question.

129 Similarly, even if Huawei and ZTE were unable to provide a meaningful answer to a request for information concerning the NRE payments, even though they were crucial, it must be observed that the Commission did not need those answers, since it relied in that regard on a body of consistent evidence, as is apparent from the examination of the sixth plea. The Commission was therefore entitled to conclude, without making a manifest error, that the investigation of the case was sufficient.

130 Lastly, as regards its access to documents which had been substantially redacted and which might potentially have contained exculpatory evidence, the applicant does not dispute that the complainant, which provided that information, considered it to be subject to legal professional privilege, and that the Commission itself did not possess unredacted versions. In any event, the Commission was, reasonably, able to conclude that it had sufficient other evidence for the carrying out of the investigation, without having to ask the complainant to provide it with versions of the documents it had submitted bearing fewer redactions.

131 For those reasons, the applicant’s request that the Commission be ordered to confirm whether the numerous redactions in documents ID 1112-00146, 1112-00148, 1112-00150, 1112-00151, 1112-00154, 1112-00185, 1112-00218, 1112-00196, 1112-00229 and 1294 are based on sound assertions of legal professional privilege and to inform Qualcomm thereof or to obtain such presentations in order to examine the validity of Nvidia’s assertions must also be rejected.

132 In the light of the foregoing, the present complaint must be rejected.

– The third complaint: the investigation was biased

133 The applicant submits that the contested decision is the result of a biased investigation, which resulted in a breach of the principle of sound administration and, to a lesser extent, of the principles of the presumption of innocence, in dubio pro reo, legal certainty and equality of arms and which infringes its rights of defence.

134 In support of the present complaint, the applicant puts forward three arguments.

135 In the first place, the Commission failed to assess impartially the arguments and evidence put forward by the applicant in the SO response and at the hearing. Rather, it engaged in a prolonged ‘fishing expedition’ by issuing multiple requests for information which resulted in vast amounts of data being gathered. Furthermore, the contested decision contained material differences and new elements compared with the SSO and the LoF, which, in turn, differed from the SO. In that regard, the applicant submitted a list of examples of such differences in Annex A.11 to the application.

136 In the second place, in the contested decision the Commission disregarded certain exculpatory evidence obtained from Huawei, in particular as regards NRE payments.

137 In the third place, the Commission met the complainant to discuss issues related to the SSO hearing, without the Hearing Officer being present, even though that matter falls under the mandate of the Hearing Officer, which raises serious questions as to the integrity and neutrality of the Commission’s investigation.

138 The Commission disputes the arguments put forward by the applicant.

139 In that regard, it should be noted that every person has the right to have his or her affairs handled impartially by the institutions of the European Union (see judgment of 2 February 2022, Scania and Others v Commission, T‑799/17, EU:T:2022:48, paragraph 145 and the case-law cited). In the present case, however, it must be stated that none of the arguments put forward by the applicant makes it possible to establish that the Commission failed to offer all the guarantees to exclude any legitimate doubt regarding its impartiality in the investigation.

140 As regards, first of all, the applicant’s argument based on the fact that the Commission issued numerous requests for information which resulted in vast amounts of data being gathered, it should be recalled that, according to recital 23 of Regulation No 1/2003, the Commission should be empowered throughout the European Union to require such information to be supplied as is necessary, in particular, to detect any abuse of a dominant position prohibited by Article 102 TFEU. It is also apparent from Article 18(1) of Regulation No 1/2003 that, in order to carry out the duties assigned to it by that regulation, the Commission may, by simple request or by decision, require undertakings and associations of undertakings to provide ‘all necessary information’.

141 Given the Commission’s broad powers of investigation and assessment, it falls to it to assess whether the information which it requests from the undertakings concerned is necessary. As regards the Court’s power of review over that assessment by the Commission, it should be recalled that, according to the case-law, the concept of ‘necessary information’ must be interpreted by reference to the objectives for the achievement of which the powers of investigation in question have been conferred upon the Commission. Thus, the requirement that a correlation must exist between the request for information and the presumed infringement will be satisfied as long as, at that stage in the procedure, the request may legitimately be regarded as having a connection with the presumed infringement, in the sense that the Commission may reasonably suppose that the information will help it to determine whether the alleged infringement has taken place (see judgment of 14 March 2014, Holcim (Deutschland) and Holcim v Commission, T‑293/11, not published, EU:T:2014:127, paragraph 110 and the case-law cited).

142 As regards the present case in particular, it is apparent from paragraph 128 of the judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission (T‑371/17, not published, EU:T:2019:232), following the applicant having brought an action for annulment of the Commission’s decision of 31 March 2017, that the scope of the Commission’s investigation was such as to justify the provision of a significant amount of information.

143 In addition, as the Court also pointed out in paragraph 201 of the judgment of 9 April 2019, Qualcomm and Qualcomm Europe v Commission (T‑371/17, not published, EU:T:2019:232), it was precisely in order to prepare its final decision on the possible existence of an infringement of Article 102 TFEU with all due care and attention and to take that decision on the basis of all the information which might have a bearing on it that the Commission was entitled to adopt that decision. The same reasoning applies by analogy to the other requests for information communicated by the Commission in the administrative procedure, there being no need to consider that there was any ‘fishing expedition’. First, a demonstration of the Commission’s bias during the investigation with regard to the applicant cannot be inferred in the abstract from the Commission having a margin of discretion in how that investigation is conducted and, second, the applicant has not adduced evidence capable of demonstrating in practice that those other requests for information could be explained only by such bias.

144 As regards, next, the applicant’s argument concerning the differences between the SO, the SSO, the LoF and the contested decision, it is apparent from the case-law of the Court of Justice that observance of the rights of the defence means that the undertaking concerned must have been afforded the opportunity, during the administrative procedure, to make known its views on the truth and relevance of the facts and circumstances alleged and on the documents used by the Commission to support its claim that there has been an infringement of the competition rules (see judgment of 5 December 2013, SNIA v Commission, C‑448/11 P, not published, EU:C:2013:801, paragraph 41 and the case-law cited; judgment of 25 March 2021, Deutsche Telekom v Commission, C‑152/19 P, EU:C:2021:238, paragraph 106).

145 That obligation is satisfied if the final decision does not allege that the persons concerned have committed infringements other than those referred to in the statement of objections and takes into consideration only facts on which the persons concerned have had the opportunity of stating their views (judgments of 24 May 2012, MasterCard and Others v Commission, T‑111/08, EU:T:2012:260, paragraph 266; of 18 June 2013, ICF v Commission, T‑406/08, EU:T:2013:322, paragraph 117; and of 13 December 2018, Slovak Telekom v Commission, T‑851/14, EU:T:2018:929, paragraph 180).

146 However, the essential facts on which the Commission is relying may be set out summarily in the statement of objections and the decision is not necessarily required to be a replica of the statement of objections, since that statement is a preparatory document containing assessments of fact and of law which are purely provisional in nature (judgment of 17 November 1987, British American Tobacco and Reynolds Industries v Commission, 142/84 and 156/84, EU:C:1987:490, paragraph 70; see also judgment of 5 December 2013, SNIA v Commission, C‑448/11 P, not published, EU:C:2013:801, paragraph 42 and the case-law cited; judgment of 24 May 2012, MasterCard and Others v Commission, T‑111/08, EU:T:2012:260, paragraph 267). Thus, it is permissible for the Commission to supplement the statement of objections in the light of the parties’ replies, whose arguments show that they have actually been able to exercise their rights of defence. The Commission may also, in the light of the administrative procedure, revise or supplement its arguments of fact or law in support of its objections (judgment of 9 September 2011, Alliance One International v Commission, T‑25/06, EU:T:2011:442, paragraph 181). Accordingly, until a final decision has been adopted, the Commission may, in view, in particular, of the written or oral observations of the parties, abandon some or even all of the objections initially made against them and thus alter its position in their favour or decide to add new complaints, provided that it affords the undertakings concerned the opportunity of making known their views in that respect (see judgment of 30 September 2003, Atlantic Container Line and Others v Commission, T‑191/98 and T‑212/98 to T‑214/98, EU:T:2003:245, paragraph 115 and the case-law cited).

147 In the present case, a demonstration of the Commission’s bias during the investigation with regard to the applicant cannot be inferred in the abstract from the fact that there were differences between the SO, the SSO, the LoF and the contested decision. In addition, the applicant has not adduced evidence making it possible to demonstrate in practice that those differences could be explained only by such bias.

148 It is, however, necessary to consider whether, in the contested decision, the Commission relied on new complaints or evidence against the applicant, in respect of which the applicant did not have the opportunity to make known its views during the administrative procedure.

149 In that regard, the applicant simply asserts, in general terms, that the contested decision contains significant differences and new elements which are not contained in the arguments put forward in the SSO and in the LoF, and does no more than cite in a footnote certain recitals of that decision and make a reference to Annex A.11 to the application.

150 Even if those elements were in fact to constitute new elements, not contained in the previous procedural documents, the fact remains that the applicant does not explain in any way how those elements would constitute new complaints or evidence against it in relation to which it has not had the opportunity to make known its views, and not simply the taking into account, in the contested decision, of the observations received by the Commission, including its own observations.

151 As is apparent from the case-law cited in paragraph 146 above, the contested decision cannot be a replica of the SO, the SSO or the LoF, precisely because the Commission is required to take account of the observations submitted by the parties during the administrative procedure. By simply drawing up a list of differences between that decision and the earlier procedural documents, without further explaining the specific implications, to its disadvantage, of such changes to the Commission’s legal reasoning and the characterisation of the facts, or how the scope of the conduct at issue was altered accordingly, the applicant has not demonstrated that the Commission conducted a biased investigation.

152 Lastly, as regards the applicant’s argument that the Commission met the complainant to discuss issues related to the SSO hearing, without the Hearing Officer being present, even though that matter falls under the mandate of the Hearing Officer, that does not call into question the Commission’s integrity or its ability to deal with the case in a neutral and objective manner. The applicant also does not demonstrate how such a meeting could have affected its rights of defence, particularly since there was nothing to prevent it from asking the Hearing Officer for an equivalent meeting if it considered that to be necessary.

153 The present complaint must, therefore, be rejected.

154 Consequently, the first part of the first plea must be rejected.

 The second part: infringement of the rights of the defence and of the principle of equality of arms

155 The third part is based on two complaints. The first alleges inadequate access to the file. The second alleges, in essence, that the content of the case file submitted was insufficient.

– Preliminary observations

156 The rights of the defence are fundamental rights forming an integral part of the general principles of law whose observance the General Court and the Court of Justice ensure (judgment of 25 October 2011, Solvay v Commission, C‑109/10 P, EU:C:2011:686, paragraph 52).

157 Observance of the rights of the defence is a general principle of EU law which applies where the authorities are minded to adopt a measure which will adversely affect an individual (judgment of 16 January 2019, Commission v United Parcel Service, C‑265/17 P, EU:C:2019:23, paragraph 28).

158 That general principle of EU law is enshrined in Article 41(2)(a) and (b) of the Charter (judgment of 25 March 2021, Deutsche Telekom v Commission, C‑152/19 P, EU:C:2021:238, paragraph 105).

159 In the context of competition law, observance of the rights of the defence requires that any addressee of a decision finding that that addressee has committed an infringement of the competition rules must have been afforded the opportunity, during the administrative procedure, to make known its views on the truth and relevance of the facts and circumstances alleged as well as on the documents used by the Commission to support its claim that there has been such an infringement, as is apparent from the case-law cited in paragraph 144 above.

160 According to settled case-law, an infringement of the rights of the defence, in particular the right to be heard, results in the annulment of the decision taken at the end of the administrative procedure at issue only if, had it not been for such an irregularity, the outcome of the procedure might have been different. An applicant who relies on infringement of his or her rights of defence cannot be required to show that the decision of the EU institution concerned would have been different in content, but simply that such a possibility cannot be totally ruled out (see judgment of 18 June 2020, Commission v RQ, C‑831/18 P, EU:C:2020:481, paragraphs 105 and 106 and the case-law cited), since that party would have been better able to ensure its defence had there been no procedural error (judgments of 2 October 2003, Thyssen Stahl v Commission, C‑194/99 P, EU:C:2003:527, paragraph 31; of 1 October 2009, Foshan Shunde Yongjian Housewares and Hardware v Council, C‑141/08 P, EU:C:2009:598, paragraph 94; and of 13 December 2018, Deutsche Telekom v Commission, T‑827/14, EU:T:2018:930, paragraph 129).

161 Such an assessment must be made in the light of the factual and legal circumstances of each case (judgment of 18 June 2020, Commission v RQ, C‑831/18 P, EU:C:2020:481, paragraph 107).

– The first complaint: inadequate access to the file

162 The applicant submits that it had access to the file only once the SO and the SSO had been adopted and after it had had to make repeated requests to that effect. It also claims that it had to devote ‘an inordinate amount of time and resources’ to consulting certain documents, which constituted an ‘unnecessary and harmful distraction’ and negatively affected its ability to defend itself.

163 The Commission disputes the arguments put forward by the applicant.

164 A corollary of the principle of respect for the rights of the defence, the right of access to the file means that the Commission must give the undertaking concerned the opportunity to examine all the documents in the investigation file which may be relevant for its defence. Those documents include both incriminating evidence and exculpatory evidence, save where the business secrets of other undertakings, the internal documents of the Commission or other confidential information are involved (judgments of 7 January 2004, Aalborg Portland and Others v Commission, C‑204/00 P, C‑205/00 P, C‑211/00 P, C‑213/00 P, C‑217/00 P and C‑219/00 P, EU:C:2004:6, paragraph 68, and of 12 July 2011, Toshiba v Commission, T‑113/07, EU:T:2011:343, paragraph 41).

165 In the present case, the applicant complains, in the first place, that it had to insist to the Commission in order to obtain access to the documents in the file, which caused it to lose time and caused memories to fade and, consequently, had a negative effect on its defence.

166 The applicant’s arguments concerning the length of the administrative procedure and its consequences for the applicant’s ability to defend itself must be rejected for the same reasons as those set out in paragraphs 116 to 121 above. Furthermore, in so far as the applicant submits that it was given access to the file only at a late stage, it is common ground that the applicant did have access to the file during the administrative procedure and that it was able to take account of the documents contained in the file when organising its defence. Accordingly, even if the applicant did not have immediate access to its file, the fact remains that it has not proved that that access was at a late stage and that its rights of defence were infringed, with the result that that argument must be rejected.

167 As regards, in the second place, the alleged inadequate access to the file, it should be noted, as is apparent from recitals 45 and 46 of the contested decision, which the applicant does not dispute, that, as regards the SO, the applicant itself acknowledged that any access to file issues were resolved before it submitted the SO response and that, as regards the SSO, it did not make use of the opportunity offered to it by the Commission of supplementing the SSO response after having had access to a revised non-confidential version of certain documents in the file, with the result that it cannot validly claim that its rights of defence were affected.

168 Consequently, the present complaint must be dismissed.

– The second complaint: the content of the file submitted was insufficient

169 The applicant submits that the Commission failed to fulfil its obligation to take detailed notes of all meetings, telephone calls and interviews, whether formal or informal, conducted for the purpose of collecting information relating to the subject matter of an investigation, and properly to provide it with those notes.

170 In the first place, the applicant submits that the Commission failed to take notes of seven conference calls and five meetings with the applicant. The Commission could then not refer to those notes, which were likely to have contained exculpatory evidence.

171 In the second place, the applicant submits that the Commission was too late in sending it certain notes of meetings or conference calls with third parties, sometimes a number of years after those meetings were held.

172 In the third place, the applicant maintains that the notes which the Commission sent it relating to certain conference calls with third parties, namely a conference call with Huawei and seven conference calls with the complainant, were too brief and did not enable it to understand the information exchanged therein.

173 The Commission disputes the arguments put forward by the applicant.

174 At the outset, it must be stated that Article 19(1) of Regulation No 1/2003 empowers the Commission to interview any natural or legal person who consents to be interviewed for the purpose of collecting information relating to the subject matter of an investigation, in order to carry out the duties assigned to it by that regulation.

175 In accordance with Article 3 of Regulation No 773/2004, the Commission may record the statements made by persons interviewed in any form when conducting an interview on the basis of Article 19 of Regulation No 1/2003. Article 3(2) of Regulation No 773/2004 also states that such an interview may be conducted by any means including by telephone or electronic means.

176 In that regard, it is apparent from the case-law that, if the Commission decides to carry out such an interview on the basis of Article 19(1) of Regulation No 1/2003, it must record the interview in full, without prejudice to the fact that it is free to decide on the type of recording. It follows that the Commission is required to record, in a form of its choosing, any interview which it conducts, under that article, for the purpose of collecting information relating to the subject matter of an investigation (judgments of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraphs 90 and 91, and of 9 March 2023, Les Mousquetaires and ITM Entreprises v Commission, C‑682/20 P, EU:C:2023:170, paragraph 89).

177 For that purpose, it is not sufficient for the Commission to make a brief summary of the subjects addressed during the interview. It must be in a position to provide an indication of the content of the discussions which took place during the interview, in particular the nature of the information provided during that interview on the subjects addressed (judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments), T‑235/18, EU:T:2022:358, paragraph 190; see also, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraphs 91 and 92).

178 In the first place, as regards the contacts which took place between the Commission and the applicant itself during the administrative procedure, the applicant complains that the file contained no notes relating to seven conference calls and five meetings.

179 It should be noted that the applicant in no way explains how, if a procedural irregularity had occurred, its rights of defence could have been affected, by demonstrating that it would have been better able to ensure its defence had the Commission taken notes of those contacts with it. Since those meetings took place between the Commission and the applicant itself, the applicant was fully aware of their content and of the subjects discussed with the Commission at those meetings. It was therefore perfectly able to use any exculpatory evidence which might potentially have been addressed during those contacts in order to be better able to ensure its defence. In addition, in such a situation, it might have been expected that the applicant would have taken the precaution of itself sending the Commission a summary of the contact in question, in order to leave a written record of any potential exculpatory evidence in the file.

180 In the second place, as regards certain notes relating to meetings or conference calls with third parties, the applicant complains that it obtained access thereto only at a late stage, which infringed its rights of defence.

181 In that regard, irrespective of whether or not the meetings and telephone calls in question constitute ‘interviews’ within the meaning of Article 19(1) of Regulation No 1/2003 which the Commission was required to record, which, moreover, the applicant does not even attempt to demonstrate, it is not disputed that the applicant had access to the notes relating to those meetings and telephone conversations on 31 July and 27 August 2018, that is to say just after the Commission sent the SSO.

182 While it is regrettable that access to the notes relating to the meetings and conference calls in question was actually granted long after those meetings and conference calls were held, the fact remains that the applicant responded to the SSO on 22 October 2018, that is to say, several weeks after receiving those notes, which gave it sufficient time to examine them, to extract any potentially exculpatory evidence from them and to rely on them in that response.

183 In that regard, it should be noted that the circumstances of the present case must be distinguished from those underlying the case which gave rise to the judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments) (T‑235/18, EU:T:2022:358), in which the General Court annulled Commission Decision C(2018) 240 final of 24 January 2018 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.40220 – Qualcomm (exclusivity payments)), in particular due to the Commission having sent certain notes relating to interviews with third parties at a late stage.

184 Thus, it is apparent from paragraphs 168 and 169 of the judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments) (T‑235/18, EU:T:2022:358), that, in the case which gave rise to that judgment, the Commission had not forwarded to Qualcomm any information during the administrative procedure concerning either the existence or the content of certain interviews with third parties and that it was only after the adoption of Decision C(2018) 240 final that it forwarded such notes, some of which had even been provided during the proceedings before the Court, in response to measures of inquiry ordered by the Court.

185 Moreover, it is not disputed that, in the case which gave rise to the judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments) (T‑235/18, EU:T:2022:358), unlike in the present case, Qualcomm did not have access to those notes during the administrative procedure and was therefore unable to make known its views on them or use any exculpatory evidence in them to defend itself before the adoption of the decision finding Qualcomm to have made exclusivity payments. Those circumstances of significance, which led the Court to annul Decision C(2018) 240 final, constitute a fundamental difference from the present case.

186 It follows from the foregoing that the applicant has failed to demonstrate that its rights of defence were infringed by the fact that the Commission sent to it the notes relating to the meetings and conference calls in question at a relatively late stage. The applicant’s argument alleging that the Commission sent those notes at a late stage must therefore be rejected.

187 In the third place, the applicant complains that certain notes relating to conference calls with third parties, namely a conference call with Huawei and seven conference calls with the complainant, are too brief.

188 In that regard, at the hearing, the Commission acknowledged, first, that at least some of the conference calls to which the applicant referred could actually be characterised as ‘interviews’ within the meaning of Article 19(1) of Regulation No 1/2003 and, second, that it had not recorded them. It should also be observed that the notes relating to those interviews, as set out in Annex A.9 to the application, are too brief to be able to compensate for that failure to make a recording.

189 As regards the inferences to be drawn from such a procedural irregularity, it is necessary to determine whether, in the light of the factual and legal circumstances specific to the present case, the applicant has demonstrated adequately that it would have been better able to defend itself had there been no such irregularity. In the absence of such demonstration, that irregularity cannot result in the annulment of the contested decision.

190 Where the undertaking concerned can give a useful indication of the authors and the nature and content of the documents which have been withheld from it, it must present the facts and adduce the evidence to prove that it could have used in its defence parts of the file to which it, in error, did not have access, whether it be inculpatory or exculpatory (see, to that effect, Opinion of Advocate General Kokott in Solvay v Commission, C‑110/10 P, EU:C:2011:257, point 37).

191 If inculpatory evidence has been withheld from it, it is sufficient for the undertaking concerned to demonstrate that the administrative procedure could have led to a different result if that evidence was disallowed, in so far as the Commission relied on that evidence (see, to that effect, judgments of 7 January 2004, Aalborg Portland and Others v Commission, C‑204/00 P, C‑205/00 P, C‑211/00 P, C‑213/00 P, C‑217/00 P and C‑219/00 P, EU:C:2004:6, paragraphs 71 and 73, and of 12 July 2011, Toshiba v Commission, T‑113/07, EU:T:2011:343, paragraph 46).

192 If exculpatory evidence has been withheld from it, the undertaking concerned must show that it would have been able to use that evidence for its defence, in the sense that, had it been able to rely on it during the administrative procedure, it would have been able to invoke evidence which was not consistent with the inferences made at that stage by the Commission and therefore could have had an influence, in any way at all, on the assessments made by the Commission in its decision (see judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 97 and the case-law cited).

193 In the present case, it must be observed that the applicant has failed to establish that, had it not been for the procedural irregularity identified, it would have been better able to defend itself.

194 In particular, despite the fact that it was, overall, possible to deduce from the notes sent by the Commission the content of the interviews in question and, consequently, to infer that evidence may exist which it could have used to defend itself, the applicant puts forward no detailed argument making it possible to understand how it would have been better able to defend itself, even when the Court expressly questioned it in that regard at the hearing.

195 Lastly, it must be observed that in the judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments) (T‑235/18, EU:T:2022:358), in which the Court annulled Decision C(2018) 240 final, in particular because the notes sent by the Commission relating to certain interviews with third parties were incomplete, the applicant had adduced, in order to substantiate the claims made in that regard in the application, an annex intended to clarify the issues which might have been discussed during those interviews and how that material could have assisted its defence. In the present case, the applicant has not put forward any such information.

196 It follows that the procedural irregularity based on the notes relating to conference calls with third parties being overly brief, which has been observed in the present case, cannot result in the annulment of the contested decision.

197 Consequently, the present complaint must be rejected, as must the second part of the first plea. Therefore, since the first part of the first plea has also been rejected (see paragraph 154 above), that plea must be rejected in its entirety.

 The second plea: ‘manifest errors of assessment’, of fact and of law, and failure to state adequate reasons as regards the definition of the relevant market and the applicant’s dominant position during the relevant period

198 The second plea comprises five parts. The first part relates to shortcomings in the contested decision as regards the definition of the relevant market. The second relates to the competitive constraints exerted directly by captive supply on the merchant market. The third relates to the competitive constraints exerted indirectly by captive supply on the merchant market. The fourth relates to the applicant’s dominant position during the relevant period. The fifth relates to the definition of the leading edge segment of the UMTS chipset market on which the Commission based its analysis.

 The first part: shortcomings in the contested decision as regards the definition of the relevant market

199 The present part of the plea is based on three complaints. The first alleges manifest errors of assessment and of law in that the Commission relied, in order to define the relevant market, on a selection of vague answers to unclear requests for information. The second alleges manifest errors of assessment and of law in that the Commission failed to examine whether a chain of substitution existed. The third alleges manifest errors of assessment and of law in that the Commission concluded that it was not required to apply the ‘small but significant and non-transitory increase in price’ test (the ‘SSNIP test’).

 Preliminary observations

200 It should be recalled that, in the application of Article 102 TFEU, the objective of the definition of the relevant market is to define the boundaries within which it must be assessed whether the undertaking concerned is able to behave, to an appreciable extent, independently of its competitors, its customers and consumers (see judgment of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraph 127 and the case-law cited).

201 The determination of the relevant market is therefore, as a general rule, a prerequisite of any assessment of whether the undertaking concerned holds a dominant position, which involves defining, first, the products or services in that market and, second, the geographical dimension of that market (see judgment of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraphs 127 and 128 and the case-law cited).

202 As regards the product or services market, the concept of the ‘relevant market’ implies that there can be effective competition between the products or services which form part of it, and this presupposes that there is a sufficient degree of interchangeability or substitutability between those products and those services. That interchangeability or substitutability is not assessed solely in relation to the objective characteristics of the products and services at issue. There must also be taken into consideration the conditions of competition and the structure of supply and demand on the market (see judgment of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraph 129 and the case-law cited).

203 Such an assessment presupposes, however, that there is a sufficient degree of interchangeability between the products or services which form part of the relevant market and those which are intended to satisfy the demand on that market. Such is the case if the person making an alternative offer is in a position, within a short period, to satisfy the demand with sufficient strength to constitute a serious counterbalance to the power held by the undertaking concerned on that market (see, to that effect, judgment of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraphs 132 and 133).

204 Lastly, according to settled case-law, in so far as the definition of the market concerned involves complex economic assessments on the part of the Commission, it is subject to only limited review by the EU Courts (judgments of 17 September 2007, Microsoft v Commission, T‑201/04, EU:T:2007:289, paragraph 482; of 7 May 2009, NVV and Others v Commission, T‑151/05, EU:T:2009:144, paragraph 53; and of 15 December 2010, CEAHR v Commission, T‑427/08, EU:T:2010:517, paragraph 66).

205 The applicant’s argument concerning the definition of the relevant market adopted by the Commission in the contested decision must be examined in the light of those considerations.

– The first complaint: manifest errors of assessment and of law in that the Commission relied on a selection of vague answers to unclear questions

206 The applicant submits that, in the contested decision, the Commission ruled out the potential substitutability between chipsets supporting the UMTS standard and chipsets supporting other standards on the basis of the biased analysis of selected responses to poorly worded questions.

207 In particular, according to the applicant, the contested decision is based on unclear or ambiguous answers to questions which were confusing and open to misinterpretation, set out in requests for information sent a number of years after the alleged infringement, with the result that a number of respondents to the questionnaires of 4 November 2014 and 30 April 2015 interpreted the questions posed as relating to the market as it stood at the point in time when those requests for information were issued, and not at the time of the alleged infringement. Furthermore, the wording of certain questions relating to substitutability between UMTS chipsets and chipsets supporting other standards led the undertakings questioned to answer the question whether they intended to switch supplier for chipsets intended to be incorporated into existing devices instead of future devices. Lastly, the drafting of the questions prejudged the existence of two distinct markets, one covering UMTS chipsets and the other covering chipsets supporting LTE, since respondents were asked whether they would switch from one type of chipset to another.

208 The Commission disputes the arguments put forward by the applicant.

209 At the outset, it should be observed that, in the contested decision, the Commission concluded on the basis of a large amount of data that there was no substitutability between UMTS chipsets and chipsets supporting other standards. That is apparent from footnotes 245, 246 and 248 to 251 to that decision as regards the lack of substitutability between UMTS chipsets and other chipsets supporting the GSM standard, from footnotes 254 to 258 to that decision as regards the lack of substitutability between UMTS chipsets and other chipsets supporting the CDMA standard, from footnotes 260 and 264 to that decision as regards the lack of substitutability between the ‘Frequency-division-duplexing’ (FDD) variant of UMTS chipsets and the same chips in the ‘Time-division-duplex’ (FDD) variant, which do not support FDD mode, and from footnotes 271 to 273 to the contested decision concerning the lack of substitutability between UMTS chipsets and other chipsets supporting the WiFi/WiMax standard. Furthermore, as is apparent from footnotes 277 and 279 to the contested decision, the Commission also relied on a large number of documents in order to conclude that UMTS chipsets supporting various versions of that standard were substitutable.

210 In addition, it should be noted that the Commission relied not only on the replies to the questionnaires sent in 2014 and 2015, but also on the replies to two questionnaires sent in 2010 which also concerned the definition of the relevant market, which the Commission expressly confirmed in response to a measure of organisation of procedure in the form of a written question put by the Court, and also relied on industry reports, as is apparent from footnotes 244, 261 and 270 to the contested decision.

211 Consequently, the Commission cannot be criticised for having relied solely on a selection of answers when it defined the relevant market.

212 In addition, even if some of the undertakings questioned had been able to reply to the questionnaires of 2014 and 2015 referring to an incorrect time frame, it must be stated that not all the respondents to whom the Commission refers in footnotes 244, 261 and 270 to the contested decision made such an error, which the applicant does not dispute, since it simply cites a number of examples of undertakings whose responses do not, in its view, relate to the relevant period.

213 As regards, in particular, the questionnaire of 30 April 2015, it should be noted that it states, by way of introduction, that replies are to be provided for the period from 2010 to 2014. As to why that questionnaire did not concern 2009, even though the second quarter of that year is included in the relevant period, when questioned in that regard in writing by the Court, the Commission stated that many respondents to that questionnaire had expressly referred to a time frame prior to 2010 or had made statements which applied generally, regardless of the reference period and which, therefore, were also valid for 2009.

214 It follows that, even if the some of the respondents to the questionnaires of 4 November 2014 and 30 April 2015 to which the applicant refers provided vague responses, as the applicant maintains, referring to a time frame which differed from the infringement time frame, such an error would not in any event be capable of calling into question the Commission’s analysis in the light of all the responses regarding the lack of substitutability between UMTS chipsets and chipsets supporting other standards, having regard to the very large amount of consistent data on which the Commission relied in that regard.

215 Furthermore, by maintaining that the wording of certain questions relating to substitutability between UMTS chipsets and chipsets supporting other standards resulted in the undertakings questioned answering the question whether they intended to switch supplier for chipsets intended to be incorporated into existing devices, rather than into future devices, the applicant submits, in essence, that the Commission ought to have examined whether, hypothetically, there was a competitive constraint exerted by suppliers of chipsets to be incorporated into future devices, namely products which did not yet exist.

216 That line of argument, however, cannot succeed.

217 It is apparent from paragraph 7 of the Commission Notice on the definition of the relevant market for the purposes of Community competition law (OJ 1997 C 372, p. 5; the ‘Notice on market definition’) that a relevant product market comprises all those products or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products’ characteristics, their prices and their intended use, which implies that the assessment is to be based on the types of products which the undertakings in question sell at that time, and not by reference to hypothetical future products, whose characteristics, price and use are not yet known to consumers.

218 It should also be observed in that regard that the applicant does not explain in any event how the hypothetical competitive constraint exerted by chipsets intended for devices which do not yet exist could have had any influence at all on the outcome of the analysis conducted by the Commission, which, as stated in paragraphs 209 and 210 above, relied on a large amount of consistent data.

219 Lastly, it is necessary to reject as unfounded the applicant’s argument that, by asking the respondents whether they would switch from one type of chipset to the other, the Commission prejudged the existence of two distinct markets, namely a UMTS chipsets market and a market for chipsets supporting LTE.

220 The questions put by the Commission in that regard refer only to specific technologies, in particular the GSM, UMTS and LTE standards, each of which has its own characteristics, and specifically concern whether or not the chipsets supporting those standards are regarded as substitutable by those who purchase them and, consequently, as potentially forming part of the same product market.

221 It follows from the above that the present complaint must be rejected.

– The second complaint: manifest errors of assessment and of law in that the Commission failed to examine whether a chain of substitution exists

222 The applicant claims that the Commission committed manifest errors of assessment and of law by failing to examine whether a chain of substitution exists between UMTS chipsets and chipsets supporting earlier or later standards.

223 According to the applicant, the Commission ought to have examined whether, at the time of the transition from one standard to another, there was what it described as a ‘chain of substitution’, that is to say, in its own words, ‘a continuum of products on offer or, in the alternative, an obvious structural “breaking point” between the two standards’. In particular, it disputes the Commission’s conclusion that there was no chain of substitution between UMTS chipsets and chipsets which had previously supported the broad GSM/EDGE standard.

224 The Commission disputes the arguments put forward by the applicant.

225 It must be stated that, in paragraph 57 of the Notice on market definition, the Commission explains that the existence of chains of substitution might lead, in certain cases, to the inclusion in a market of certain products which are not directly substitutable, where product B is a demand substitute for products A and C. In that situation, even if products A and C are not direct demand substitutes, they might be found to be in the same relevant product market since their respective pricing might be constrained by substitution to B.

226 Clearly, in the present complaint, the applicant is, in essence, criticising the Commission for having concluded, first, that chipsets supporting an earlier standard and UMTS chipsets were not substitutable and, second, that UMTS chipsets and chipsets supporting a later standard were not substitutable. In other words, it complains that the Commission carried out two ‘traditional’ assessments of substitutability between two products, and did not assess possible substitutability via a chain of substitution between two products which were not directly substitutable but which would become so indirectly, by virtue of their both being substitutable for a third product.

227 In that regard, it is sufficient to recall that the Commission examined in the contested decision the substitutability between UMTS chipsets and chipsets supporting other standards, either earlier or later than the UMTS standard, relying on a large amount of consistent data, as is apparent from paragraphs 209 and 210 above.

228 It follows from the above that the present complaint must be rejected.

– The third complaint: manifest errors of assessment and of law in that the Commission concluded that it was not required to apply the SSNIP test

229 The applicant is of the view that, by failing to apply the SSNIP test in the contested decision, the Commission made a manifest error of assessment and an error of law. According to the applicant, while the judgment of 11 January 2017, Topps Europe v Commission (T‑699/14, not published, EU:T:2017:2), to which the Commission refers in that decision, does indeed refer to tools being available to the Commission other than that test, including market studies or an assessment of consumers’ and other competitors’ points of view, the fact remains that the Commission is still required to rely on suitable, convincing and reliable evidence, which it did not do in the present case, since it relied on a selection of answers to confusing questions set out in requests for information. The applicant also criticises the Commission for having concluded, in recital 248 of that decision, that such a test was not in any event appropriate since prices of UMTS chips may already have been set at a supra-competitive level.

230 The Commission and the intervener dispute the applicant’s arguments.

231 As a preliminary point, it should be recalled that the SSNIP test consists of examining whether a small increase in the price of a product, in the order of 5 to 10%, would lead to a significant number of customers choosing another product, which will in that case be regarded as substitutable for the first product.

232 It should also be noted that, although the SSNIP test is a recognised method for defining the market at issue, it is not the only method available to the Commission. It may also take into account other tools for the purposes of defining the relevant market, such as market studies or an assessment of consumers’ and competitors’ points of view (judgments of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 82, and of 5 October 2020, HeidelbergCement and Schwenk Zement v Commission, T‑380/17, EU:T:2020:471, paragraph 331 (not published)), which the applicant does not dispute.

233 Furthermore, it is clear both from the case-law and from paragraph 25 of the Notice on market definition that there is no rigid hierarchy between the different criteria available to the Commission (see, to that effect, judgments of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 82, and of 5 October 2020, HeidelbergCement and Schwenk Zement v Commission, T‑380/17, EU:T:2020:471, paragraph 331 (not published)), which the applicant also does not dispute.

234 In addition, it should be recalled that the Commission has a certain margin of discretion with regard to the definition of the relevant market, in so far as such definition involves complex economic assessments (see, to that effect, judgment of 15 December 2010, CEAHR v Commission, T‑427/08, EU:T:2010:517, paragraph 66 and the case-law cited).

235 It follows that the Commission is not obliged, when defining the relevant market for the purposes of applying Article 102 TFEU, to carry out a SSNIP test.

236 The Commission therefore did not err in law when it stated, in recital 180 of the contested decision, that it was able in the present case to define the relevant market without having to carry out a SSNIP test.

237 Furthermore, it is apparent from the analysis of the first complaint relied on by the applicant in support of the first part of its first plea that, when it defined the relevant market in the contested decision, the Commission relied on suitable, convincing and reliable evidence, and not, as the applicant claims, on a selection of answers to confusing questions set out in requests for information. The applicant therefore has failed to demonstrate that the Commission erred in failing to carry out the SSNIP test in the present case.

238 Moreover, the applicant has not even attempted to demonstrate that the use of the SSNIP test would have made any difference to the conclusions reached by the Commission in the contested decision (see, to that effect, judgment of 5 October 2020, HeidelbergCement and Schwenk Zement v Commission, T‑380/17, EU:T:2020:471, paragraph 331 (not published)).

239 The present complaint must therefore be rejected, there being no need to give a ruling on the substance of one of the justifications given by the Commission for its failure to use the SSNIP test in the present case, which is disputed by the applicant, namely the fact that such a test was not in any event appropriate in the present case since the price of UMTS chipsets was already set at a supra-competitive level.

240 The first part of the second plea must therefore be rejected.

 The second part: the competitive constraints exerted directly by captive supply on the merchant market

241 In support of the present part of the plea, the applicant puts forward four complaints.

242 In the first place, the applicant maintains that, in order to conclude that there was no competitive constraint directly from the captive supply of certain vertically integrated original equipment manufacturers in the UMTS chipset market, the Commission erred in relying on the inconsistent and irrelevant statements of those manufacturers, but did not question customers about their arrangements and ability, in the event of an increase in the price of chipsets on that market, to switch to self-supply or to increase self-supply for those already vertically integrated.

243 In the second place, the applicant submits that the Commission erred in excluding [confidential]1F(1)’s captive production from the relevant market on the ground that it was minimal during the relevant period. It refers in that regard to its replies to the SO and to the SSO, in which it demonstrated that [confidential] exerted a significant competitive constraint.

244 In the third place, the applicant criticises the Commission’s rejection of certain evidence relating to market dynamics and trends in the sector, whereas, in a sector as dynamic as that concerned by the contested decision, it was necessary to take account of those developments and trends. That was all the more so in the present case since, as regards other aspects of that decision, in particular in order to substantiate the significant growth potential of the leading edge segment of the UMTS chipsets market, the Commission took account of future projections.

245 In the fourth place, the applicant complains that the Commission failed to acknowledge commercial realities, in particular the fact that it had lost approximately 70% of its sales of UMTS chipsets for Huawei MBB devices to captive supply from Huawei’s subsidiary.

246 The Commission disputes the arguments put forward by the applicant.

247 According to paragraph 7 of the Notice on market definition, a relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products’ characteristics, their prices and their intended use.

248 Moreover, according to paragraph 13 of the Notice on market definition, undertakings are subject to three sources of competitive constraints, namely demand substitutability, supply substitutability and potential competition, and in the context of market definition, demand substitution constitutes the most immediate and effective disciplinary force on the suppliers of a given product. An undertaking can have a significant impact on the prevailing conditions of sale only if customers are in a position to switch easily to available substitute products, which results in identifying the effective alternative sources of supply for the customers of the undertakings involved.

249 Moreover, as is stated in paragraph 20 of the Notice on market definition, supply-side substitutability may also be taken into account when defining the relevant market in those situations in which its effects are equivalent to those of demand substitution in terms of effectiveness and immediacy. That means that suppliers are able to switch production to the relevant products and market them in the short term without incurring substantial additional costs or risks in response to small and permanent changes in relative prices (judgments of 17 September 2007, Microsoft v Commission, T‑201/04, EU:T:2007:289, paragraph 484, and of 9 September 2009, Clearstream v Commission, T‑301/04, EU:T:2009:317, paragraph 50).

250 In addition, it is apparent from paragraph 24 of the Notice on market definition that potential competition is not taken into account when defining markets, since the conditions under which potential competition will actually represent an effective competitive constraint depend on the analysis of specific factors and circumstances relating to the conditions of entry.

251 Lastly, it is clear in particular from points 16, 20, 21 and 23 of the Notice on market definition that the necessary substitutability for the purposes of the definition of the relevant market must materialise in the short term (see, to that effect, judgment of 29 March 2012, Telefónica and Telefónica de España v Commission, T‑336/07, EU:T:2012:172, paragraph 123).

252 In the present case, it is apparent from recital 215 of the contested decision, which the applicant does not, in essence, dispute, that, during the relevant period, only three vertically integrated undertakings self-supplied UMTS chipsets, [confidential] In the light of [confidential]’s minimal production of UMTS chipsets during that period, the Commission stated in that recital that it would not take [confidential]’s chipsets into account in its analysis.

253 As regards the first complaint, alleging that the Commission erred in relying on the inconsistent and irrelevant statements of certain manufacturers of vertically integrated original equipment manufacturers in the UMTS chipset market in order to conclude that there was no competitive constraint directly from the captive supply of those manufacturers, the Commission cannot be criticised for having questioned [confidential] during its examination as to the destination of the chipsets which they self-produced and as to their intentions in that regard. As is apparent from paragraph 13 of the Notice on market definition, in order to be able to determine whether customers would be in a position to switch easily to available substitute products, it is necessary to identify the effective alternative sources of supply for those customers.

254 In that regard, the responses from [confidential] are very clear, since it is apparent from recital 218 of the contested decision that [confidential] and since it is apparent from recital 219 of that decision that [confidential]. In addition, it is apparent from recital 220 of that decision that [confidential]. Such statements do not in any way appear to be inconsistent or without basis, in contrast to the applicant’s general and unexplained assertions.

255 It follows that the captive production of [confidential] did not constitute an effective source of supply for third-party buyers of UMTS chipsets during the relevant period, with the result that those customers were not in a position to switch easily to that captive production as referred to in paragraph 13 of the Notice on market definition. In response to a measure of organisation of procedure, the Commission also stated that, in accordance with paragraph 24 of that notice, when defining the relevant market, it did not have to take into account products which were potentially (but not currently) in competition, which was the situation for the chipsets produced by [confidential], which were, during that period, only at a very early stage of their development. Consequently, the Commission did not make a manifest error of assessment and did provide an adequate statement of reasons when it found, on the basis of those responses, that captive production was not generally available for supply to third-party customers, which is sufficient to rule out the possibility that such captive production could have directly exerted a competitive constraint on UMTS chipset producers during that period.

256 It is also necessary to reject the applicant’s criticism that the Commission ought rather to have questioned customers about their arrangements and ability to switch to self-supply or to increase self-supply for customers who are already vertically integrated, should the price of UMTS chipsets on the merchant market increase. Having regard to the characteristics of the UMTS chipsets market and, in particular, to the high barriers to entry represented, in particular, by the significant costs of research and development (R&D) necessary for the development of such chipsets (Section 11.4.1 of the contested decision), it is difficult to imagine that non-vertically integrated customers could, in the short term, have started to produce such chipsets.

257 As regards vertically integrated customers, it is apparent from recital 220 of the contested decision that [confidential]. It was therefore no longer conceivable for [confidential] to increase its self-supply in the short term during the relevant period. As for [confidential], it is apparent from recital 230 of the contested decision, based on [confidential]’s answers to questions asked in that regard by the Commission, [confidential]. It is therefore reasonable to infer that, even if the price of UMTS chipsets on the merchant market were to increase, that manufacturer would not in any event have increased its captive production in the short term, which resulted in the creation of a direct competitive constraint on the producers of UMTS chipsets active on that market.

258 It is therefore necessary to reject the present complaint.

259 As regards the second complaint, alleging that the Commission erred in excluding [confidential]’s captive production, it should be noted that the applicant does not dispute that [confidential]’s captive production, which, as is apparent from footnote 285 to the contested decision, represented only 0.1% of total UMTS chipset sales on the merchant market in 2010 and 2011, was minimal.

260 The applicant also does not dispute that [confidential] used its captive production during the relevant period exclusively or almost exclusively for internal purposes and that it was only from [confidential], that is to say well after that period, that that manufacturer began to supply third-party customers on the merchant market, as is apparent from recital 225 of the contested decision.

261 It follows from the minimal nature of [confidential]’s self-supply and from its use of captive production exclusively or almost exclusively for internal purposes during the relevant period, which are not disputed by the applicant, that it is difficult to imagine that [confidential] was able, during the relevant period, to discipline the prices offered by the applicant on the UMTS chipsets market, contrary to the applicant’s allegations, and that the Commission gave sufficient reasons for, and did not make a manifest error of assessment in, excluding that captive production from its analysis, which is sufficient to reject the present complaint.

262 In any event, it should be observed that, even if the Commission had included [confidential]’s captive production in the relevant market, having regard to the minimal nature of that production, which is not disputed by the applicant, such inclusion would not have had the slightest influence on the Commission’s conclusion regarding the applicant’s dominant position on that market.

263 As regards the third complaint, alleging that the Commission did not take sufficient account of market dynamics and trends in the sector, it should be recalled that it is apparent from point 24 of the Notice on market definition that potential competition is not taken into account when defining markets. The Commission was therefore not required to take account of such developments subsequent to the relevant period.

264 Contrary to the applicant’s allegations, such an approach is not incompatible with the fact that, in recital 363 of the contested decision, the Commission took account of the significant growth potential of the leading edge segment of the UMTS chipset market, since that recital is found in Section 12 of that decision which deals with abuse, in which the Commission may take account of projections, in particular with a view to assessing the potential effects of the conduct at issue, and is not found in Section 10 of that decision which deals with market definition, in which the Commission must determine which products are, during the relevant period, regarded by customers as substitutable for the purposes of including them in the same product market. Since those are two different analyses, pursuing different objectives, it is normal for the Commission to take different factors into account when undertaking such analyses, without that giving rise to any inconsistency or incompatibility at all.

265 Lastly, it should be noted that the applicant simply disputes in general terms the failure to take into account what it describes as ‘evidence on market dynamics and industry trends’, and does not explain in any way what evidence is involved, how such evidence could have influenced the Commission’s analysis and, in particular, how that evidence ought to have led the Commission to conclude that the captive production of vertically integrated undertakings was a direct competitive constraint on producers active on the UMTS chipset market.

266 The present complaint must, therefore, be rejected.

267 As regards the fourth complaint, alleging that the Commission failed to acknowledge commercial realities, the applicant relies solely on the fact that it had lost approximately 70% of its sales of UMTS chipsets for Huawei MBB devices to captive supply by Huawei’s subsidiary. It is apparent from Huawei’s reply to question 25 of the request for information dated 19 July 2013, to which the applicant refers, that it was only in 2013 that HiSilicon provided 69% of Huawei’s supplies as captive sales, that is to say, after the relevant period. The Commission, therefore, made no manifest error of assessment in failing to take that circumstance into account, which is sufficient to reject the present complaint.

268 Accordingly, it must be held that the Commission did not make a manifest error of assessment and did not fail to fulfil its obligation to state reasons when it concluded that there was no competitive constraint directly from the captive supply on the UMTS chipset market. The second part of the second plea must therefore be rejected.

 The third part: the competitive constraints exerted indirectly by captive supply on the merchant market

269 In support of the present part of the plea, the applicant puts forward four complaints.

270 In the first place, the applicant argues that the Commission erred in limiting its analysis of whether there was an indirect competitive constraint on the downstream market for mobile phones incorporating a UMTS chipset, on the pretext that during the relevant period self-supply of UMTS chipsets for other devices, including MBB devices, was negligible. In the applicant’s view, such an approach, furthermore, contradicts the ‘theory of harm’ developed by the Commission, which focuses on ‘leading edge chipsets’ used in MBB devices.

271 In the second place, the applicant submits that, in the absence of the SSNIP test, the Commission’s assertion, in recital 237 of the contested decision, that a hypothetical price increase for UMTS chipsets would not lead to a sharp drop in demand for such chipsets on the merchant market, is unsubstantiated and immaterial.

272 In the third place, the rejection of indirect constraints stemming from competition on the downstream market is irreconcilable with the ‘theory of harm’ developed by the Commission for two reasons.

273 First, the Commission cannot, without contradicting itself, argue that a hypothetical price increase of 5 to 10% in the upstream market, namely the UMTS chipset market, would have had a negligible impact on prices on the downstream market, namely the market for devices incorporating such chipsets (recital 235 of the contested decision) while also criticising the applicant for having engaged in indirect predation.

274 Second, the Commission cannot, without contradicting itself, argue that, even if [confidential]’s captive production were included in the relevant market, that would not, given that production was minimal, have had any impact on the applicant’s market share and therefore acknowledge that captive production exerted some degree of competitive constraint on the merchant market (footnote 311 to the contested decision) while also concluding that that captive production did not exert a competitive constraint on the merchant market during the relevant period.

275 In the fourth place, the applicant claims that the Commission erred in stating in footnote 311 to the contested decision that [confidential]’s captive production could be included in the relevant market, but [confidential]’s captive production could not, on the ground that [confidential] sourced UMTS chipsets during the relevant period for its MBB devices solely on the merchant market.

276 According to the applicant, the Commission could not exclude [confidential]’s captive production because [confidential] did not self-supply chips for its MBB devices, when the relevant market definition adopted by the Commission in the contested decision was not limited to UMTS chips for MBB devices, but also included UMTS chips for other devices. In addition, it is apparent from recital 220(b) of that decision that, while [confidential] announced in 2009 that it was ceasing captive production, in fact it self-supplied until the third quarter of 2013.

277 The Commission disputes the arguments put forward by the applicant.

278 As a preliminary point, it should be observed that, in Section 10.2.9.2 of the contested decision, the Commission explained the reasons which led it to conclude that the captive production of the only two vertically integrated undertakings during the relevant period did not exert an indirect competitive constraint on the UMTS chipset market, via competition on the downstream market.

279 In order to reach such a conclusion, the Commission first explained that it had carried out its analysis of a possible indirect competitive constraint taking into account only possible downstream competition between mobile phones including a UMTS chipset, to the exclusion of other devices, including MBB devices. In recital 230 of the contested decision, it stated that it had done so since that was the only possible indirect competitive constraint, since the self-supply of UMTS chipsets for devices other than mobile phones was, in any event, negligible during the relevant period.

280 Furthermore, the Commission also explained that it had confined itself to examining the indirect competitive constraint exerted by [confidential], since, during the relevant period, [confidential]’s captive production was minimal and therefore could not exert an indirect competitive constraint on the merchant market.

281 The Commission then explained, first, that, given that the dilution factor of the products concerned was low, since the price of a UMTS chipset represented, during the relevant period, only approximately 6% of the retail price of the mobile phone incorporating it, an increase of 5 to 10% in the price of that chipset would have only a negligible impact on the price of the finished product, with the result that such an upstream increase in the price of UMTS chipsets on the merchant market would remain profitable for the chip producer, who would not see the volume of its sales negatively affected by such an increase (recital 235 of the contested decision).

282 Second, the Commission has demonstrated that, in view of the fact that mobile phones are highly differentiated products depending on the brand, with highly specific characteristics, an increase in the price of a mobile phone brand following an increase in the price of UMTS chipsets incorporated in them on the merchant market does not mean that customers would switch to another mobile phone brand and, in particular, that they would instead buy a mobile phone incorporating a self-supplied chipset, which was not subject to such a price increase. That confirms that an upstream increase in the price of UMTS chipsets on the merchant market would remain profitable for the chipset manufacturer, who would not see the volume of its sales negatively affected by such an increase (recital 237 of the contested decision).

283 From that, the Commission inferred that captive production was not capable of exerting an indirect competitive constraint on the UMTS chipset market during the relevant period.

284 As regards the first complaint, alleging that the Commission erred in limiting its analysis of whether there was an indirect competitive constraint on the downstream market for mobile phones incorporating a UMTS chipset, it should be noted that captive production of UMTS chipsets was the highest for precisely those devices, estimated, for example, [confidential], with the result that limiting the Commission’s analysis to that segment of the market was the situation most likely to demonstrate that there was an indirect competitive constraint and, consequently, the most favourable to the applicant.

285 Furthermore, the fact that such limiting allegedly contradicts the ‘theory of harm’ developed by the Commission, focusing on ‘leading edge chipsets’ used in MBB devices, in no way alters the fact that that approach is more favourable to the applicant and in no way affects its validity. In that regard, it should be observed that captive production of ‘leading edge chipsets’ to be incorporated into those devices was, according to the Commission, much more limited than the captive production of UMTS chipsets to be incorporated into mobile phones, which the applicant, moreover, does not dispute.

286 As regards the second complaint, alleging that the Commission’s assertion in recital 237 of the contested decision that a hypothetical increase in the price of UMTS chipsets would not lead to a sharp drop in demand for such chips on the merchant market on account of the absence of the SSNIP test is unfounded and immaterial, it is sufficient to recall that the Commission is not required to carry out such a test (see paragraph 235 above). In addition, it must be stated that the applicant has put forward no evidence other than the absence of a SSNIP test in the present case in order to call into question that assertion by the Commission.

287 As regards the third complaint, alleging that the rejection of indirect constraints on the merchant market stemming from downstream competition is irreconcilable with the ‘theory of harm’ developed by the Commission, it must be stated that the indirect predation which the Commission alleges against the applicant cannot, in itself, invalidate the analysis of whether there was an indirect competitive constraint, for two reasons.

288 First, contrary to the applicant’s allegations, the Commission did not err and did not contradict itself when it (i) concluded that a hypothetical price increase of 5 to 10% on the upstream market, namely the UMTS chipset market, would have had only a negligible impact on the prices of devices incorporating those chipsets on the downstream market, namely the market for devices incorporating those chips, and (ii) also criticised the applicant for having engaged in indirect predation. It is not because a general increase in the price of UMTS chipsets on the upstream market would have had almost no downstream impact on the price of devices incorporating such chips that a selective reduction in the price of UMTS chipsets offered by the applicant solely to Huawei would not have favoured Huawei, allowing it better to compete with its rival ZTE on the downstream market.

289 Second, it must be observed that, in footnote 311 to the contested decision, the Commission in no way implied that [confidential]’s captive production represented some degree of competitive constraint on the merchant market. In that footnote, the Commission simply stated that, in any event, even if [confidential]’s entire captive supply of UMTS chipsets, that is to say the chipsets intended for mobile phones and the chipsets intended to be incorporated into other devices, were to be included in the relevant market, the applicant’s market share would remain unchanged in 2009 and would fall only marginally in 2010 and 2011, and would in any event remain above the 50% threshold. The applicant’s argument is therefore based on a misreading of that footnote.

290 As regards the fourth complaint, alleging that the Commission erred in stating in footnote 311 to the contested decision that [confidential]’s captive production could be included in the relevant market, but not [confidential]’s, on the ground that [confidential] sourced UMTS chipsets during the relevant period for its MBB devices solely on the merchant market, that is also based on a misreading of that footnote. In that footnote, the Commission in no way concluded that [confidential]’s captive production could be included in the relevant market but [confidential]’s could not be.

291 It should be stated, first of all, that footnote 311 to the contested decision relates to recital 240 of that decision, and in particular to whether, hypothetically, an indirect constraints analysis focusing on MBB devices would have led to a different outcome.

292 In that regard, in respect of [confidential], the Commission simply indicated that Huawei’s captive production of chipsets intended for both mobile phones and other devices was so minimal during the relevant period that, even if it had been included in the relevant market (whether in the indirect constraints analysis focused on mobile phones or the indirect constraints analysis focused on MBB devices), that would not have resulted in the applicant’s market share falling below the presumed dominance threshold of 50%.

293 However, as regards [confidential], the Commission indicated in footnote 311 to the contested decision that, in the hypothetical scenario of an indirect constraints analysis focused on MBB devices, [confidential]’s captive production would not have been taken into account, because [confidential] sourced all UMTS chipsets for its MBB devices from the merchant market and had announced in 2009 that it would cease its captive production, as is apparent from recital 220(b) of that decision. By contrast, in the analysis which the Commission did undertake of whether there was any indirect downstream constraint focused on mobile phones, it did take account of [confidential]’s captive production of UMTS chipsets intended for its mobile phones, since [confidential] did actually self-supply for those devices.

294 In addition, the fact that it is apparent from recital 220 of the contested decision that [confidential] shipped the last of those UMTS chipsets [confidential] in 2013 does not invalidate either the finding that [confidential] obtained its supplies solely on the merchant market for UMTS chipsets for its MBB devices and saw its captive production limited to chipsets for mobile phones, or the finding that [confidential] had decided in 2009 to put an end to its captive production and had ceased development of new UMTS chipsets in July 2010, since a number of years may well elapse between (i) that decision and the development of new UMTS chipsets coming to an end and (ii) the delivery of the last UMTS chipsets produced.

295 It follows from the foregoing that none of the complaints set out in support of the present part of the plea is capable of calling into question the Commission’s analysis in Section 10.2.9.1 of the contested decision, which concluded that during the relevant period captive production of UMTS chips on the merchant market was not an indirect constraint.

296 Consequently, the third part of the second plea must be rejected.

 The fourth part: the applicant’s dominant position during the relevant period

297 In support of the present part of the plea, the applicant puts forward the four complaints set out below.

298 In the first place, the applicant submits that, if the product market had been correctly defined and had included the captive sales of vertically integrated manufacturers, its market shares would have been below 40%, that is to say below the dominance threshold.

299 In the second place, the applicant argues that the Commission erred by overstating the importance of its market shares. According to the applicant, the baseband chips sector is characterised by short innovation cycles, with the result that market shares are ephemeral and do not correctly reflect market power. In addition, it maintains that, if the Commission had taken proper account of the constraints imposed on it, it would have come to the conclusion that the applicant did not have significant and durable market power.

300 In the third place, the applicant submits that the Commission erred when it concluded that the relevant market was characterised by high barriers to entry and expansion. In the view of the applicant, new entrants were encouraged to enter the market without being impeded by the need to make significant R&D investments on account of the fact that much of the relevant technology is standardised. It asserts that the grant-back network from which it benefited is also not a barrier to entry, which is demonstrated by the fact that only a small minority of respondents to a question from the Commission concerning barriers to entry mentioned that grant-back network as being such a barrier.

301 In the fourth place, the applicant complains that the Commission failed to quantify the other barriers to entry identified by it, namely the need to have baseband chipsets certified by mobile network operators and OEMs, its brand image, reputation and strong business relations.

302 The Commission disputes the arguments put forward by the applicant.

303 In accordance with the case-law, the dominant position referred to in Article 102 TFEU relates to a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by affording it the power to behave to an appreciable extent independently of its competitors, its customers and ultimately of consumers (judgment of 13 February 1979, Hoffmann-La Roche v Commission, 85/76, EU:C:1979:36, paragraph 38).

304 As a general rule, the existence of a dominant position may derive from several factors which, taken separately, are not necessarily decisive (judgments of 14 February 1978, United Brands and United Brands Continentaal v Commission, 27/76, EU:C:1978:22, paragraph 66, and of 15 December 1994, DLG, C‑250/92, EU:C:1994:413, paragraph 47). Among those factors, the existence of large market shares is highly important (judgments of 12 December 1991, Hilti v Commission, T‑30/89, EU:T:1991:70, paragraph 90, and of 25 June 2010, Imperial Chemical Industries v Commission, T‑66/01, EU:T:2010:255, paragraphs 255 and 256).

305 According to the case-law, very large market shares are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position (judgment of 13 February 1979, Hoffmann-La Roche v Commission, 85/76, EU:C:1979:36, paragraph 41). In particular, a market share of 50% in itself, and save in exceptional circumstances, constitutes a dominant position (see, to that effect, judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraph 60).

306 Other factors may also be taken into account when analysing the existence of a dominant position, such as the relationship between the market shares of the undertaking concerned and of its immediate competitors (see, to that effect, judgment of 13 February 1979, Hoffmann-La Roche v Commission, 85/76, EU:C:1979:36, paragraph 48) or the existence of barriers to entry or expansion, resulting in particular from the need to make significant investments (see, to that effect, judgment of 14 February 1978, United Brands and United Brands Continentaal v Commission, 27/76, EU:C:1978:22, paragraph 122) or even, in certain circumstances, of ownership of intellectual property rights (see, to that effect, judgment of 1 July 2010, AstraZeneca v Commission, T‑321/05, EU:T:2010:266, paragraph 270 and the case-law cited).

307 As regards the first complaint, alleging that, if the relevant product market had been correctly defined and had included vertically integrated manufacturers’ captive sales, the applicant’s market shares would have been below 40%, it is sufficient to recall that, as is apparent from the analysis of the first three parts of the second plea, the Commission did not err in defining the relevant product market as the UMTS chipset market, which permits the rejection of that complaint.

308 As regards the second complaint, alleging that, in the light of the specific characteristics of the sector, the importance of the applicant’s market shares, which the applicant regards as ephemeral, did not show that it had significant and durable market power, it must be stated that, even if the relevant market were characterised by short innovation cycles, that would not constitute exceptional circumstances within the meaning of the case-law cited in paragraph 305 above, allowing a finding that market shares of such a size do not, by themselves, constitute evidence of the existence of a dominant position, especially where the market shares of the closest competitor are much smaller, as in the present case. In any event, it is apparent from the data set out in Tables 3 and 5 of the contested decision, the essence of which the applicant does not dispute, that the applicant had a market share of 61.2% in 2009, 59.7% in 2010 and 62.5% in 2011 in terms of revenue and a market share of 58.9% in 2009, 57.8% in 2010 and 65.3% in 2011 in terms of volume. Such relatively stable market shares cannot therefore be described as ephemeral, contrary to the applicant’s assertions.

309 As regards market power, it is also apparent from the data set out in Tables 3 and 5 of the contested decision, the essence of which the applicant does not dispute, that its two closest competitors in terms of revenue were, in 2009 ST-Ericsson with 15.1% market share and Intel with 12.6% market share, in 2010 Intel with 18.9% market share and ST Ericsson with 8.9% market share and in 2011 Intel with 19.1% market share and Broadcom with 4.4% market share. In terms of volume, its two closest competitors were, in 2009, ST Ericsson with 20 to 30% market share and Intel with 10 to 20% market share, in 2010 Intel with 10 to 20% market share and ST Ericsson and Marvell with 5 to 10% market share and in 2011 Intel with 20 to -30% market share and Marvell with 5 to 10% market share. Contrary to the applicant’s assertions, those tables clearly demonstrate that during the relevant period, the applicant faced competition from only a few competitors, who were in a much weaker position and had relatively unstable market shares, in contrast with the applicant itself, which confirms that the applicant had significant and durable market power and makes it hard to believe that its competitors would be able in practice to constrain it. The second complaint must therefore be rejected.

310 As regards the third complaint, alleging, in essence, that the Commission erred in regarding the investments necessary in order to design a UMTS chipset, in particular R&D investments, as barriers to entry or expansion, and the fourth complaint, alleging that the Commission failed to quantify the other barriers to entry identified, namely the need to have baseband chipsets certified by mobile network operators and OEMs, the applicant’s brand image, reputation and strong business relations, it is sufficient to note that they relate to other factors which may be taken into account when analysing whether there is a dominant position (see paragraph 306 above) and are not able to call into question the evidence of the existence of a dominant position, as has been established in accordance with the case-law cited in paragraph 305 above.

311 It follows from the foregoing that the fourth part of the second plea must be rejected.

 The fifth part: the definition of the leading edge segment of the UMTS chipset market on which the Commission based its analysis

312 In support of the fifth part of the plea, the applicant complains, in essence, that the Commission based its analysis on the leading edge segment of the UMTS chipsets market, even though it is a construct of convenience which never existed and which is at odds with the technical and commercial realities of the relevant period.

313 That line of argument must be regarded as ineffective if it is relied on in order to call into question the definition of the relevant market or the applicant’s dominant position. Indeed, the Commission did not define the leading edge segment of the UMTS chipsets market in Section 10 of the contested decision which dealt with that definition, or in Section 11 of that decision which dealt with that dominant position, but rather in Section 12 of that decision which dealt with abuse, and in particular in its analysis of the applicant’s strategy to counter Icera in that leading edge segment (Section 12.4 of the contested decision).

314 When questioned in that regard by way of a measure of organisation of procedure, the applicant stated that, while it had set out its criticism of the definition of the leading edge segment of the UMTS chipset market on which the Commission based its analysis in the second plea, it did so in order to raise that fundamental point at the earliest opportunity in its written pleadings, because such a criticism potentially influenced the other pleas set out subsequently and because it wished to highlight the inconsistencies in the definition of the relevant market, including in relation to that leading edge segment which forms part of the relevant market.

315 Such a response is, however, not sufficient to circumvent the fact that the present part of the plea is ineffective, since, even if the complaint set out in support thereof were to succeed, that cannot call into question the conclusion reached by the Commission as regards the definition of the relevant market and the applicant’s dominant position.

316 It follows that the fifth part of the second plea must be rejected. Therefore, since the other parts of that plea have also been rejected (see paragraphs 240, 268, 296 and 311 above), that plea must be rejected in its entirety.

 The fifth plea: ‘manifest errors of assessment’ and failure to state reasons as regards the reconstruction of alleged ‘effectively paid’ prices

317 The fifth plea comprises two parts. The first part alleges that the ‘price restatement’ conducted in the contested decision was unnecessary. The second part alleges errors made in that decision as regards that ‘restatement’.

 The first part: the ‘price restatement’ conducted in the contested decision was unnecessary

318 In support of the present part of the plea, the applicant submits, in essence, that, by failing to rely on Qualcomm’s accounting data in order to determine chipset prices, the Commission undertook an unnecessary and unjustified ‘restatement’ of its price data. It thereby calls into question the contested decision’s reconstruction of the prices actually paid to Qualcomm by Huawei and ZTE during the relevant period.

319 In the first place, the applicant submits that, in the price reconstruction carried out in the contested decision, the Commission failed to take sufficient account of its objections to the price-cost test conducted in the SO, which criticised the fact that (i) that reconstruction excluded certain revenue by incorrectly treating certain limited instances of average sales price volatility as accounting errors and (ii) it failed to take into account the effects of delayed revenue recognition.

320 In the second place, the applicant submits that the Commission erred in its assessment of Qualcomm’s price data in its price-cost test by using a quarterly reference period. According to the applicant, the most meaningful and commercially relevant period for assessing pricing data was six months, or even one year. It states that it does not agree prices with its customers or revise them on a quarterly basis, but typically concludes agreements for much longer periods.

321 The applicant adds that the Commission incorrectly rejected the argument that aggregation over time would address problems arising from delayed revenue recognition, given that that risk concerns only limited amounts, and given that the risk that predatory pricing periods would be difficult to identify because their average over a longer reference period would exist only if the price-cost test were conducted not only over the period of predation, but also over the entirety or a portion of the recoupment period. According to the applicant, since the Commission-alleged predation in the present case took place over two years, it would have been possible to conduct the price-cost test on the basis of six-month reference periods.

322 In the third place, the applicant maintains that the exercise of restatement of average sales prices in the contested decision is unnecessary since, as regards the average sales prices recorded in its accounting data, the discrepancy compared with the result of the Commission’s calculations is, in any event, minimal. In addition, it submits that the price-cost test carried out by the Commission in the SO, which relied on its data, indicated, at most, intermittent below-cost pricing of a very short duration, which is inconsistent with the ‘theory of harm’ developed by the Commission based on anticompetitive foreclosure. In the applicant’s view, that suggests that the Commission restated average sales prices solely in order to avoid a finding of intermittent predation and, consequently, that there was no infringement.

323 The Commission disputes the arguments put forward by the applicant.

324 As a preliminary point, it should be recalled that Section 12.5 of the contested decision sets out the Commission’s methodology and the calculation of the quarterly average sales prices which Huawei and ZTE effectively paid per unit for the chipsets under investigation during the relevant period.

325 In the SO, the Commission had based its calculations mainly on Qualcomm’s accounting data. However, due to the application of United States Generally Accepted Accounting Principles (‘US GAAP’) and the arguments put forward in the SO response, the Commission concluded that it was inappropriate to base its price-cost test on those accounting data without making adjustments to take account of the revenue accounting principle and to reflect business reality more accurately during the relevant period.

326 In particular, according to the Commission, (i) the revenues initially recorded during a given quarter potentially understated the final effective price of units sold during that quarter and (ii) the delayed recognition of the reserves accumulated by Qualcomm which had not yet been paid inflated the revenues of the quarter in which they were recorded. That delayed recognition was due to the USA GAAP applied by the applicant, under which, at the time of physically shipping a chipset, Qualcomm had to account only for revenue which was certain and measurable. Reserves re-enter Qualcomm’s accounts only at the time when the precise quantity of chipsets for which a financial incentive applies is verified.

327 For that reason, the Commission concluded that it was inappropriate to rely on Qualcomm’s accounting data without making adjustments to take account of the difference between the quarter during which chipsets were sold and the quarter during which revenues were recorded.

328 Therefore, the Commission assessed the prices ‘effectively paid’ by taking into account the reserves which would re-enter Qualcomm’s accounts, so as to calculate the gross revenues collected in the quarter during which the sale took place.

329 In recitals 608 to 610 of the contested decision, the Commission also stated that it had relied on quarterly reference periods in the light of the fact that (i) it was apparent from certain documents of the [confidential], that Qualcomm’s pricing decisions were based on calendar quarters, (ii) customers also submitted their incentive claims on a quarterly basis and (iii) that period was consistent with the AVC data provided by Qualcomm, which were also recorded at a quarterly level.

330 As regards the first complaint, alleging, in essence, that the Commission did not correctly take into consideration the arguments which the applicant had set out in the SO response, it must be stated that the applicant does not explain how the Commission failed to take appropriate account of its observations or why it was not necessary to reconstruct prices charged in the contested decision in order to respond to the criticisms set out in that response.

331 In any event, it is sufficient to recall that, as stated in the contested decision, due to the application of US GAAP and the arguments put forward in the SO response, the Commission concluded that it was inappropriate to base its price-cost test on Qualcomm’s accounting data without making adjustments to take account of the revenue accounting principle. The Commission accordingly took that circumstance into account in the price reconstruction conducted in that decision by matching the adjustments made in the accounts to the sales concerned, so as to reflect business reality more accurately during the relevant period. In particular, it adjusted those accounting data, relying on documents held by the applicant, in order to reallocate the various incentive payments to the corresponding units and thereby to calculate the price effectively paid by Huawei and ZTE.

332 In addition, it should be noted, as the Commission did, that the Commission had, in the past, already made certain adjustments based on the price and cost statements of the dominant undertaking and on all other relevant information provided by that undertaking, in order to reflect business reality. That approach has been endorsed by the case-law of the General Court, for example, in the judgments of 30 January 2007, France Télécom v Commission (T‑340/03, EU:T:2007:22, paragraphs 131 to 137); of 10 April 2008, Deutsche Telekom v Commission (T‑271/03, EU:T:2008:101, paragraphs 208 to 211); and of 13 December 2018, Slovak Telekom v Commission (T‑851/14, EU:T:2018:929, paragraphs 220 to 235).

333 In those circumstances, the first complaint must be rejected.

334 As regards the second complaint, alleging that the quarterly reference period was not the most appropriate for conducting the price-cost test, it should be recalled that, in areas giving rise to complex economic assessments, the EU Court must not only establish whether the evidence put forward is factually accurate, reliable and consistent but must also determine whether that evidence contains all the relevant data that must be taken into consideration in appraising a complex situation and whether it is capable of substantiating the conclusions drawn from it. In that regard, the EU Courts must carry out their review of legality on the basis of the evidence adduced by the applicant in support of the pleas in law put forward (see judgment of 10 July 2014, Telefónica and Telefónica de España v Commission, C‑295/12 P, EU:C:2014:2062, paragraphs 54 and 56, and the case-law cited).

335 In the present case, in recitals 609, 610 and 630 to 633 of the contested decision, the Commission set out the evidence which led it to adopt a quarterly reference period and the reasoning followed in order to reject the objections raised by the applicant.

336 It should be observed at the outset that, while it is true that the data reproduced in Tables 28 to 30, 44 and 45 of the contested decision do not demonstrate complete uniformity in the determination of purchase periods, the fact remains that it follows from those data that the most frequently used period is in fact the quarterly period. It should be added, first, that the period in which customers were to submit their incentive claims was the quarterly period and, second, that Qualcomm itself provided the cost data relating to AVC on a quarterly basis. In addition, the applicant has not put forward, in its written pleadings or in its SO or SSO response to which it referred, any argument or evidence supporting its contention that the Commission ought to have used a reference period of six months or one year.

337 The other factors relied on by the applicant also do not permit the inference that the Commission erred in adopting a quarterly reference period in order to conduct the price-cost test.

338 First, the applicant has failed to demonstrate how the issue of delayed recognition, which affects the whole of the relevant period, could have been resolved by altering the reference period, as it argues.

339 Second, even if it were true, as the applicant maintains, that taking a longer reference period into account could distort the analysis only if prices relating to a period outside the predation period were taken into account, that does not mean that the choice of reference period must necessarily be that suggested by the applicant. The choice between a quarterly, half-yearly or annual reference period was made by the Commission in the present case on the basis of the most relevant data and circumstances concerning the applicant’s business. In that regard, it must be stated that the applicant has not succeeded in demonstrating that the choice of a quarterly reference period was vitiated by error.

340 In those circumstances, the second complaint must be rejected.

341 As regards the third complaint, alleging that the Commission’s restatement of average sales prices is unnecessary since, as regards the average sales prices recorded in the applicant’s accounting data, the discrepancy compared with the result of the Commission’s calculations is, in any event, minimal, it is apparent from recitals 614, 739, 740, 774, 775 and 922 of the contested decision that the difference between the applicant’s accounting prices and the prices reconstructed by the Commission was very often small. That fact was, in essence, confirmed by the parties at the hearing.

342 However, the applicant has not established that that fact affects the lawfulness of the contested decision. While the difference between the average sales prices recorded in the accounting data and the average sales prices reconstructed by the Commission is only minimal, it cannot be inferred therefrom that the Commission erred in reconstructing the average sales prices. In those circumstances, the present complaint must be rejected.

343 In addition, the applicant in no way substantiates its argument that a price-cost test which relied on its accounting data indicated, at most, intermittent below-cost pricing of a very short duration, and is therefore inconsistent with a finding of an infringement. Indeed, the applicant does not explain what it regards as intermittent predation and why that type of predation could not constitute an infringement of competition law.

344 In the light of the foregoing considerations, the third complaint must be rejected, as must the first part of the fifth plea in its entirety.

 The second part: errors in the contested decision regarding the ‘restatement’ conducted

345 In support of the present part of the plea, the applicant puts forward five complaints.

346 In the first place, the applicant maintains that, in order to calculate the ‘effectively paid’ prices, the Commission incorrectly allocated certain financial incentives from the quarters when they were recorded in Qualcomm’s accounting system to the quarters where the units concerned were shipped to the customer on the basis of an incorrect manipulation of incentive claims submitted to Qualcomm by Huawei and ZTE, found in ‘document packets’ which were unsuitable for the purposes of conducting a reliable price-cost test.

347 The applicant also states that the reasons why the ‘document packets’ in question were unsuitable had been set out in the SSO response, in which it had explained that they were prepared in great detail by customers, who were no longer able to provide clarification of them at the time when the Commission requested such clarification from them, in 2017. In addition, those ‘document packets’ were incomplete and sometimes vitiated by inconsistencies and errors.

348 In the second place, the applicant complains that the Commission’s approach unduly introduced ex post hindsight, in that it transferred reserves based on the assumption that Qualcomm could have forecasted incentive claims with perfect certainty at the time of sale.

349 In that regard, the applicant argues that the finding in recital 620 of the contested decision – that what matters is that Qualcomm was ready to incentivise the purchase of all units falling, a priori, within the scope of certain agreements providing for incentives – makes no sense. The application of US GAAP means that, at the time chipset orders were received and dispatched, Qualcomm was obliged to assume that all chipset units would qualify for each potentially applicable incentive, without allowing it the slightest uncertainty. That does not mean that Qualcomm accepted that all units would be granted the maximum incentive.

350 In the third place, the applicant submits that the Commission’s approach is inconsistent, in that the Commission, on the one hand, rejected Qualcomm’s accounting prices as ‘unsuitable’ and, on the other hand, selectively resorted to the use of those same prices in two crucial calculations, namely the calculation of the price of chipsets for which no ‘document packets’ existed and the calculation conducted in order to allocate incremental R&D costs for the purposes of the price-cost test.

351 In the fourth place, the applicant maintains that the Commission’s approach contradicts Decision C(2014) 7465 final adopted in the case of 15 October 2014 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.39523) – Slovak Telekom), in which the Commission concluded that it was preferable, with a view to legal certainty, to make assessments on the basis of the cost data as used by the undertaking in question rather than on the basis of data resulting from complex ex post adjustments and calculations.

352 In the fifth place, the applicant observes that the Commission’s methodology results in prices which, in certain instances, are demonstrably not the prices ‘actually paid’ by Qualcomm’s customers. For instance, there is no uncertainty as to the actual price paid by Huawei for backlog orders of [confidential] units of MDM8200 chipsets sold by Qualcomm, for which the agreed final, flat price was [confidential] (USD) per chipset, net of any incentives; in contrast, the price reconstructed by the Commission following its manipulations is USD [confidential] per chipset. Similarly, the price paid by Huawei in the first quarter of 2011 for the MDM8200 chipset was USD [confidential] per chipset, whereas the Commission concluded, on the basis of its incorrect methodology, that the average net sales price for that chipset was USD [confidential] per chipset.

353 In addition, the applicant argues that the Commission is not able to explain why the average sales price net of the Master Incentive in the first quarter of 2011 was inconsistent with the average sales price net of the Master Incentive quoted in the [confidential] slide: the two reformulated prices, which the Commission attempts to justify by relying on a slide in the [confidential] presentation deck dated 4 August 2010, are clearly not ‘actual’ prices.

354 The Commission disputes the arguments put forward by the applicant.

355 As regards the first complaint which, in essence, alleges that the ‘document packets’ containing the incentives claims submitted to Qualcomm by Huawei and ZTE were unsuitable for the purposes of conducting a reliable price-cost test, it is sufficient to observe, as the Commission did, that the documents concerned were used by the applicant itself when issuing billing credit notes and when establishing client account statements, as indicated in recital 597 of the contested decision, which was not disputed by the applicant. Therefore, in the absence of other evidence put forward by the applicant, the present complaint must be rejected.

356 As regards the second complaint, alleging that the Commission’s approach unduly introduced ex post hindsight, in that it transferred reserves based on the assumption that Qualcomm could have forecast the incentive claims with perfect certainty at the time of sale, it must be stated that, contrary to the applicant’s assertions, the Commission’s methodology is not based on parameters which Qualcomm was unable to assess. Qualcomm’s prices were structured as a gross price to which the incentives, that is to say discounts, were applied. Accordingly, when Qualcomm took pricing decisions, it knew what the lowest unit price could be if all incentives were applied. That calculation, as the Commission maintains, was intended simply to establish, on the basis of the information available to Qualcomm, on which the latter relied for drawing up its accounts, the price actually paid by Huawei and ZTE. Those considerations are sufficient to reject the present complaint.

357 As regards the third complaint, alleging that the Commission’s approach to Qualcomm’s prices is inconsistent, it is sufficient to reject the argument that it is based on the incorrect premiss that the contested decision found that Qualcomm’s accounting prices were ‘unsuitable’. In recitals 603, 613 and 618 of that decision, to which the applicant refers, the Commission clearly identifies which elements in those accounting prices were unsuitable, without, however, rejecting the possibility of using those elements after having made the necessary adjustments on the basis of other sources of information.

358 As regards the fourth complaint, alleging a contradiction with Decision C(2014) 7465 final of 15 October 2014 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.39523 – Slovak Telekom), it should be observed that, while it is true that, in certain cases, the Commission initially relied on the figures in the accounts of the dominant undertaking, which was sufficient on occasion, where those figures were not available or did not reflect market realities, it constructed appropriate representative values, also based on all relevant information provided by the dominant undertaking. As has been pointed out in paragraph 332 above, that approach has been endorsed by the Court. The present complaint must therefore be rejected.

359 As regards the fifth complaint, alleging that the Commission’s methodology results in prices which, in certain instances, are manifestly not the prices ‘actually paid’ by Qualcomm’s customers, it should be observed that, as regards the first example relied on by the applicant, the slide on which the Commission’s defence is based (Annex A.2.2.19 to the application) actually demonstrates that the price paid by Huawei for the MDM8200 chipset after the incentive was applied was USD [confidential] (rounded to USD [confidential]) per chipset which corresponds to the price calculated by the Commission and indicated in recital 743 of the contested decision. As regards the second example, it should be noted that the document referred to by the Commission (Annex A.2.4.12 to the application) to demonstrate that the prices approved by the [confidential] had undergone successive variations, as clarified in the Commission’s reply to questions put by the Court, confirms that the price communicated to Huawei for the MDM8200 chipset in the first quarter of 2011 was lower than that presented to the [confidential] and corresponded to the actual price of USD [confidential] as calculated by the Commission.

360 It follows that the applicant has not demonstrated that there were errors in the methodology used by the Commission and in the results arrived at by the Commission in the contested decision. In those circumstances, the present complaint must be rejected.

361 It follows that the second part of the fifth plea must be rejected. Therefore, since the first part of the fifth plea has also been rejected (see paragraph 344 above), that plea must be rejected in its entirety.

 The sixth plea: incorrect ‘allocation of non-recurring engineering expenses’

362 By the present plea, the applicant submits that, in its assessment of the NRE payments, the Commission committed ‘manifest errors of fact and of law, and breach[ed] the principle of legal certainty, the presumption of innocence and the principle in dubio pro reo’. In addition, the contested decision fails to provide a sufficient statement of reasons, including because it does not ‘address’ numerous important arguments made by Qualcomm during the administrative procedure. The applicant argues that the Commission thereby failed to fulfil its ‘duty of sound administration’.

363 The Commission disputes the arguments put forward by the applicant.

364 Pursuant to Article 21 of the Statute of the Court of Justice of the European Union and Article 76(d) of the Rules of Procedure, each application is required to state the subject matter of the proceedings and a summary of the pleas in law on which the application is based and it is necessary, in order for an action to be admissible, that the basic legal and factual particulars relied on be indicated, at least in summary form, coherently and intelligibly in the application itself.

365 That information must be sufficiently clear and precise to enable the defendant to prepare its defence and the General Court to rule on the action. The same considerations also apply to all claims, which must be accompanied by pleas and arguments enabling both the defendant and the Court to assess their validity (judgment of 13 December 2018, Slovak Telekom v Commission, T‑851/14, EU:T:2018:929, paragraphs 74 and 75 (not published)).

366 In the present case, it must be observed that, while the applicant’s arguments raised in the context of the present plea may be read as seeking to highlight alleged errors of assessment by the Commission concerning (i) the NRE payment to ZTE (first part) and (ii) the NRE payment to Huawei (second part), the applicant has not provided any information accompanying those alleged infringements enabling an assessment of their substance to be made. It follows that the applicant’s arguments alleging those infringements, in particular the infringement of the principle of legal certainty, infringement of the principle of the presumption of innocence and infringement of the principle in dubio pro reo, must be regarded as being inadmissible in that they are not accompanied by sufficient details to enable an assessment of their substance to be made.

367 Moreover, it should be recalled that, in accordance with the case-law, although Article 296 TFEU requires the Commission, when taking a decision in the context of the application of the competition rules, to state the factual matters justifying the adoption of a decision together with the legal considerations which have led to its adopting it, that provision does not require the Commission to discuss all the matters of fact and law which may have been dealt with during the administrative proceedings. The statement of the reasons on which a decision adversely affecting a person is based must allow the Court of Justice to exercise its power of review as to the legality of the decision and must provide the person concerned with the information necessary to enable him or her to decide whether or not the decision is well founded (judgment of 11 July 1985, Remia and Others v Commission, 42/84, EU:C:1985:327, paragraph 26). In the light of that case-law, the applicant cannot therefore legitimately rely on the fact that the Commission failed to respond specifically in the contested decision to all the observations of fact or of law which it had raised during the administrative procedure.

 The first part: the NRE payment to ZTE

368 In support of the first part of the sixth plea, after describing, without disputing, the chronology of the events relating, in particular to (i) the agreement regarding non-recurring engineering expenses concluded with ZTE and (ii) the NRE payment to ZTE, the applicant submits that the Commission erred in treating that payment as a per-unit rebate of USD [confidential] for MDM6200 chipsets purchased by ZTE in 2010, even though the approval of December 2009 was never formalised in an agreement or communicated to ZTE.

369 According to the applicant, if the Commission had not taken into account that per-unit rebate to ZTE, it would not have found predation in the price-cost test for MDM6200 chipsets.

370 In the first place, the applicant considers that the approval, at the end of 2009 and in May 2010, of a set of potential incentives for ZTE was part of a wider package of incentives for ZTE, which were not limited to the leading edge segment of the UMTS chipsets market.

371 In the second place, as regards the Commission’s conclusion that the NRE payment to ZTE was equivalent to a per-unit rebate for MDM6200 chipsets purchased by ZTE in 2010, the applicant puts forward the arguments set out below.

372 First, the applicant maintains that the terms of the approval changed between December 2009 and May 2010, following a considered reassessment, which meant that the NRE payment to ZTE could no longer be perceived as a rebate. It adds that the Commission treats the USD [confidential] lump-sum payment as if it were a rebate of USD [confidential] per unit, that is to say much higher than the rebate of USD [confidential] per unit allegedly foreseen by Qualcomm in December 2009.

373 Second, the Commission adduces no evidence that Qualcomm intended to offer, or was even aware that it was offering, to ZTE an incentive equivalent to a rebate of USD [confidential] per unit, or that ZTE was aware of such a per-unit rebate. It follows that the NRE payment to ZTE could not have incentivised ZTE to purchase MDM6200 chipsets in 2010. Conversely, the Commission itself provides evidence demonstrating that Qualcomm perceived the price of those chipsets for sale to ZTE to be higher than that which would have resulted from applying a USD [confidential] per-unit rebate.

374 Third, by treating the NRE payment to ZTE as a rebate received for each chipset purchased by ZTE, the contested decision arrived at an artificially inflated per-chipset rebate of USD [confidential], precisely because that payment was independent of the number of units sold and, at the same time, ZTE’s demand for MDM6200 chips in 2010 represented a minute fraction of the demand anticipated by Qualcomm in December 2009, which led the Commission to spread that USD [confidential] payment over a small fraction of the 1 250 000 units mentioned in the December 2009 emails. If the payment in question had actually incentivised ZTE to purchase more units of the MDM6200 chipset, the implied per-unit rebate would have been lower and a finding of below-cost pricing less likely.

375 Fourth, the applicant maintains that, although, in the contested decision, the Commission concedes that the inflated figure is the result of weak demand by ZTE for the MDM6200 chipset, the decision then speculates that that lower demand may have been one of the reasons why Qualcomm changed the terms of the approval in the [confidential] meeting of 24 May 2010. Qualcomm identified correspondence dating from May 2010, from which it is apparent that ZTE had forecast demand for 475 000 units of the MDM6200 chipset for that year. According to Qualcomm, applying that figure to the USD [confidential] payment leads to a hypothetical rebate of USD [confidential] per unit at the time of the May 2010 approval, and prices paid by ZTE for the MDM6200 chipset which are consistently above the contested decision’s LRAIC.

376 Fifth, even if Qualcomm had intended the NRE payment to constitute a per-unit rebate of USD [confidential], that would, in any event, have been economically irrelevant, since that payment never constituted an incentive for ZTE to purchase MDM6200 chipsets instead of chipsets supplied by a competitor, as the amount of the incentive was fixed and would not increase with the purchase of more chipsets.

377 Sixth, the fact that, in February 2011, Qualcomm acceded to ZTE’s request to extend the carrier qualification deadline for MDM8200A-based devices until 30 June 2011, but refused an equivalent request for MDM6200-based devices, demonstrates that the lump-sum payment was contractually and effectively linked to MDM8200A-based devices, which is also confirmed by Qualcomm internal correspondence.

378 In the third place, the applicant criticises an inconsistency in the contested decision between (i) the findings relating to predation and (ii) the finding of an increase in the price and margin for the MDM6200 chipsets sold to ZTE. According to the applicant, there is also an anomaly in the allocation of the alleged R&D costs between the MDM6200 and the MDM6600 chipsets in 2010, since the prices used in the Commission’s calculation for the MDM6200 chipsets sold to ZTE are not reduced by the NRE rebate.

379 The Commission disputes the arguments put forward by the applicant.

380 As a preliminary point, it should be stated that, in particular in recitals 677 to 693 of the contested decision, the Commission calculated the prices paid by ZTE and Huawei and, in accordance with the considerations set out in particular in Sections 12.4.2.3, 12.4.2.4, 12.4.2.6 and 12.4.2.11 of that decision, reallocated two rebates deriving from the non-recurring engineering expenses granted by Qualcomm to those two customers. According to the Commission, while each rebate was formally linked to the MDM8200A chipsets sold to those customers, contemporaneous evidence demonstrated that Qualcomm’s real intention was (i) to grant a lump-sum payment of USD [confidential] (initially USD [confidential]) to ZTE, to be applied to sales made in 2010, with the aim of encouraging ZTE to buy and develop solutions based on the MDM6200 chipset and (ii) to grant a retroactive rebate in the form of a lump-sum payment of USD [confidential] to Huawei in order to reduce the price of the MDM8200 chipset units which it had purchased from Qualcomm, but which had proved to be too costly to make it possible for Huawei to win downstream tenders.

381 As regards the NRE payment to ZTE, in order to reach its conclusion, the Commission, in Sections 12.4.2.4, 12.4.2.6 and 12.4.2.11 of the contested decision, refers to internal Qualcomm documents demonstrating, in its view, that Qualcomm was not in a position to offer ZTE a price lower than that offered to Huawei, and feared lowering at too early a stage the price of the MDM6200 chipset, which at that time had not yet been launched. Therefore, it was proposed, from December 2009, to deal with the question of setting prices via one-off rebates arising from non-recurring engineering expenses, which as to one part was subject to the condition that ZTE obtain carrier qualification for MDM8200A chipset devices and as to the other part was subject to the condition that ZTE obtain carrier qualification for MDM6200 chipset devices, in both cases prior to 31 December 2010 (a deadline which was then deferred to 30 June 2011 for the qualification for the MDM8200A chip). According to the Commission, the purpose of that payment was, in reality, to reduce the price of the MDM6200 chipset by USD [confidential] per chipset which Qualcomm intended to sell to ZTE in 2010. Following a significant decrease in ZTE’s orders compared to Qualcomm’s forecasts, the amount of the reduction actually granted was USD [confidential] per chipset.

382 In the first place, it should be stated that the fact that Qualcomm implemented a package of incentives to ZTE in relation to various chipsets, including chipsets not used for the leading edge segment of the UMTS chipsets market, does not mean that some of them cannot be taken into account for the purposes of the price-cost test conducted in order to assess predation.

383 In the second place, it is clear from the applicant’s description that the difference between the incentive approved in December 2009 and the incentive approved in May 2010 lies in the fact that the former incentive entailed (i) a payment of USD [confidential], to be made upon signature of the agreement, and (ii) a payment of USD [confidential], to be made subject to ZTE obtaining carrier qualification for MDM8200A chipset devices, while the second incentive entailed a payment of USD [confidential], payable on condition that ZTE obtain carrier qualification for MDM6200 chipset devices. However, the applicant does not explain why the fact that ‘the terms of approval changed from December to May demonstrates a considered reassessment’ or how that is favourable to it for the purposes of assessing predation.

384 Conversely, as described in recitals 501 and 502 of the contested decision, internal communications from December 2009 between senior directors (such as the QCT Product Management Director and the Vice-Presidents of Finance and of Sales), report that, in the first place, the prices offered to ZTE for the MDM6200 chipset could not be lower than those offered to Huawei, and that Qualcomm was not in favour of making ‘big price moves’; in the second place, Qualcomm feared losses for the MDM6200 chipset resulting from the collaboration between ZTE and Icera and, in the third place, the envisaged solution was an NRE payment tied to MDM6200 or MDM8200A chipsets, to prevent ‘Icera gain[ing] a lot of business’. The solution to that concern was identified in the ‘ZTE MDM6200 Price Proposal’ dated December 2009 which, to some extent, was not a ‘big price move’ in that it was not a price move but rather an incentive, and which included a payment tied to carrier qualification for the MDM8200A chipset and the purchase of a number of units of the MDM6200 chipset, with the possibility of clawing back payment if one or other of the conditions were not met.

385 Moreover, it is apparent from the presentation to the [confidential] meeting of 8 February 2010, which therefore post-dated the December 2009 approval, that the payment which, at the time, was supposed to be formally envisaged for both the MDM8200A chipset and the MDM6200 chipset, was in fact intended solely to reduce the price of the latter chipset.

386 In addition, it should be noted, as the Commission did, that the condition initially proposed, requiring ZTE to purchase [confidential] units of the MDM6200 chipset in 2010 (corresponding to ZTE’s demand forecasts at the time), was not formalised in the [confidential]’s approval. However, in May 2010, taking into account the fact that ZTE’s demand forecasts had fallen to 475 000 units, the [confidential] modified the NRE payment structure. In addition, contrary to the applicant’s assertions, there is nothing to indicate that those modifications corresponded to a change in the intention underlying those payments which was to encourage ZTE to purchase MDM6200 chipsets. Indeed, if the initial objective had been, in December 2009, to promote the MDM8200A chipset, or if that was the objective in May 2010, it would not have been necessary or logical to reduce the rebate for MDM8200A chipsets from USD [confidential] to USD [confidential] (a 60% decrease). By contrast, having significantly revised ZTE’s forecasted demand for MDM6200 chipsets from [confidential] units to fewer than [confidential] units, it was no longer justified, from Qualcomm’s point of view, to offer to ZTE the initially planned rebate of USD [confidential].

387 The modification of the NRE payment structure nevertheless made it possible to maintain a payment equivalent to a rebate of USD [confidential] per unit formally tied to MDM8200A chipsets (but actually intended for MDM6200 chipsets), while also offering to double the payment (which would therefore reach USD [confidential]) if the customer obtained carrier qualification for the MDM6200 chipset before the end of 2010, that is to say, if ZTE stimulated the development of an MDM6200 solution during 2010. The Commission correctly states that Qualcomm’s refusal to extend the carrier qualification deadline for MDM6200 chips in February 2011 is consistent with that structure and Qualcomm’s conclusion that ZTE had not done enough to promote the MDM6200 chipset in 2010 to merit receiving the remainder of the rebates resulting from non-recurring engineering expenses.

388 As regards the applicant’s argument that the fact that Qualcomm refused in February 2011 to extend the deadline for carrier qualification for devices containing an MDM6200 chipset demonstrates that the NRE payment to ZTE was contractually, and actually, tied to the MDM8200A chipset, it must be pointed out that that circumstance and, consequently, the fact that Qualcomm did not pay ZTE the USD [confidential] amount relating thereto, are perfectly consistent with Qualcomm’s initial intention to prescribe a clawback mechanism for part of the NRE payment if ZTE did not purchase the expected volumes of MDM6200 chipsets.

389 As regards the NRE payment to ZTE being treated as a per-unit discount, it must be observed that there can be no finding of error on the part of the Commission as regards the accounting treatment of that payment to the applicable units, or as regards the calculation of the rebate or the assessment of the context.

390 First, in order to be able to quantify the size of the NRE payment to ZTE, any payment relating to the MDM6200 chipsets ordered and delivered, namely 145 775 units, must be accounted for by dividing the total amount of the payments by the total number of units ordered and purchased.

391 The fact that the NRE payment to ZTE ultimately gives rise to a per-unit rebate three times higher than that initially provided for results simply from the significant reduction in the quantity of MDM6200 chipsets ordered and purchased by ZTE compared with Qualcomm’s forecast, namely [confidential] units. Since the amount of the incentive had been established a priori, each chipset benefited from a much larger price reduction than originally planned. In other words, that increase in the rebate applied to each chipset is no more than a by-product of Qualcomm’s incorrect sales forecast regarding ZTE.

392 As regards the fact that ZTE was not aware that the NRE payment intended for it was to be treated as a per-unit rebate, it must be stated that the case-law does not require the customer of the undertaking in a dominant position to be aware of the methodology used to arrive at a price below cost being applied. It is sufficient, on the one hand, that the price-cost test reveals a price higher than the measure of cost chosen and, on the other hand, that there is an intention to eliminate the targeted competitor (see, to that effect, judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 71 and 72).

393 In addition, as regards the argument that the payment did not incentivise ZTE to purchase more chipsets, it should be noted that in the judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 53 and the case-law cited), the Court of Justice held that the characterisation of a practice of a dominant undertaking as abusive does not mean that it is necessary to show that the result of a practice of such an undertaking, intended to drive its competitors from the market concerned, has been achieved and, accordingly, to prove an actual exclusionary effect on the market. The purpose of Article 102 TFEU is to penalise abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it, irrespective of whether or not such abuse has proved successful.

394 In the third place, the applicant provides only a bare outline of its argument alleging inconsistencies in the contested decision, such that it is difficult to understand how such inconsistencies could affect the lawfulness of that decision, with the result that that argument must be declared inadmissible. In any event, the explanation given by the Commission, namely that the increase in the price and margin of the MDM6200 chipsets sold to ZTE during the relevant period is a consequence of the discounts stemming from NRE payments, is entirely consistent with the remainder of its reasoning. As regards the alleged anomaly in the allocation of the alleged R&D costs between the MDM6200 and MDM6600 chipsets in 2010, it is sufficient to note that recital 841 of that decision provides a detailed and credible explanation of the reasons which prompted the Commission to do so, reasons which, moreover, the applicant disputes only in generic terms.

395 In the light of the foregoing, the first part of the sixth plea must be rejected.

 The second part: the NRE payment to Huawei

396 According to the applicant, even if it had intended the NRE payment to Huawei to act as ‘compensation’ for the inventory of MDM8200 chipsets which Huawei had accumulated, that payment pertained to chipsets already purchased and could not therefore have incentivised Huawei to make further purchases of those chips. In addition, as is also apparent from Huawei’s response to a Commission request for information, the terms of the non-recurring engineering expenses agreement concluded with Huawei and that payment, linked to carrier qualification of MDM8200A-based chipset devices, did not incentivise Huawei to purchase such chipsets, but instead to purchase MDM8200A-based chipsets, for an indefinite period.

397 More generally, the applicant criticises the Commission for having removed from recital 1142 of the contested decision the sentence containing an explicit ‘quasi-allegation’ of ‘obfuscation’ introduced into the SSO with regard to the NRE payments to ZTE and Huawei, which Qualcomm had addressed during the administrative procedure, while the remainder of the recital concerning those NRE payments remained unchanged. In its view, that removal constitutes a failure to provide a proper statement of reasons and further illustrates the Commission’s failure to establish Qualcomm’s intention to exclude Icera from the market.

398 The Commission disputes the arguments put forward by the applicant.

399 As a preliminary point, it should be noted that, as regards the NRE payment to Huawei, the Commission refers, in Section 12.4.2.3 of the contested decision, to internal Qualcomm documents, which, in its view, demonstrate that, since a number of senior directors of Qualcomm were opposed to giving in to the pressure from Huawei, which was requesting significant price reductions on MDM8200 chipsets, Qualcomm eventually granted a retroactive price reduction for shipments of the MDM8200 chipset after 23 November 2009, as well as ‘MDM8200A NRE’, that is to say, a payment officially intended to cover non-recurring engineering expenses incurred in connection with the certification of a specific end device incorporating the MDM8200A chipset, in the amount of [confidential] (subsequently USD [confidential]). According to the Commission, although the NRE payment was formally associated with obtaining carrier qualification for the latter chipset, the NRE payment was clearly linked to Huawei’s stock of the MDM8200 chipset and enabled a retroactive per-unit price reduction to be applied.

400 In the first place, the applicant’s argument – that Huawei had already purchased the MDM8200 chipsets and could not therefore have been incentivised by the NRE payment which was intended for it to purchase more of them – cannot succeed. It is apparent from the documents and correspondence referred to, in particular, in recitals 480 to 486 of the contested decision that that incentive was granted to Huawei in order to assist it in disposing of an excess of those chipsets, in a context in which Huawei relied increasingly on the competitive pressure exerted by Icera, in particular via the ICE8042 chip, in order to obtain more favourable conditions.

401 In addition, it is apparent from the documents and correspondence referred to in particular in recitals 480 to 486 of the contested decision that Qualcomm was not in a position to ignore Huawei’s requests, which demonstrates that price negotiations were still taking place when Qualcomm decided to grant the NRE payment to Huawei, and that the conditions of sale of MDM8200 chipsets had, at that time, not yet been finally determined.

402 As regards the applicant’s argument based on the fact that the terms of the non-recurring expenditure agreement and the NRE payment to Huawei did not incentivise Huawei to purchase MDM8200 chipsets, but rather MDM8200A chipsets, it must be stated, as the Commission did, that Huawei’s responses to the Commission’s questions do not demonstrate that Huawei believed that payment to relate to MDM8200A chipsets. On the contrary, as stated in recital 681 of the contested decision, Huawei’s response simply repeats the terms of that agreement, which are not evidence concerning the true purpose of that payment. It is apparent from the evidence referred to in recitals 482 and 679 of that decision, and from Qualcomm’s own admission in the SO response, that that payment was to serve as a retroactive rebate for the MDM8200 chipsets. In addition, Qualcomm’s assertion that Huawei believed the payment in question to relate to the MDM8200A chipsets is contradicted by the contemporaneous evidence referred to in recitals 491, 527 and 543 of that decision, in which Huawei requested Qualcomm to modify rebates stemming from non-recurring engineering expenses in order to take account of additional stock units for MDM8200 chipsets which it had identified. It is apparent from those elements that Huawei was perfectly aware that the true intention underlying the NRE payment was that it was to act as a rebate for MDM8200 chipsets and not for MDM8200A chipsets.

403 Finally, as regards the explicit ‘quasi-allegation’ of ‘obfuscation’ introduced in the SSO in relation to the NRE payments to ZTE and Huawei, it is true, first, that the last sentence of paragraph 771 of the SSO asked whether the fact that Qualcomm had not properly accounted for non-recurring engineering expenses for chipsets was an attempt to conceal the rebate tied to the payment of those expenses, and, second, that that question was not set out in the contested decision. However, the applicant’s criticism – concerning the removal of that ‘quasi-allegation’, according to which that removal constitutes a failure to provide a proper statement of reasons and also illustrates the Commission’s failure to establish Qualcomm’s intention to exclude Icera from the market – cannot succeed.

404 It is sufficient to note in that regard that, in the light of the case-law cited in paragraph 307 above, the Commission was not required to respond specifically in the contested decision to all the observations of fact and law which it had raised during the administrative procedure. Moreover, as the Commission submits, the removal in the contested decision of the last sentence of paragraph 771 of the SSO shows, contrary to the applicant’s assertions, that the Commission took account of Qualcomm’s response during the administrative procedure, without, however, that having affected the evidence of Qualcomm’s intention to exclude Icera from the relevant market.

405 In the light of the foregoing, the second part of the sixth plea must be rejected. Therefore, since the first part of the sixth plea has also been rejected (see paragraph 395 above), that plea must be rejected in its entirety.

 The seventh plea: failure ‘to establish an appropriate cost benchmark’

406 By the seventh plea, which is divided into three parts, the first alleging infringement of the principle of legal certainty and the principle in dubio pro reo, the second alleging that LRAIC is not the appropriate cost benchmark and the third alleging that the LRAIC calculated by the Commission is not ‘true’ LRAIC, the applicant submits that the treatment of costs in the contested decision is vitiated by numerous flaws and breaches of general principles of law.

 Preliminary observations

407 In Section 12.6 of the contested decision, the Commission explained, in relation to the chipsets under investigation, the reasons why it had concluded that the most appropriate cost benchmark for the price-cost test was LRAIC. In the calculation of LRAIC, it took into account (i) manufacturing costs, which vary according to the quantity of chipsets produced and thus represent the relevant variable component of LRAIC (Section 12.6.2 of that decision) and (ii) Qualcomm’s R&D costs as recorded in the Qualcomm [confidential], thereby capturing the most important fixed portion of the incremental production cost of a chipset (Section 12.6.3 of that decision). It added that other types of fixed costs, such as commercialisation costs, were not included, with LRAIC calculated by the Commission therefore being below Qualcomm’s ATC.

408 According to the Commission, having regard to the criteria established by the EU Courts in that field, LRAIC was the most appropriate cost measure in the present case in order to calculate the minimum required rate of recovery of costs by Qualcomm for the relevant products. In its view, since Qualcomm is a multi-product undertaking benefiting from economies of scope, a number of products could share the same production activities, with the result that the relevant costs do not vary with the number of products supplied (‘common costs’). It follows, according to the Commission, that those costs are not taken into account in LRAIC, since LRAIC only comprises the production costs specific to the products under investigation. Therefore, the average of all variable and fixed costs incurred by Qualcomm to produce a particular product (namely LRAIC) is below ATC for each product.

409 The Commission’s view in the contested decision is that LRAIC being below ATC for each product is not called into question by the fact that each chipset was likely to generate, or to have received, R&D spillovers for future chips. According to the Commission, the spillovers from which a particular chip benefited are likely to be roughly balanced by the spillovers generated in turn by that particular chip for other later chips. It stated that the R&D costs incurred for a chip were not, therefore, discounted by the spillovers likely to have been generated by that chip in favour of a later chip. However, with regard to the MDM8200 and MDM8200A chipsets, the Commission adjusted its allocation of development costs to account for the fact that evidence in the file indicated that the MDM8200 chipset generated substantially more spillovers for the MDM8200A chipset than it had itself received.

410 In practice, in order to calculate the AVC, the Commission relied on the ‘average unit cost’ parameter taken from Qualcomm data which could be presented according to two accounting criteria: one which reflected the average unit cost of chips sold in a given quarter and a second which reflected the unit cost of the chips purchased by Qualcomm (from foundries) in a given quarter. Initially, in the SO, the Commission used the first approach. Subsequently, in the SSO and the contested decision, in order to take account of certain observations made by Qualcomm, the Commission instead used the second approach, subject to certain adjustments, aimed primarily at resolving an inventory accounting issue.

411 By contrast, in order to calculate Qualcomm’s R&D costs to be attributed to each chipset, the Commission identified certain cost elements in the [confidential] as being ‘incremental’ to that chipset, which Qualcomm used to record and allocate certain fixed costs to certain chips. In particular, the ‘incremental’ costs were identified based on an internal Qualcomm document in which that term was used, namely an internal presentation entitled [confidential], concerning the [confidential] (‘the [confidential]’), and the explanations given by Qualcomm during the administrative procedure regarding the categories of costs it regarded as ‘incremental’ in that document.

 The first part: infringement of the principle of legal certainty and the principle in dubio pro reo

412 The applicant submits that the Commission infringed the principle of legal certainty, the principle of the presumption of innocence and the principle in dubio pro reo by using two different approaches in order to quantify Qualcomm’s total R&D investment, namely a ‘top down’ approach in the SO and a ‘bottom-up’ approach in the SSO and the contested decision. In particular, the Commission acknowledged in the SO that the [confidential] was unsuitable for use in a price-cost test. By contrast, in the SSO and the contested decision, the Commission used data from that database and the methodology employed led to very different margins between chipsets and from one quarter to another relative to the SO.

413 The applicant also maintains that the difference between the methodology used in the SO, on the one hand, and the methodology used in the SSO and the contested decision, on the other hand, meant that it was unable to foresee the methodology which would ultimately be used by the Commission or its results when it made the relevant chipset pricing decisions or, a fortiori, R&D investment decisions. Indeed, the Commission itself did not anticipate even the fundamentals of the methodology ultimately employed in that decision, or its results, after over half a decade of investigation.

414 The Commission disputes the arguments put forward by the applicant.

415 As a preliminary point, it should be observed, as was done in paragraphs 364 to 366 above, that the applicant’s complaints alleging infringement of the principle of the presumption of innocence and of the principle in dubio pro reo must be regarded as being inadmissible in that sufficient details are not provided to enable an assessment of their substance to be undertaken.

416 Furthermore, in the first place, as regards the applicant’s argument concerning, in essence, the difference between the analysis methodologies used by the Commission in the SO, on the one hand, and in the SSO and the contested decision on the other, it should be recalled that, in accordance with the case-law cited in paragraph 146 above, until a final decision has been adopted, the Commission may, in view, in particular, of the written or oral observations of the parties, abandon some or even all of the objections initially made against them and thus alter its position in their favour or decide to add new complaints, provided that it affords the undertakings concerned the opportunity of making known their views in that respect. The same is true of the methodology used by the Commission in its price-cost test.

417 It should be recalled that, according to settled case-law, while the principle of legal certainty requires that legal rules be clear and precise, and aims to ensure that situations and legal relationships governed by EU law remain foreseeable (see judgments of 15 September 2005, Ireland v Commission, C‑199/03, EU:C:2005:548, paragraph 69 and the case-law cited, and of 29 March 2012, Spain v Commission, T‑398/07, EU:T:2012:173, paragraph 107), the fact remains that it follows from the provisional nature of the methodology used in the SO in order to determine the most appropriate measure of costs for the price-cost test that the Commission’s final decision cannot be annulled on the sole ground that the results arising from applying another methodology in the SSO and in the contested decision are not an exact match for the results which would have arisen from the methodology initially used in the SO (see, by analogy, judgment of 5 December 2013, SNIA v Commission, C‑448/11 P, not published, EU:C:2013:801, paragraph 43).

418 In addition, it must be noted that, in the present case, the applicant does not deny that it had the opportunity to explain its views during the administrative procedure and, in particular, after the SSO was sent, on the methodology to be applied in the contested decision.

419 Moreover, the applicant simply refers to the ‘top down’ and ‘bottom up’ nature of the methodologies used by the Commission and to a difference between the margins required between the chips and from one quarter to another, without explaining, except by making a reference to its written submissions submitted during the administrative procedure and annexed to the application, the specific differences between the ‘top down’ and ‘bottom up’ approach and their practical implications. In that regard, it must be stated, as the Commission did, that the methodology adopted in the contested decision (and in the SSO) is based on the actual R&D costs which Qualcomm itself posted in its accounts for each chipset. For that reason, the Commission concluded that that methodology offered a more accurate reflection of the R&D costs actually incurred in developing each chipset than the methodology applied in the SO.

420 Consequently, even if the analysis methodologies used by the Commission in the SO, on the one hand, and in the SSO and the contested decision, on the other, were to produce different results, as the applicant maintains, that does not permit the inference in the present case that the Commission erred in adopting in the contested decision a methodology based on actual R&D costs which Qualcomm itself had posted in its accounts for each chipset. In those circumstances, that argument cannot succeed.

421 In the second place, as regards the applicant’s argument that it was impossible for it to know, when making its decisions on chipset pricing and R&D investments, the methodology which would be used by the Commission in order to determine the relevant cost benchmark, put forward solely in support of the alleged infringement of the principle of legal certainty, it should be recalled that, in the SSO, the Commission had stated that the modification of the LRAIC calculation had been made in order to take account of the objections raised by the applicant in its SO response concerning the LRAIC calculation made in the SO.

422 It should be noted in that regard that, while it is true that the account taken of the costs of the dominant undertaking allows that undertaking, in the light of its special responsibility under Article 102 TFEU, to assess the lawfulness of its own conduct, and is therefore consistent with the general principle of legal certainty (see, to that effect, judgments of 14 October 2010, Deutsche Telekom v Commission, C‑280/08 P, EU:C:2010:603, paragraph 202, and of 17 February 2011, Teliasonera Sverige, C‑52/09, EU:C:2011:83, paragraph 44), that does not mean it is impossible for the Commission to make certain adjustments, based on the price and cost statements of the dominant undertaking and on all other relevant information provided by the dominant undertaking (see, to that effect, judgments of 30 January 2007, France Télécom v Commission, T‑340/03, EU:T:2007:22, paragraphs 131 to 137; of 10 April 2008, Deutsche Telekom v Commission, T‑271/03, EU:T:2008:101, paragraphs 208 to 211; and of 13 December 2018, Slovak Telekom v Commission, T‑851/14, EU:T:2018:929, paragraphs 220 to 235).

423 It follows that the principle of legal certainty does not require the dominant undertaking to have detailed forecasts relating to the exact methodology to be used by the Commission in order to calculate its costs. The methodology adopted by the Commission must take account of the particular circumstances of the case and, in particular, the available information provided by the dominant undertaking.

424 In the light of the foregoing, the first part of the seventh plea must be rejected.

 The second part: LRAIC is not the appropriate cost benchmark

425 According to the applicant, the Commission ought to have concluded that the most appropriate cost benchmark for the purposes of the price-cost test in the present case was AVC or average avoidable costs (‘AAC’), and not LRAIC.

426 In the first place, it argues that pricing decisions are based on competitive circumstances existing at the time when Qualcomm’s R&D costs had already become ‘sunk’, in the sense that they could no longer be avoided. Prices below LRAIC will, according to the applicant, often be profit maximising in the short term and not imply profit sacrifice.

427 In the second place, the applicant argues that LRAIC is designed for a static world, since it focuses on a given product and does not take account of past and future related products. In a dynamic sector such as the semiconductors sector, with inter-temporal effects, their use is not appropriate.

428 In the third place, the extremely narrow allegations of abuse in the present case (relating to the sale of three chipsets to only two customers for only certain quarters) means that true LRAIC is in fact AAC or AVC, or extremely close thereto.

429 In that regard, the Commission erred in concluding that products belonging to the relevant market were characterised by low marginal or variable costs, whereas the Commission had itself, on a number of occasions, stated that AVC was an important parameter of competition.

430 In the fourth place, the applicant submits that the Commission’s decision to base its analysis on LRAIC rather than on ATC is not, contrary to the Commission’s assertion, more favourable to the applicant. According to the applicant, it is unlikely that ATC, correctly calculated, would be higher than LRAIC, since LRAIC also includes common costs, fails to take into account the revenues derived from its patent licensing business and has undergone post hoc manipulations, resulting in LRAIC of the MDM8200A chip being significantly inflated.

431 The Commission and the intervener dispute the applicant’s arguments.

432 It should be recalled, first of all, that when an action for the annulment of a decision applying Article 102 TFEU is brought before it under Article 263 TFEU, the Court must as a general rule undertake, on the basis of the evidence adduced by the applicant in support of the pleas in law put forward, a full review of the question whether or not the conditions for the application of that provision are met and that, in carrying out such a review, the Court cannot use the Commission’s margin of discretion, by virtue of the role assigned to it in competition policy by the EU and FEU Treaties, as a basis for dispensing with the conducting of an in-depth review of the law and of the facts (see, by analogy, judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraphs 155 and 156).

433 Furthermore, it should be observed that the Commission’s use of a methodology based on prices above AVC and below ATC simply follows from the case-law. According to the Court of Justice, prices below AVC by means of which a dominant undertaking seeks to eliminate a competitor must be regarded as abusive. A dominant undertaking has no interest in applying such prices except that of eliminating competitors so as to enable it subsequently to raise its prices by taking advantage of its monopolistic position, since each sale generates a loss, namely the total amount of the fixed costs (that is to say, those which remain constant regardless of the quantities produced) and at least part of the variable costs relating to the unit produced. Moreover, prices below ATC, including fixed costs and variable costs, but above AVC, must be regarded as abusive if they are determined as part of a plan for eliminating a competitor. Such prices could drive from the market undertakings which are perhaps as efficient as the dominant undertaking but which, because of their smaller financial resources, are incapable of withstanding the competition waged against them (judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 71 and 72).

434 That case-law has been confirmed in a number of subsequent judgments which have consistently held that, first, prices below AVC must in principle be regarded as abusive in so far as, in applying such prices, an undertaking in a dominant position is presumed to pursue no economic purpose other than that of eliminating its competitors. Second, prices below ATC but above AVC are only to be considered abusive if they are determined in the context of a plan having the purpose of eliminating a competitor (judgments of 14 November 1996, Tetra Pak v Commission, C‑333/94 P, EU:C:1996:436, paragraph 41, and of 2 April 2009, France Télécom v Commission, C‑202/07 P, EU:C:2009:214, paragraph 109).

435 In the light of the case-law resulting from the judgment of 3 July 1991, AKZO v Commission (C‑62/86, EU:C:1991:286, paragraphs 71 and 72), the Commission, in order to find an abuse of a dominant position, was therefore obliged in the present case, first, to find that the applicant’s prices were below ATC and, second, to prove it intended to exclude a competitor. The Commission found in the contested decision, as recalled in paragraphs 408 and 409 above, that Qualcomm’s LRAIC was, for each product, below its ATC.

436 In that regard, contrary to the applicant’s assertions in its fourth complaint, it must be held that, since ATC includes, inter alia, all common costs, whereas LRAIC includes only the costs associated with the specific products concerned, it cannot, in essence, be accepted that ATC is below LRAIC. Even if the Commission erred in regarding certain common costs as specific to the products in question, the applicant cannot legitimately maintain that LRAIC could be above ATC.

437 Indeed, it must be stated that the Commission calculated ATC in Section 12.7.5.2 of the contested decision and concluded that it was above LRAIC. Qualcomm has neither disputed that point nor proposed any other calculation methodology. Therefore, the Commission cannot be criticised for having used LRAIC instead of ATC for the purposes of its price-cost test in so far as LRAIC is more favourable to the applicant than ATC. Moreover, as the intervener correctly points out, it was not necessary for the Commission to determine whether the applicant’s prices were also below AVC or LRAIC, since the Commission chose to ascertain whether Qualcomm intended to exclude a competitor. It follows that the applicant’s fourth complaint cannot succeed.

438 As regards the other three complaints put forward by the applicant disputing the appropriateness of the LRAIC as a cost benchmark, it must be stated, as the Commission did, that a price calculation based solely on variable costs is unsuitable for identifying predatory prices in a sector where, as is apparent in particular from recitals 109 to 119 and 280 to 284 of the contested decision, R&D activity and the importance of intellectual property rights generate high R&D costs, which would not be taken into account in a calculation based solely on variable costs.

439 In particular, it is not disputed that, as the Commission indicated in recital 787 of the contested decision, the semiconductor industry is characterised by low variable costs (for example, the costs of manufacturing chipsets) and high fixed costs (for example, the costs of R&D investment necessary to design and develop chipsets) which are mostly sunk by the time the products are commercialised. Therefore, the failure to include product-specific sunk costs, such as R&D investments, does not reflect the market reality for the costs associated with entering the market and competing on it, thereby making it very difficult, if not impossible, to detect predation aimed at eliminating a competitor.

440 Since LRAIC covers the fixed and variable costs specific to each product incurred both before and during the period when the abusive conduct took place, it is, therefore, the most appropriate cost benchmark in the present case in order to calculate the minimum required rate of recovery of costs for the products under investigation (recital 780 of the contested decision).

441 Moreover, the applicant’s argument that selling at a price above AVC but below LRAIC can be profit maximising in the short term cannot succeed. It should be noted, as the Commission did, that the applicable case-law recognises that pricing above AVC but below ATC (which, in the present case, is above LRAIC) is abusive only if done with the intention to exclude a competitor (see, to that effect, judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 71 and 72). As recalled in recital 785 of the contested decision, the Commission demonstrated exactly that. First, the applicant charged prices below LRAIC and, second, it did so in the context of a plan intending to exclude Icera. Therefore, the argument that prices above AVC could be used for legitimate purposes is simply a recognition of the distinction made in the applicable case-law and does not demonstrate that LRAIC would be an inappropriate benchmark in the present case.

442 As regards the applicant’s assertion that LRAIC is an inappropriate reference in the present case because of the limited scope of the infringement, it must be stated that, in relation to a two-year predatory practice in a sector which, as the applicant itself confirmed (see, for example, paragraphs 333 and 337 of the applicant’s SSO response, Annexes A.2.2 and A.2.4 to the application and paragraphs 715 to 723 of the applicant’s SO reply), is characterised by short innovation cycles requiring significant R&D investments, to ignore all costs associated with the development of the products which are the subject of the alleged predation would amount to ignoring a significant part of the costs which determine the pricing decisions of both the dominant undertaking and those of its rivals. If it was true, as the applicant maintains, that in ‘R&D intensive’ sectors, undertakings could set optimal prices at levels which do not permit recovery of all R&D costs which could be associated with a certain product, the profitability of their businesses would then be seriously challenged. However, those costs play an important role in the strategy for defining prices, particularly in the case of markets characterised by a rapid succession of products, since relatively old products are rapidly succeeded by more innovative products within relatively short periods of time. The fact that an R&D activity may benefit from previous results or facilitate future developments or discoveries does not mean that those investments cannot be taken into account, but rather means that costs must be allocated carefully between the various products.

443 In the light of those considerations, the applicant’s complaints disputing the appropriateness of LRAIC as a cost benchmark must be rejected. Indeed, while the Commission has a margin of discretion in competition policy, which is subject to in-depth review of the law and of the facts by the General Court (see, to that effect, judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraph 156), the fact remains that the evidence adduced by the applicant does not permit the inference that, in choosing not to use AVC or AAC as the cost benchmark, the Commission erred.

444 In the light of all of the foregoing, the second part of the seventh plea must be rejected.

 The third part: LRAIC calculated by the Commission is not ‘true’ LRAIC

445 In support of the third part of the plea, the applicant relies on eight complaints.

446 In the first place, the applicant complains that the Commission erred in its definition of LRAIC, in that it failed to take account of the fact that, where there were spillover benefits from one chip to another, the costs liable to be so classified could not, by their non-specific nature, be attributed to a particular chipset and could not therefore simply be included in the calculation of LRAIC.

447 In the second place, the applicant complains of an inconsistency between the justifications given by the Commission for its calculations in the SSO and the justifications for its calculations in the contested decision. According to the applicant, in the SSO, the Commission refers to the need to resolve the ‘accounting-induced distortions’ identified in the SO response, whereas, in the contested decision, the Commission refers rather to the need to address the concerns raised. In particular, the applicant criticises the fact that (i) the AVC reconstruction by the Commission in that decision is excessively complex and (ii) the Commission did not use average unit costs (AUC) in its calculations, contrary to the methodology the applicant had defended in its replies to the SO and the SSO.

448 In the third place, the applicant maintains that the [confidential] is unsuitable for identifying its incremental R&D costs, in that that [confidential] is used solely for internal management and accounting purposes and is not designed or suitable for the purposes of the Commission’s analysis.

449 In the fourth place, the applicant complains that the Commission treated certain costs recorded in the [confidential], namely the [confidential] and [confidential] cost categories, as ‘incremental’ costs, even though they were not, since they were associated with a given chip solely as a result of [confidential] allocations, and not because they related specifically to that chip.

450 In that regard, the applicant argues, first, that a distinction must be made between costs allocated directly to a chip and those allocated ex post, that is to say, not because they were considered to be specifically linked to that particular chip, but solely because of a simple internal allocation mechanism. According to the applicant, QTC employees [confidential].

451 Accordingly, hardware-related costs not recorded directly against a chip, namely, [confidential] in the [confidential] category, [confidential].

452 Similarly, from September 2009, software-related costs are the result [confidential].

453 Due to the specific nature of R&D in the semiconductor sector, even the costs recorded in the [confidential] directly against a chip may not be truly specific to that chip. In any event, even if the Commission had regarded such directly recorded R&D costs as ‘incremental’ or ‘product-specific’, it would not have been able to find predation.

454 In practice, the Commission provides no robust justification for classifying the [confidential] and [confidential] cost categories as ‘incremental’ and simply relies on (i) the use of the term ‘incremental’ in a single Qualcomm document in its possession, namely the [confidential], and (ii) the explanations given by Qualcomm during the administrative procedure, which confirm that the Commission did not conduct any independent assessment.

455 Second, the applicant submits, as regards the [confidential] used by the Commission to identify the category of incremental costs within the cost categories borne by the applicant, that that document concerned a specific chipset, the [confidential], not covered by the investigation, which was not developed at the same time as the investigated chipsets and which, unlike them, was not a ‘thin’ modem. However, a number of documents in the Commission’s possession, including documents relating to the chipsets covered by the investigation, make no reference to incremental costs.

456 In addition, the [confidential], which dates from April 2011, post-dates the major R&D investment decisions for the chips in question.

457 Moreover, in recital 847 of the contested decision, the Commission ignores and misconstrues the [confidential], where it contains a reference to ‘incremental R&D’. According to the applicant, the Commission ought also to have taken account of the reference, in the same document, to the concept of ‘incremental portfolio margin’, which would have enabled a better understanding of the term ‘incremental’. Similarly, the applicant claims that it never stated, as the Commission asserts in recital 848 of the contested decision, that the expression [confidential] was of more general application within Qualcomm than in that document.

458 Third, the applicant submits that it has, on a number of occasions, explained that more than USD [confidential] of the USD [confidential] incorrectly recorded in the [confidential] and [confidential] cost categories, which the Commission regarded as ‘incremental’ for the MDM8200 and MDM8200A chipsets, in fact relate to the development of [confidential], which is definitely not specific to any chip or chipsets. In that regard, [confidential] notes that [confidential], which demonstrates that that USD [confidential] is not specific to a given chip.

459 In addition, according to the applicant, the justifications given in the contested decision for rejecting its arguments in that regard are implausible.

460 In the fifth place, the applicant complains that the Commission erred in disregarding the revenues generated by Qualcomm through its patent licensing business. In essence, its view is that the Commission ought to have excluded the R&D costs associated with the chipsets in question in so far as those costs stemmed from obtaining patents which Qualcomm had included in its licensed patent portfolio and which could not therefore be regarded as specific to the chips concerned. According to the applicant, the Commission erred in concluding, first, [confidential] and, second, that the incremental revenues generated by those licences were unlikely to have a measurable impact on the revenues generated by that activity.

461 The applicant observes that the Commission, however, itself, stated in footnote 1238 to the contested decision, that [confidential]. It adds that the relevant question is not whether the revenue generated by patent licensing activity is likely to have a measurable impact or not on the revenue generated through Qualcomm’s technology licensing business. According to the applicant, the relevant question is rather whether the R&D is truly ‘incremental’ to the three chipsets in question and whether it results in patents which are then licensed, with the result that that R&D is not specific to those chipsets.

462 According to the applicant, it is also not possible to draw a direct link between the chipsets in question and software-related patents, since the software-related costs are not recorded directly against a specific chip.

463 In the sixth place, the applicant maintains that the contested decision relies on an extract from the [confidential] which does not link any costs therein to any specific customers. Nevertheless, the Commission finds that R&D costs associated with the development of the relevant chipsets were ‘incremental’ costs for Huawei and ZTE. According to the applicant, a price-cost test targeting only Huawei and ZTE ought, however, to exclude all R&D costs, which militates further in favour of using AVC or AAC.

464 In particular, first, according to the applicant, the fact that sales to customers other than ZTE and Huawei would not have justified developing the chipsets in question does not mean that sales to Huawei and ZTE would have been sufficient to justify the development of those chipsets. Second, the Commission itself acknowledged, in recitals 131, 132 and 354 of the contested decision, that the MDM8200 chipset had been developed in cooperation with operators other than ZTE and Huawei. Third, data for sales of the MDM8200A chipset to ZTE and Huawei, as set out in recital 980 and Table 58 of that decision, hardly support the claim that R&D associated with that chipset had to be regarded as specific to those two customers. Fourth, the documents referred to by the Commission in footnote 1277 to that decision provide no support whatsoever for the Commission’s contention. Fifth, Qualcomm [confidential].

465 In the seventh place, the applicant criticises the post hoc manipulation carried out by the Commission by reallocating the costs from the [confidential] between different chipsets, in particular by reallocating costs from the MDM8200 chipset to the MDM8200A chipset, thereby increasing LRAIC for the MDM8200A chipset from USD [confidential] to USD [confidential]; in the absence of such manipulation, prices would not be below costs for the latter chipset. In essence, first, the applicant submits that it is common ground that Qualcomm’s decision to develop the MDM8200A chipset post-dated, and was independent of, Qualcomm’s decision to develop the MDM8200 chipset and that those two chips were not a joint project, as is apparent from the table set out in recital 880(a) of the contested decision.

466 Second, the applicant states that it developed numerous chipsets after the MDM8200 chipset and the MDM8200A chipset, which also supported the HSPA+ standard and more advanced standards. Moreover, there is nothing ‘unique’ in the relationship between the MDM8200 chipset and the MDM8200A chipset: the existence of a close technical relationship between two chipsets is nothing unique or even unusual.

467 Third, while it is true that until some point in time between 2013 and 2015, [confidential], that was the result purely of a simple clerical error which was not, at the time, especially problematic. The contested decision is, the applicant argues, also inconsistent in its treatment of, on the one hand, the MDM8200 and MDM8200A chipsets and, on the other hand, the MDM6200 chipset, which is also alleged to be a ‘leading edge’ chipset closely related to other Qualcomm chipsets, in particular the QSC6295 chipset, which was developed before it.

468 Fourth, the Commission made no effort to investigate the technology and associated R&D effort in question or to tie back its manipulations to any specific entries or cost categories in the [confidential].

469 In the eighth place, the applicant complains that the Commission incorrectly allocated ‘incremental’ costs on the basis of revenue, rather than volume.

470 The methodology in question is, first of all, devoid of support from legal or economic literature. It also allocates additional costs to sales simply because prices in a given period are relatively high, which creates a strong bias towards finding predatory pricing where there is none. Finally, a genuine test of predation would involve ascertaining that the undertaking subsequently increased its prices (which the applicant did not do), which, according to that methodology, would mean that less R&D cost would be allocated to the allegedly predatory sales, due to the fact that they would instead be transferred over the recoupment period. That makes no sense as a matter of economics and is inconsistent with the concept of ‘recoupment’.

471 In the view of the applicant, the Commission also errs in defining the methodology in question as a ‘proxy for the price evolution of a chipset’, thereby failing to take into account that LRAIC is a cost measure and not a price proxy.

472 In addition, the applicant submits that the methodology at issue also results in a ‘moving target’, by virtue of which, if it had increased its prices in order to avoid alleged predation, that would at the same time have resulted in its revenues increasing, which would have led to higher R&D costs being allocated for each quarter, with the result that it would always have been guilty of predation. The applicant also criticises the ‘unpredictable and counter-intuitive results’ to which the methodology, as adapted by the Commission, gives rise. The applicant adds that adopting a volume-based allocation methodology results in almost no below-cost pricing.

473 The Commission and the intervener dispute the applicant’s arguments.

474 As regards the first complaint, alleging an error in the definition of LRAIC, it must be observed that, in the contested decision, the Commission acknowledged that spillovers were likely to occur in the chipset industry as the products improved. However, that does not prevent the applicant from bearing costs specifically for the development of a particular chipset or from being able to launch that chip on the market without bearing those costs. It follows that, even if there are spillovers, the Commission did not err in choosing to take account of the incremental development costs specifically linked to the product in question.

475 Moreover, by basing the measure of costs on the [confidential], the Commission, as is apparent from recital 835 of the contested decision, took into account only the costs which the applicant had directly or indirectly recorded against the development of the chipsets in question. Where there was specific evidence of spillovers, the Commission took that evidence into account, as is apparent, in relation to R&D cost spillovers between the MDM8200 chipset and the MDM8200A chipset, from Section 12.6.3.2 of that decision.

476 As regards possible spillovers more generally, in the absence of information which would have enabled the Commission to quantify those spillovers, as the Commission found in recital 783 of the contested decision, the most reasonable approach was to assume that the spillovers from which a particular chip benefited were roughly balanced by the spillovers generated by that particular chip for other chips. Accordingly, the development costs borne for that chip were not offset by the spillovers which it probably generated.

477 Consequently, the present complaint must be dismissed.

478 As regards the second complaint, which alleges an inconsistency between the justifications given by the Commission for its calculations in the SSO and those given for its calculations in the contested decision, it is sufficient to state that that complaint, even if it were well founded, could not in any way result in the methodology at issue being unlawful. Consequently, it must be rejected as being ineffective.

479 As regards the third complaint, alleging that the [confidential] is unsuitable for identifying its incremental R&D costs, it should be recalled that in that database the applicant records the costs, chipset by chipset, for internal management and accounting purposes. Given that cost allocation by chipset and the fact that that database is used internally by Qualcomm, the applicant fails to explain why it ought to be regarded as inappropriate for the Commission’s cost reconstruction. Moreover, as stated in recital 845 of the contested decision, the applicant itself, in response to information requests relating to the quantification of costs, extracted information from that database.

480 In addition, as regards Qualcomm’s assertion that none of the data in the [confidential] are designed to calculate LRAIC for the purposes of competition law, it is sufficient to note that that fact does not preclude the Commission from relying on such a database in so far as it contains information relevant to its investigation.

481 Lastly, even if the Commission had initially decided not to use the [confidential] in the SO and did not provide any explanation as to its change of opinion in that regard in the SSO, it is sufficient to state that, in the light of the case-law cited in paragraph 367 above, the applicant cannot properly rely on the fact that the Commission did not respond specifically in the contested decision to all the observations of fact and law which it raised during the administrative procedure. It follows that the present complaint must be rejected.

482 As regards the fourth complaint, alleging an error in treating certain costs recorded in the [confidential] as ‘incremental’ costs, it should be recalled, first of all, that Section 12.6.3.1 of the contested decision provides an explanation of the measure of the fixed part of the R&D costs incurred by Qualcomm for the production of each chipset which was taken into consideration for the calculation of LRAIC. Recital 835 of that decision lists the various R&D cost categories which, in that database, are allocated to each chip, or recorded directly against the development of a specific product, or are allocated ‘ex post’.

483 Furthermore, as set out in paragraph 411 above, recitals 836 and 837 of the contested decision state that in its choice of the elements to be included as incremental costs in the cost analysis, the Commission was guided by the use of the term ‘incremental’ in the [confidential], in relation to which it requested Qualcomm, in question 8.3 of the request for information made by decision dated 30 January 2017, to provide the same information on incremental R&D for the chipsets at issue as provided for the [confidential]. The information provided in response is set out in Annex A.15.1 to the application.

484 Lastly, the Commission reconstructed the incremental costs by adding together the [confidential] costs and [confidential] costs, that is to say, all the cost components in the database set out in Annex A.15.1 to the application, excluding those which were recorded as [confidential] costs, since the latter costs, although included in that database, had been subtracted by Qualcomm from the total ‘Incremental (i.e. following the same [methodology] of the [confidential])’ costs in that database.

485 It must be held, in that regard, that the elements which are included in the database extract criticised by the applicant are summarised in recital 835 of the contested decision and were identified by the Commission on the basis of information provided by Qualcomm.

486 Irrespective of whether that information ought to be classified in accordance with [confidential], the applicant always provided the same information in: (i) its response to the request for information of 10 July 2013, dating from December 2013, and, in particular, in its reply to question 19 concerning the main development stages of the first MDM chipsets and the associated costs; (ii) its response to questions 8 and 9 of the request for information of 13 October 2014 (Annex A.4.8.b to the application), to paragraph 38 et seq.; and (iii) its reply to question 8.3 of the request for information made by decision of 30 January 2017 (Annex A.4.10.b to the application).

487 In other words, in the applicant’s responses, the various cost categories are identified and allocated to a particular chipset in a regular manner and, in the absence of inconsistencies between the tables set out in the various responses (except for [confidential] costs), those categories can therefore be regarded as costs ‘incremental’ to such a chipset. The only difference between taking into account the methodology and not taking it into account is the inclusion or not of Hardware Tech Allocations, which, in any event, the Commission excluded from its calculations.

488 In those circumstances, it must be held that, in order for a certain cost to be included in the calculation of LRAIC, the simple use, or failure to use, the term ‘incremental’ in respect of a cost in the [confidential] cannot constitute the decisive criterion. Rather, it is the intrinsic nature of the cost itself (that is to say, whether it is apparent from one of the cost categories generally regarded as incremental costs), and the classification methodology used by the applicant itself in its responses, which must be taken into consideration.

489 Therefore, it must be held, on the basis of the responses and explanations provided by the applicant during the administrative procedure, that the contested decision correctly reconstructed the elements to be taken into account in calculating LRAIC. Furthermore, it should be noted that, in its response to question 19 of the request for information of 10 July 2013 (the first response which is relevant in that regard, dating from December 2013) (Annex A.4.3 to the application), the applicant made no reference to cost allocation or to the fact that, in practice, some of those costs were not attributable to specific chipsets.

490 Moreover, as regards software development costs, the applicant stated, in essence, in paragraphs 90 and 92 of the response to question 19 of the request for information of 10 July 2013 (Annex A.4.3 to the application), that the life of a piece of software was closely linked to that of a particular chipset and that the related costs follow the commercial life of that chip, thereby supporting the conclusion in recital 860 of the contested decision that that part of the [confidential] is a solid basis for determining the R&D costs related to software development to be attributed to a specific chip. However, that does not preclude the possibility that part of a piece of software may then be used for subsequent chipsets.

491 In addition, it should be noted that the applicant’s reservation concerning cost allocation was formulated relatively imprecisely during the administrative procedure and in the application.

492 That lack of precision in identifying non-incremental cost categories undermines the credibility of the applicant’s arguments, especially since, if its reasoning were to be followed, almost no R&D costs could be attributed to a specific chipset.

493 As regards the probative value of the evidence used by the Commission, it must be recalled that the reliability and, thus, the probative value of a document depends on its origin, the circumstances in which it was drawn up, the person to whom it is addressed and its content (judgments of 15 March 2000, Cimenteries CBR and Others v Commission, T‑25/95, T‑26/95, T‑30/95 to T‑32/95, T‑34/95 to T‑39/95, T‑42/95 to T‑46/95, T‑48/95, T‑50/95 to T‑65/95, T‑68/95 to T‑71/95, T‑87/95, T‑88/95, T‑103/95 and T‑104/95, EU:T:2000:77, paragraphs 1053 and 1838, and of 12 July 2011, Hitachi and Others v Commission, T‑112/07, EU:T:2011:342, paragraph 70).

494 In the present case, the Commission relied, in its price reconstruction, on the tables and responses provided by the applicant itself, including figures taken from the [confidential] and, in addition, it simply applied classification criteria already used by the applicant. Given that those data come from the applicant itself, they must be regarded as sufficiently credible to support the Commission’s price reconstruction.

495 In that context, it is not sufficient for the undertaking concerned to refer to a circumstance which might affect the probative value of that evidence. On the contrary, it is for the undertaking concerned to prove to the requisite legal standard, first, the existence of the circumstance relied on by it and, second, that that circumstance calls into question the probative value of the evidence relied on by the Commission (see, to that effect and by analogy, judgment of 26 September 2018, Infineon Technologies v Commission, C‑99/17 P, EU:C:2018:773, paragraphs 65 to 67).

496 In the present case, the applicant simply reiterates the arguments submitted during the administrative procedure in its response to the request for information of 13 October 2014, insisting that the data provided cannot be used for the purposes of an antitrust analysis, without adding in support of its argument any document demonstrating how another classification methodology would have been more appropriate than that used by the Commission. In the absence of such evidence, it must be concluded that, for both direct and indirect recording, the costs which the applicant itself identified in response to the Commission’s question are in principle to be regarded as incremental as regards each of the chipsets in question.

497 Furthermore, it must be pointed out, as the Commission did, that the choice of a specific recording method (direct or indirect, that is to say, by allocation) on the part of the dominant undertaking could affect the analysis of its costs. Even if a certain cost element followed in fact from an allocation, that could not, in itself, prevent it from being included in the calculation of LRAIC. If that were the case, a dominant undertaking could always frustrate any price-cost test by claiming that it was impossible to make allocations to specific costs. In addition, in the present case, the allocations reflect the assessments made by the applicant itself of the costs to be linked to a specific chipset.

498 As regards the USD [confidential] of R&D costs, recorded against the MDM8200 chipset, but which actually relate to the [confidential], it should be noted that the applicant raised that aspect for the first time in its SSO response. Previously, the Commission had no reason to question the accuracy of the data received, especially since Article 18(4) of Regulation No 1/2003 imposes on the undertakings concerned an obligation to provide complete, correct and non-misleading information.

499 The fact that those costs had been recorded against the MDM8200 chipset is clear from the [confidential] and from the applicant’s responses, which must be regarded as particularly reliable evidence in that they run counter to the interests of the declarant (see, to that effect, judgment of 3 March 2011, Siemens v Commission, T‑110/07, EU:T:2011:68, paragraphs 54 and 75). In response to those elements, the only evidence adduced by the applicant is a document dated 9 July 2009 entitled [confidential] (Annex A.18.1 to the application), which contains footnote 20 concerning USD [confidential] having been excluded since it related to [confidential]. It should be noted, in that regard, that that document, as noted by the Commission in recital 851 of the contested decision, provides a table comparing the R&D costs for the MDM8200 chipset with those relating to the MDM8200A chipset, which was regarded as ‘not cost competitive’. That table appears to demonstrate only that, for the purposes of comparing the costs of the two chipsets in the presentation, that [confidential] amount was not taken into consideration, and it does not demonstrate that, in calculating the costs relating to the MDM8200 chipset, those specific costs were not, or ought not to have been, taken into account. As to the remainder, the applicant’s arguments are not supported by any documentation and, in themselves, cannot call into question the evidence previously provided by the applicant itself. In those circumstances, they must be rejected. It follows that the present complaint must be rejected.

500 As regards the fifth complaint, alleging an error in that the revenues obtained by Qualcomm through its patent licensing business were disregarded, it should be noted that, even if the applicant had changed its licensing policy, that does not mean that it did not grant licences relating to its entire patent portfolio during the relevant period. In any event, as is apparent from footnote 1238 to the contested decision, the applicant does not challenge the fact that [confidential].

501 Moreover, it should be noted that, first, the applicant maintains that it identified approximately [confidential] patents and patent applications of which ‘many’ are ‘likely’ to have been generated ‘in whole or in part’ by work which the Commission regards as specific to the chips at issue. Second, the applicant does not identify those patents or patent applications, nor does it quantify the revenue generated by them. Rather, it simply argues that, particularly as regards software-related patents, it is impossible to identify those linked to the chipsets in question. In those circumstances, the applicant’s arguments are not sufficient to call into question the Commission’s findings in that regard. It follows that the present complaint must be rejected.

502 As regards the sixth complaint, alleging, in essence, errors concerning the taking into account of R&D costs related to the development of Qualcomm’s chipsets, it must be held, as is clear from the contested decision, in particular from recital 862 thereof, and more generally from Section 12.6.3 of the contested decision, that sales to Huawei and ZTE were fundamental for the recovery of R&D costs.

503 Moreover, as is indicated in recital 888 of the contested decision, it was appropriate to include the incremental costs incurred in R&D in the measure of costs in order to assess whether the sales to Huawei and ZTE had been made at prices below cost, as the forecasted sales to Huawei and ZTE accounted for a substantial part of the expected demand which justified developing those chips in the first place. Furthermore, contrary to the applicant’s assertions, while the forecasted sales of units of the MDM8200, MDM6200 and MDM8200A chipsets to other customers also contributed to the recovery of the incremental development costs, those sales were of an order of magnitude which would not have justified developing those products. That conclusion is also borne out by the documents referred to in footnote 1277 to that decision.

504 Consequently, the present complaint must be dismissed.

505 As regards the seventh complaint, alleging post hoc manipulation carried out by the Commission in reallocating the costs recorded from the [confidential] between different chipsets, it must be observed that the applicant itself acknowledged that, until December 2013, the costs related to the software used for the MDM8200 and MDM8200A chipsets, as recorded in that [confidential], were not distinguished according to whether they related to one or other of those two chipsets (footnote 78 to the reply to the request for information of 13 October 2014).

506 It should also be noted that, in the table taken from the [confidential], contained in Annex A.15.1 to the application, although the development of the MDM8200A chipset commenced in April 2009, no costs are associated with it before the last quarter of 2009 and the costs associated with that chipset, even from the last quarter of 2009 onwards, remain low compared with those of the MDM8200 chipset and the MDM6200 chipset. That fact, in view of the technical similarity of the two chipsets and their joint commercialisation, supports the Commission’s arguments in recital 880 of the contested decision, demonstrating that there was a unique relationship between the MDM8200 chipset and the MDM8200A chipset, which is not comparable to any relationship which might have existed between the MDM8200 chipset and any other chipset supporting HSPA+ technology.

507 The intervener also observes, correctly, that the contested decision contains a ‘lifetime profitability analysis’ which involved examining the profitability of all sales of the MDM8200 chipset and the MDM8200A chipset throughout their lifetime and in relation to all customers, and that it is apparent from that analysis that the lifetime revenues of those two chipsets were not sufficient to cover their R&D manufacturing and development costs. Therefore, Qualcomm’s pricing for those two chipsets would never have allowed an equally efficient supplier to cover its costs for a competing chipset.

508 In those circumstances, the present complaint must be rejected.

509 As regards the eighth complaint, alleging that ‘incremental’ costs were incorrectly allocated on the basis of revenue, rather than volume of sales, first, it should be noted that, although the Commission cites only one source in favour of the methodology adopted, namely the Organisation for Economic Co-Operation and Development (OECD) Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2017 in footnote 1293 to the contested decision, the applicant does not refer to any study or principle supporting the substitute methodology it proposes. Furthermore, in so far as, by its argument that the methodology adopted by the Commission creates a strong bias towards finding predatory pricing where there is none, the applicant alleges that that methodology leads to predatory prices being identified artificially, it adduces no evidence to support its allegation.

510 Second, it is sufficient to note, as the Commission did, that the applicant’s argument that the methodology adopted by the Commission resulted in lower R&D costs being attributed to the allegedly predatory sales due to the fact that they would instead be transferred over the recoupment period does not, however, mean that that methodology is unsuitable. It should be noted that the methodology adopted by the Commission results in those costs being allocated according to the fact that prices over a given period, in particular the first period of the commercialisation of chipsets, are relatively high, and therefore revenues are also high. The methodology proposed by the applicant, on the other hand, results in a uniform, pro rata allocation of costs so that each unit (regardless of when it is sold during the chipset’s life cycle) has to recover the same amount of costs. As the Commission correctly states in recital 917 of the contested decision, given the evolution of chipset prices, such an approach was not suitable in the present case because it would have led to a large part of the development costs being shifted towards the end of the product’s life cycle, where prices and margins tend to be low, thus generating a large number of false positives in the price-cost test.

511 Third, as regards the Commission’s use of the expression ‘proxy for the price evolution of a chipset’ to define the methodology adopted, the applicant’s argument that the Commission thereby failed to take into account that LRAIC was a measure of costs and not a price proxy cannot succeed, since that expression simply refers to the fact that that methodology seeks to take account of the price profile over time, with the result that periods during which chipsets are assumed to generate higher revenues also bear a greater share of the costs.

512 Fourth, as regards the applicant’s argument that the methodology adopted by the Commission is a ‘moving target’, it must be stated that, as indicated in recital 935 of the contested decision, the methodology proposed by the applicant suffers from the same problem in that whenever prices increase in some or all quarters of the product’s life cycle, the volumes sold of that product would drop, so that each unit sold would have to bear a larger share of the product’s fixed development cost. Accordingly, higher prices might still not be sufficiently high to exceed the cost benchmark obtained through a volume-based allocation of development costs.

513 Fifth, the applicant’s argument that adopting a volume-based allocation results in almost no below-cost pricing is unfounded. As the Commission points out, Table 13 in Annex A.2.4.5 to the application demonstrates that that is the case for the MDM8200A chipsets, but not for the MDM8200 or MDM6200 chipsets. Indeed, Qualcomm’s predatory sales of MDM6200 chipsets to ZTE were priced so low, below AVC, that a simple allocation of costs based on volume rather than revenue would not change the fact that Qualcomm was still pricing below cost (recital 891 of the contested decision). The same applies to MDM8200 chipsets. The allocation of costs by volume rather than revenue does not produce significantly different results for Qualcomm. The present complaint must therefore be rejected.

514 In the light of all of the foregoing, the third part of the seventh plea must be rejected. Therefore, since the other parts of that plea have also been rejected (see paragraphs 424 and 444 above), that plea must be rejected in its entirety.

 The ninth plea: ‘manifest errors of law and of appraisal’ as regards the finding that the prices charged by the applicant foreclosed Icera and caused consumer harm

515 The present plea comprises four parts. The first alleges that the Commission failed to conduct a so-called ‘as-efficient’ competitor analysis. The second part alleges a contradiction between the finding of foreclosure set out in the contested decision and Icera’s actual performance during the relevant period. The third alleges that there is no evidence of alleged financial predation, having regard to the fact that, during that period, Icera raised additional funds, was acquired by Nvidia and remained active on the market. The fourth alleges that the applicant could not have eliminated competition from Icera’s technology.

516 Since the second, third and fourth parts all concern, in essence, the alleged lack of effects of the conduct at issue, they will be examined together.

 The first part: the Commission failed to conduct a so-called ‘as-efficient’ competitor analysis

517 According to the applicant, the Commission failed to carry out a so-called ‘as-efficient’ competitor analysis both on the relevant market and in the leading edge segment of that market and, as a result, did not comply with the applicable legal standard, as is apparent from the arguments put forward in support of the third plea. In so doing, the Commission also failed to examine whether the share of the market covered by the conduct at issue was of sufficient magnitude to have anticompetitive effects, as the Court of Justice held in the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632).

518 The applicant also believes that Icera was not in any event an ‘as-efficient’ competitor, since, being uncompetitive outside the leading edge segment of the UMTS chipsets market, Icera would not have been able to satisfy a significant share of demand on the relevant market. In its view, an ‘as-efficient’ competitor should, at the very least, be able to replicate the dominant undertaking’s entire offering.

519 The Commission and the intervener dispute the applicant’s arguments.

520 According to the case-law on predatory pricing, prices below AVC must be regarded as abusive, since a dominant undertaking has no interest in applying such prices except that of eliminating its competitors. Moreover, prices below ATC, but above AVC, must be regarded as abusive if they are determined as part of a plan for eliminating a competitor, given that such prices can drive from the market undertakings which are perhaps as efficient as the dominant undertaking but which, because of their smaller financial resources, are incapable of withstanding the competition waged against them (see, to that effect, judgment of 3 July 1991, AKZO v Commission (C‑62/86, EU:C:1991:286, paragraphs 70 to 72).

521 It follows, first, that there is a presumption of abuse where an undertaking in a dominant position prices below AVC, there being no need for the Commission to undertake any analysis other than such a comparison of the prices charged by the dominant undertaking and of some of its costs. Second, where an undertaking in a dominant position prices below ATC but above AVC, the Commission is required, for the purposes of demonstrating that there is abuse, to compare the prices charged by the dominant undertaking and some of its costs and to prove that there is a plan to eliminate a competitor, given that such prices are, in essence, capable, as such, of driving equally efficient competitors from the market. The Commission is therefore under no obligation to carry out analyses other than such a demonstration, in order, in particular to adduce evidence of the anticompetitive effects of the contested practice.

522 Contrary to the applicant’s assertions, the Commission is, consequently, not required, when examining whether an undertaking in a dominant position charged predatory prices, to examine whether the share of the market covered by the contested practice is of sufficient magnitude for that practice to have anticompetitive effects. In the present case, the Commission cannot therefore reasonably be criticised for having failed to undertake such an examination.

523 In that regard, it is also apparent from paragraphs 72 and 73 of the Communication regarding Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings (OJ 2009 C 45, p. 7; ‘the Enforcement Priorities Communication’) that it may be easier for the dominant undertaking to engage in predatory conduct if it selectively targets specific customers with low prices, as that will limit the losses incurred by the dominant undertaking, and that it is less likely that it engages in predatory conduct if the conduct concerns a low price applied generally for a long period of time. It follows that, as regards predatory pricing, the share of the market covered by the conduct at issue is generally low and if penalties could be imposed in respect only of predatory conduct relating to a sufficiently large market share, any selective predatory practice could escape the prohibition laid down in Article 102 TFEU, even though it might lead to an as-efficient competitor being eliminated. It is clear from the case-law that it must be possible to impose penalties in respect of predatory pricing whenever there is a risk that competitors will be eliminated (see, to that effect, judgment of 14 November 1996, Tetra Pak v Commission, C‑333/94 P, EU:C:1996:436, paragraph 44).

524 Furthermore, it is also apparent from the case-law that the application of the ‘as-efficient’ competitor test consists in examining whether the pricing practices of a dominant undertaking could drive an equally efficient competitor from the market and that that test is based on a comparison of the prices charged by a dominant undertaking and certain costs incurred by that undertaking as well as its strategy (judgments of 27 March 2012, Post Danmark, C‑209/10, EU:C:2012:172, paragraph 28, and of 6 October 2015, Post Danmark, C‑23/14, EU:C:2015:651, paragraphs 53 and 54).

525 It follows that, in the context of an investigation into potential predatory prices, the analysis by which the Commission compares the prices charged by an undertaking in a dominant position with some of its costs for the purposes of assessing whether that undertaking priced below ATC but above AVC already includes an ‘as-efficient’ competitor analysis.

526 In so far as an undertaking in a dominant position sets its prices below ATC but above AVC, a competitor as efficient as that undertaking will not, in principle, be able, because of its lesser financial resources, to compete with those prices without incurring losses which are unsustainable in the long term. Such prices are therefore capable of excluding an ‘as-efficient’ competitor, which corresponds to what the Commission must demonstrate when applying the ‘as-efficient’ competitor test in order to prove that an anticompetitive practice has foreclosure potential.

527 Therefore, where the Commission has validly proven, as in the present case, that the undertaking in a dominant position priced below ATC, but above AVC, accordingly, it has implicitly applied the ‘as-efficient’ competitor test, which is sufficient to reject the first part of the ninth plea.

 The other parts: in essence, the conduct at issue had no effects

528 In support of the second part of the plea, the applicant argues that Icera actually prospered during the relevant period, which undermines the Commission’s assertion that the conduct at issue resulted in Icera being prevented, at a crucial stage of its development, from gaining access to Huawei or ZTE, on which its prospects for growth in the leading edge segment of the UMTS chipsets market depended.

529 In support of the third part of the plea, the applicant submits that, during the relevant period, Icera raised additional funds, in particular as a result of its acquisition by Nvidia, which undermines the Commission’s allegation that the conduct at issue resulted in Icera being required to make sales at a loss and to exhaust its venture capital funding.

530 In support of the fourth part of the plea, the applicant claims that the conduct at issue could not have resulted in competition from Icera’s technology being eliminated. In the applicant’s view, Icera’s withdrawal from the market had a limited, or even no, impact on the prices which it charged for its ‘leading edge’ UMTS chipsets or for any other UMTS chipsets, in particular because Icera’s technology and know-how were transferable, with the result that competition would not have been affected by Icera disappearing from the market.

531 The Commission and the intervener dispute the applicant’s arguments.

532 It must be observed that, as the Commission states in its written pleadings, it was only for the sake of completeness that it, in Section 12.7.4 of the contested decision, demonstrated that the conduct at issue had had anticompetitive effects, which the applicant disputes in the context of those parts. It is settled case-law that complaints directed against grounds of a Commission decision included purely for the sake of completeness will be rejected outright since they cannot lead to that decision being set aside (judgments of 8 May 2003, T. Port v Commission, C‑122/01 P, EU:C:2003:259, paragraph 17, and of 8 July 2004, Dalmine v Commission, T‑50/00, EU:T:2004:220, paragraph 146), which, in any event, renders those parts ineffective (see, to that effect, judgment of 29 March 2012, Telefónica and Telefónica de España v Commission, T‑336/07, EU:T:2012:172, paragraph 283).

533 It follows that the second to fourth parts of the applicant’s ninth plea must be rejected. Therefore, since the first part of that plea has also been rejected (see paragraph 527 above), that plea must be rejected in its entirety.

 The tenth plea: ‘manifest errors of assessment’, of fact and of law, failure to state reasons and infringement of the right to be heard and of the principle of sound administration, in that the Commission concluded that the applicant’s pricing practices were the implementation of a plan to exclude Icera

534 The tenth plea comprises three parts. The first part alleges that the documents cited in the contested decision do not establish a coherent ‘theory of harm’ developed by the Commission. The second alleges misinterpretation and misrepresentation of certain internal documents on which the Commission relies in that decision for the purposes of demonstrating an intention to eliminate a competitor. The third part alleges that the Qualcomm employee who authored a number of internal documents quoted by the Commission for the purposes of demonstrating that there was a predatory strategy was a relatively low-level employee with no power to make pricing decisions.

535 The Commission and the intervener dispute the applicant’s arguments.

536 In recital 1118 of the contested decision, the Commission stated that, in order to establish the applicant’s intention to eliminate a competitor, it had relied not only on the documents expressly cited in the context of Section 12.8 of that decision, specifically relating to such an intention, but also on the other evidence set out in Section 12.4.2 of that decision and on the links it had been able to establish between that evidence and its price-cost test in the decision as regards the MDM8200 chipset (recitals 954 to 956 of the contested decision), as regards the MDM6200 chipset (recitals 968 to 971 of the contested decision) and as regards the MDM8200A chipset (recitals 977 to 978 of the contested decision). In that regard, it indicated, in recital 1118 of the contested decision, without being contradicted on that point by the applicant, that all of that evidence and those recitals formed an integral part of its analysis aimed at demonstrating the applicant’s eliminatory intention.

537 The Commission then demonstrated there was intention on the part of the applicant to eliminate a competitor, which underpinned its predatory pricing towards Huawei and ZTE for the MDM8200, MDM6200 and MDM8200A chipsets, on the basis of direct evidence, namely internal Qualcomm documents (Section 12.8.1 of the contested decision) and indirect evidence, namely contextual elements (Section 12.8.2 of the contested decision).

538 As regards, in particular, the direct evidence, it is apparent from recital 1120 of the contested decision that the Commission selected, from the evidence summarised in Section 12.4.2 of that decision, nine exchanges of internal emails which, in its view, were particularly illustrative of there being a plan stemming from Qualcomm’s senior management to eliminate Icera, which it set out in recitals 1121 to 1137 of that decision.

539 In particular, the following exchanges of emails are cited:

– an exchange of emails from December 2008, containing inter alia, the statements ‘we should not give Icera any opportunity in Huawei strategically’, ‘in the case Icera gets ZTE, we can push ZTE back by working with Huawei in the market place’ and ‘please consider to give another 2-3% [price reduction] to make sure we have a 100% share in Huawei’ (recital 1121 of the contested decision);

– an exchange of emails from February 2009 referring to preventive action in order to prevent Icera from being able to achieve sufficient volumes, a price war, the threat posed by Icera and the fact that Icera ought not to make a breakthrough with any of the major original equipment manufacturers, and to ‘crush Icera at ZTE’ (recital 1122 of that decision);

– an exchange of emails from December 2009 concerning a reduction in the price of the MDM6200 chipset for ZTE and the fact that the applicant could not allow Icera to gain a large amount of business, which would have helped it become stronger (recital 1124 of that decision);

– an exchange of emails from December 2009 dealing with making price a non-issue to let the MDM8200 chipset ‘fill the void’ (recital 1125 of the contested decision);

– an exchange of emails from January 2010 in which Icera’s 8042 chipset is presented as the main threat to the applicant and proposing, inter alia, to work with Huawei to beat Icera with the MDM6200 chipset, to exhaust Icera’s limited funds from venture capital and to squeeze ICE8042/ZTE market space (recital 1127 of the contested decision);

– an exchange of emails from January 2010, containing the statements ‘we have to protect our sockets at major accounts’, ‘ZTE and Huawei are the top priorities and the China team is working on issues at these accounts’ and, ‘for the short term, adjusting MDM6200 and MDM8200a pricing has to be done to contain volume growth at Icera in 2010’ as well as an attached presentation in which it is suggested that Icera should be squeezed for approximately six months to burn out its very limited venture capital funds (recital 1128 of the contested decision);

– an exchange of emails from June 2010, containing the statement ‘the strategy of squeezing [the] ICE8042 [chipset with] both MDM6200 chips and MDM8200A made effect’ (see recital 1132 of the contested decision);

– an exchange of emails from December 2010, relating to the MDM8200A chipset strategy with Huawei and ZTE, which includes Icera’s projected sales volumes to ZTE in the scenario where the applicant did nothing and did not grant ZTE special support other than normal pricing adjustments (recital 1134 of the contested decision);

– an exchange of emails from May 2011 in which a senior Qualcomm employee expresses his hesitation, in view of Icera’s difficult financial situation, to grant further price reductions for the MDM8200A chipset (recital 1136 of the contested decision).

540 As regards the indirect evidence, the Commission relied on the following five contextual elements, namely the highly targeted nature of the conduct at issue, the significant extent of the applicant’s sales at a loss in the strategically important leading edge segment of the UMTS chipset market, the fact that the conduct at issue continued without interruption during the relevant period, the two NRE payments intended to provide a disguised price reduction for the MDM8200 and MDM6200 chipsets, and the applicant’s financial sacrifices in the area of prices and supply terms. In that regard, the Commission stated, in recital 1138 of the contested decision, that that indirect evidence was, in itself, sufficient to demonstrate the existence of an intention on the part of the applicant to eliminate Icera from the relevant market.

541 According to settled case-law, in so far as certain grounds of a decision in themselves provide a sufficient legal basis for that decision, any errors in other grounds of the decision have no effect in any event on its operative part (judgments of 21 September 2005, EDP v Commission, T‑87/05, EU:T:2005:333, paragraph 144, and of 29 March 2012, Telefónica and Telefónica de España v Commission, T‑336/07, EU:T:2012:172, paragraph 283).

542 In the present case, it is therefore necessary to examine whether the present plea is effective in so far as the applicant argues that the Commission made errors when it concluded that the pricing practices were the implementation of a plan to eliminate Icera.

543 At the outset, it should be noted that, as regards direct evidence, the applicant expressly disputes, in the context of the second part of the present plea, only the Commission’s interpretation of four of the nine email exchanges considered in the contested decision to be particularly illustrative of there being an intention to eliminate Icera, namely the February 2009 exchange (the second item of direct evidence), the January 2010 exchanges (the fifth and sixth items of direct evidence) and the June 2010 exchange (the seventh item of direct evidence).

544 As regards the allegedly low-level employee referred to in the third part of the present plea, he was involved only in the February 2009 email exchanges (the second item of direct evidence), the second email exchange of December 2009 (the fourth item of direct evidence) and the two email exchanges of January 2010 (the fifth and sixth items of direct evidence), but was not involved in the email exchanges of December 2008 (the first item of direct evidence), the first email exchange of December 2009 (the third item of direct evidence), the email exchange of June 2010 (the seventh item of direct evidence), the email exchange of December 2010 (the eighth item of direct evidence) and the email exchange of May 2011 (the ninth item of direct evidence).

545 From that, it is apparent that, in the application, in support of the present plea, the applicant did not criticise the Commission’s interpretation in the contested decision of the emails exchanged in December 2008 (the first item of direct evidence), the first email exchange of December 2009 (the third item of direct evidence), the email exchange of December 2010 (the eighth item of direct evidence) or the email exchange of May 2011 (the ninth item of direct evidence).

546 The email exchanges referred to in paragraph 548 above attest, in themselves, to there being a strategy on the part of the applicant seeking to eliminate Icera.

547 As is apparent from paragraph 539 above, the exchange of emails of December 2008 (the first item of direct evidence) refers to the fact of not allowing Icera any ‘opportunity’ with Huawei, of working with Huawei to push ZTE back if Icera were to win ZTE, and of granting Huawei an additional 2 or 3% price reduction to ensure that it would have a 100% share with that customer. It follows that Icera is clearly identified in that exchange, as is the need for the applicant to take action in respect of it, by collaborating with Huawei and granting Huawei an additional price reduction in order to allow Icera no ‘opportunity’ with that customer.

548 It should also be observed that the Commission was entitled to rely on the December 2008 email exchange even though it occurred a few months before the beginning of the relevant period. It is apparent from the case-law that the Commission may take account of evidence from outside the infringement period if that evidence forms part of the body of evidence relied on by the Commission in order to prove that infringement (see, to that effect, judgments of 2 February 2012, Denki Kagaku Kogyo and Denka Chemicals v Commission, T‑83/08, not published, EU:T:2012:48, paragraph 193, and of 16 June 2015, FSL and Others v Commission, T‑655/11, EU:T:2015:383, paragraph 178). Moreover, it seems entirely conceivable that the actual implementation of a strategy to eliminate a competitor may require some time and will, consequently, occur a few months after the strategy was developed.

549 The first email exchange of December 2009 (the third item of direct evidence) refers to a reduction in the price of the MDM6200 chipset for ZTE and to the fact that the applicant could not allow Icera to gain a large amount of business, which would help it become stronger. It follows that Icera is clearly identified in that exchange, as is the need for the applicant to take action in respect of Icera, by granting ZTE a price reduction to prevent Icera from becoming stronger by winning contracts from that customer in particular.

550 The email exchange of December 2010 (the eighth item of direct evidence) refers to special support to ZTE other than normal price adjustments. In the same vein, it follows from that exchange that Icera is clearly identified, as are Icera’s projected shipments to ZTE if special support for ZTE other than normal price adjustments were not granted to ZTE.

551 The email exchange of May 2011 (the ninth item of direct evidence) refers to Icera’s difficult financial situations which caused a senior Qualcomm employee doubts regarding granting additional price reductions for the MDM8200A chipset, demonstrating that Icera’s financial situation had influenced the applicant’s pricing policy during the relevant period.

552 The fact that the applicant does not call into question the Commission’s interpretation in the contested decision of the email exchanges referred to in paragraph 545 above therefore renders the present plea ineffective in so far as the applicant relies on errors made by the Commission when it concluded that the pricing practices were the implementation of a plan to eliminate Icera. Even if the applicant’s criticism of the Commission’s interpretation of certain email exchanges on which the Commission relies as direct evidence were to be accepted, the fact remains that the evidence which it does not dispute constitutes, in itself, a demonstration of the applicant’s intention to eliminate Icera.

553 In any event, the Commission also relied, in the contested decision, on items of indirect evidence in order to demonstrate the applicant’s intention to eliminate a competitor.

554 As is apparent from recital 1138 of the contested decision, which is not disputed by the applicant, indirect evidence is sufficient to demonstrate the applicant’s intent to eliminate Icera.

555 In addition, the applicant puts forward three parts in support of the present plea, none of which criticises the body of indirect evidence referred to in Section 12.8.2 of the contested decision, since the first relates to the Commission’s ‘theory of harm’, the second calls into question the Commission’s interpretation and presentation of certain items of direct evidence, namely certain internal Qualcomm documents, and the third concerns the fact that an employee who authored certain documents referred to by the Commission as items of direct evidence was a low-level employee.

556 Therefore, since the body of indirect evidence on which the Commission relies in Section 12.8.2 of the contested decision is sufficient, in itself, to demonstrate the existence of an intention on the part of the applicant to eliminate Icera, the fact that the applicant does not call that evidence into question in the context of the present plea renders that plea ineffective in so far as the applicant relies on errors made by the Commission when it concluded that the applicant’s pricing practices were the implementation of a plan to eliminate Icera.

557 In any event, according to the case-law, where an undertaking in a dominant position prices below ATC, in order to demonstrate that there is abuse, the Commission must rely on a body of sound and consistent evidence making it possible to establish that undertaking’s intention to eliminate a competitor (see, to that effect, judgments of 6 October 1994, Tetra Pak v Commission, T‑83/91, EU:T:1994:246, paragraph 151, and of 30 January 2007, France Télécom v Commission, T‑340/03, EU:T:2007:22, paragraph 197). That demonstration must be aimed at establishing, on the basis of specific, tangible points of analysis and evidence, that that conduct, at the very least, is capable of producing exclusionary effects (see judgment of 21 December 2023, European Superleague Company, C‑333/21, EU:C:2023:1011, paragraph 130).

558 Faced with, prima facie, a sufficient body of evidence, it is for the applicant to put forward in the application another full and consistent explanation which might lend the various elements referred to in the contested decision a different significance (see, to that effect, judgment of 12 December 2012, Almamet v Commission, T‑410/09, not published, EU:T:2012:676, paragraph 145).

559 In the present case, the Commission found, in particular, that the applicant priced below LRAIC but above AVC. In order to establish an infringement in the present case, the Commission was therefore required to provide, inter alia, sound evidence of the applicant’s intention to eliminate Icera, which it did in Section 12.8 of the contested decision.

560 As is apparent from paragraph 538 above, in Section 12.8.1 of the contested decision, the Commission refers to a number of internal Qualcomm documents, namely nine email exchanges which, in its view, are particularly illustrative of the existence of a strategy to eliminate Icera from the relevant market.

561 As noted in paragraph 545 above, in support of the present plea, the applicant made no criticism at all of the Commission’s interpretation in the contested decision of the emails exchanged in December 2008 (the first item of direct evidence), the first email exchange of December 2009 (the third item of direct evidence), the email exchange of December 2010 (the eighth item of direct evidence) and the email exchange of May 2011 (the ninth item of direct evidence). As has been held in paragraph 546 above, those email exchanges attest, in themselves, to there being a strategy to eliminate Icera, particularly in the light of the documents referred to in Section 12.4.2 of that decision.

562 The five other email exchanges – the Commission’s interpretation of which is, on this occasion, expressly disputed in the application – constitute a body of sound and consistent evidence making it possible to establish the applicant’s intention to eliminate Icera.

563 At the outset, it is important in that regard to recall that what the Commission sought to demonstrate by relying on that direct evidence in Section 12.8.1 of the contested decision, was the applicant’s intention to eliminate Icera rather than the other constituent elements of the infringement, such as, in particular, whether the pricing of the three chipsets referred to was predatory, which are addressed in other parts of that decision. Therefore, what must be determined in the present case is whether the internal documents on which the Commission relies do constitute a body of sound and consistent evidence making it possible to establish, overall, the applicant’s intention to eliminate Icera, all other considerations being irrelevant for that purpose.

564 In that regard, the terms used in each of those email exchanges are particularly indicative of the applicant’s intention to eliminate Icera.

565 In the first place, the email exchange of February 2009 expressly describes Icera as a ‘threat’ and refers to preventive measures to be taken in order to prevent Icera from reaching sufficient volumes and from making a breakthrough with major original equipment manufacturers. Furthermore, it is suggested in that exchange of emails that Icera be ‘crush[ed]’ in relation to ZTE and that Icera should be allowed no ‘opportunity’ in relation to Huawei. It follows that Icera is clearly identified in that exchange, as is the need for the applicant to take measures to prevent it from attaining sufficient volumes and to eliminate Icera, in particular as regards two customers, Huawei and ZTE.

566 The applicant disputes the conclusions drawn by the Commission from the email exchange of February 2009, submitting that that exchange took place three months before the beginning of the infringement, that it contains no reference to the pricing of the three chips concerned, and that it undermines the ‘theory of indirect predation’.

567 In that regard, it should be recalled that, as is apparent from the case-law cited in paragraph 548 above, the Commission may take account of evidence from outside the infringement period if that evidence forms part of the body of evidence relied on by the Commission to prove that infringement. In addition, the intention to eliminate a competitor may well be expressed some months before the actual implementation of the strategy aimed at such elimination. Furthermore, the fact that the email exchange of February 2009 contains no reference to the pricing of the three chipsets concerned or that it allegedly contradicts the ‘theory of indirect predation’ is irrelevant, since those elements are not capable of calling into question the existence of an intention on the part of the applicant to eliminate Icera. Moreover, even though that exchange does not set out the details of the measures which were to be taken against Icera, the fact remains that that exchange expressly suggests adopting preventive measures against Icera.

568 In the second place, the first email exchange of January 2010 (the fifth item of direct evidence) again identifies Icera, and more specifically its ICE8042 chipset, as the main threat to the applicant and also refers to actions to be taken by the applicant in order to eliminate Icera, namely working together with Huawei and exhausting Icera’s limited funds. It follows that Icera is clearly identified in that exchange, as is the need for the applicant to take action in order to eliminate it, in particular as regards Huawei.

569 The second exchange of emails in January 2010 (the sixth item of direct evidence) refers to measures to be taken to contain Icera’s growth in volume terms and to protect the applicant’s position with its major customers, Huawei and ZTE being described in that regard as ‘priority’ customers, including an adjustment to the price of MDM6200 and MDM8200A chipsets. Reference is also made, in a presentation attached to that exchange, to ‘squeez[ing]’ Icera for six months to exhaust its limited funds. It follows that Icera is once more clearly identified in that exchange, as is the need for the applicant to take action for the purpose of eliminating it as regards Huawei and ZTE, in particular by means of a price adjustment for the MDM6200 and MDM8200A chipsets.

570 The applicant attempts to call into question the Commission’s conclusion regarding the email exchanges of January 2010, relying on a long series of different allegations, such as: (i) the Commission’s failure to take account of the competition provided by HiSilicon in the ‘theory of harm’ developed by it, whereas that is referenced in the email of 1 January 2010, (ii) the failure to mention the MDM8200 chipset at all and (iii) the fact that the main subject of those emails was competition from Vodafone. Such allegations must, however, be regarded as irrelevant since they are not capable of calling into question the existence of an intention on the part of the applicant to eliminate Icera.

571 In addition, the applicant does not dispute that the email exchanges of January 2010 refer to actions, in particular pricing actions, to be taken in relation to Huawei and ZTE in order to contain Icera’s growth and to oblige it to exhaust its limited funds. In that regard, although the applicant disputes the Commission’s interpretation of the word ‘squeeze’ which, in its view, does not mean ‘cut funding’, the fact remains that the sentence which includes that term, namely ‘squeeze Icera for approximately [six] months to burn out its very limited venture capital funds’ refers unequivocally to a strategy financially to exhaust Icera. It should also be noted that the applicant puts forward no other interpretation of that term ‘squeeze’ which would call into question the harmful nature of such a strategy in relation to Icera.

572 In the third place, the email exchange of June 2010 (the seventh item of direct evidence) expressly refers to the fact that the Icera ICE8042 chipset was losing momentum with Huawei and ZTE and to the fact that the ‘strategy of squeezing [the] ICE8042 [chipset] w[ith] both MDM6200 and MDM8200A [chipsets]’ had ‘made effect’. It follows that the applicant did indeed implement a strategy in relation to Huawei and ZTE, based on the MDM6200 and MDM8200A chipsets and targeting Icera.

573 In that regard, the applicant relies on a varied series of allegations which are irrelevant, since they do not call into question the existence, as such, of an intention on its part to eliminate Icera, such as the failure to mention the MDM8200 chipset at all, the agreement on non-recurring engineering expenses concluded with ZTE or the additional equity financing announced by Icera in May 2010. In those allegations, the applicant does not dispute that the email exchange of June 2010 refers to a strategy in relation to Huawei and ZTE against Icera, using the MDM6200 and MDM8200A chipsets, which was actually implemented, since it had effects.

574 In the fourth place, in support specifically of the third part of the present plea, the applicant claims that the Commission uses as its basis documents drafted by a low-level employee, who had no authority over or influence on prices and was therefore incapable of playing an essential role in implementing the strategy to eliminate Icera. In that regard, it is apparent from Section 12.8.1 of the contested decision that that employee was involved in the email exchanges of February 2009, December 2009 and January 2010.

575 As regards the email exchange of February 2009 (the second item of direct evidence), it is apparent from recital 1122 of the contested decision that the employee concerned sent his ‘analysis on Icera threat’ to a member of the applicant’s senior management, who agreed with that analysis and sent it to other members of senior management.

576 As regards the second email exchange of December 2009 (the fourth item of direct evidence), it is not disputed by the applicant that it was a member of its senior management who suggested making price a non-issue to leave the MDM8200 chipset to ‘fill the void’.

577 Lastly, as regards the January 2010 email exchanges (the fifth and sixth items of direct evidence), containing the suggestion by the employee concerned to ‘squeeze’ Icera in order to exhaust its limited funds, it is common ground that those emails were addressed to a number of members of the applicant’s senior management. In addition, that suggestion was also repeated in a presentation prepared by that employee attached to an ‘Icera analysis’ report, sent by a member of senior management to the applicant’s executive management.

578 It follows that members of the applicant’s senior management, with authority and influence over price, were indeed involved in those email exchanges from which an intention to eliminate Icera is clear.

579 It follows from the foregoing that the Commission established to the requisite legal standard, in Section 12.8.1 of the contested decision, by means of direct evidence constituting a body of sound and consistent evidence, the existence of an intention on the part of the applicant to eliminate Icera and, accordingly, that the Commission did not fail to fulfil its obligation to state reasons.

580 Lastly, it must be stated that the applicant has put forward no evidence whatsoever in order to demonstrate that the Commission infringed the right to be heard and the principle of sound administration when it concluded that the applicant’s pricing practices were the implementation of a plan to eliminate Icera.

581 The present plea must therefore be rejected.

 The eleventh plea: a ‘manifest error of appraisal’, of fact and of law, and failure to state reasons as regards the Commission’s rejection of the objective justification put forward by the applicant

582 In support of the eleventh plea, the applicant puts forward the arguments set out below.

583 In the first place, the applicant’s view is that the Commission failed to take account of the fact that (i) in granting the price reductions at issue, it was simply aligning with the ‘aggressive’ prices charged by its competitor Icera and (ii) the prices it charged were above its AVC and, therefore, were profit-maximising and rational. Referring to the fourth and eighth to tenth pleas, it reiterates that, in any event, it did not implement a ‘plan’ to eliminate Icera.

584 In the second place, the applicant maintains that the Commission ought to have taken into account the fact that the price reductions to Huawei for the MDM8200 chipset were granted in response to intense pressure from Huawei on it and were intended (i) to assist Huawei in depleting its order backlog and obsolete inventory and (ii) to compete on price terms with ZTE’s devices incorporating an Icera chip.

585 In the third place, the applicant states that, during the administrative procedure, it provided the Commission with a detailed explanation that it regarded its own stocks of MDM8200 chipsets as excess and obsolete. According to the applicant, that was the reason why it reduced the price of that chipset for all its major customers, which placed Huawei at a disadvantage and ultimately resulted in it granting Huawei a price reduction for orders already placed. It argues that the fact that orders for those chipsets were placed in 2010 and 2011 in no way detracts from the objective justification for its decision to reduce the price of that chipset in order to stimulate new demand. Lastly, it states that, contrary to the Commission’s allegations, it did grant price reductions to all its other significant customers, who actually benefited from prices consistent with the average sales prices granted to Huawei.

586 In the light of the foregoing, the applicant submits that the conduct at issue was objectively justified and that the Commission’s conclusion to the contrary in the contested decision was incorrect and lacked a valid statement of reasons.

587 The Commission and the intervener dispute the applicant’s arguments.

588 It is apparent, in essence, from Article 2 of Regulation No 1/2003 that, in all proceedings for the application of Articles 101 and 102 TFEU, the burden of proving an infringement of those provisions is to rest on the party or the authority alleging the infringement, but that it is, on the other hand, for the undertaking pleading an objective justification to bear the burden of proving it.

589 Furthermore, it is apparent, in essence, from recital 5 of Regulation No 1/2003 that, while it is for the party or the authority alleging an infringement of Article 101 or 102 TFEU to prove such an infringement, it is, by contrast, for the undertaking invoking the benefit of a defence against a finding of infringement to demonstrate to the required legal standard that the conditions for applying such defence are satisfied.

590 It follows that an undertaking, such as the applicant, which pleads an objective justification, including efficiencies, for conduct which is a priori contrary to Article 102 TFEU must itself demonstrate, to the required legal standard, such objective justification.

591 In the present case, it is therefore for the applicant to demonstrate to the requisite legal standard that such conduct must nevertheless be regarded as objectively justified.

592 In that regard, it is apparent from settled case-law that conduct is not abusive if it is justified by pro-competitive advantages or if it serves legitimate interests. In particular, the dominant undertaking may demonstrate, for that purpose, either that its conduct is objectively necessary or that the exclusionary effect produced may be counterbalanced, outweighed even, by advantages in terms of efficiency which also benefit consumers. To that end, it is for the dominant undertaking concerned to demonstrate (i) that its conduct can allow efficiency gains to be achieved, by establishing the actual existence and the extent of those gains, (ii) that such efficiency gains counteract any likely negative effects of that conduct on competition and consumer welfare on the market or markets concerned, (iii) that such conduct is necessary for the achievement of those gains in efficiency and (iv) that it does not eliminate effective competition, by removing all or most existing sources of actual or potential competition (see, to that effect, judgment of 27 March 2012, Post Danmark, C‑209/10, EU:C:2012:172, paragraphs 40 to 42 and the case-law cited). The justifications put forward by the applicant must be assessed in the light of those principles.

593 At the outset, it should be noted that the applicant provides no explanation whatsoever of how the alleged predatory conduct was ‘objectively necessary’ or what advantages consumers could have obtained from it which were capable of counterbalancing the actual or potential effect of eliminating Icera from the market, or even that that conduct was ‘necessary’ for the purpose of achieving alleged efficiency gains for consumers, in accordance with the case-law cited in paragraph 592 above.

594 The three justifications put forward by the applicant for its predatory conduct, namely (i) alignment with Icera’s ‘aggressive’ prices and charging prices above AVC, (ii) price reductions for the MDM8200 chipset granted to Huawei for the purpose of assisting it in depleting its excess and obsolete stock of that chipset and (iii) reductions in the price of that chipset granted to Huawei for the purpose of depleting Qualcomm’s own excess and obsolete stock of that chipset, do not demonstrate either that the conduct at issue was objectively necessary or that the exclusionary effect it produces vis-à-vis Icera could be counterbalanced, outweighed even, by advantages in terms of efficiency which also benefit consumers.

595 When questioned in that regard at the hearing, the applicant simply referred to the pressure exerted by Huawei in order to obtain lower prices for the MDM8200 chipset and the risk that Huawei would cancel certain orders, if necessary by invoking contractual terms. In the light of the case-law cited in paragraph 592 above, such elements do not provide an objective justification for an abuse caught by the prohibition in Article 102 TFEU.

596 It follows that the applicant has not adduced sufficient evidence, in accordance with Article 2 of Regulation No 1/2003 and the case-law cited in paragraph 592 above, that the conduct at issue was objectively justified, which is sufficient to reject the eleventh plea, there being no need to examine whether the Commission made errors of fact, law and appraisal or breached its obligation to state reasons by rejecting the justifications put forward by the applicant during the administrative procedure.

597 In any event, the grounds set out in Section 12.9 of the contested decision are sufficiently reasoned and are not vitiated by errors of assessment or fact.

598 As regards alignment with Icera’s ‘aggressive’ prices, it is sufficient to observe that the case-law does not recognise any absolute right of an undertaking in a dominant position to align its prices on those of its competitors, in particular where that right would in effect justify the use of predatory pricing otherwise prohibited under the Treaty (see, to that effect, judgment of 30 January 2007, France Télécom v Commission, T‑340/03, EU:T:2007:22, paragraph 182).

599 Moreover, the fact that it was allegedly ‘rational’ for the applicant to align with Icera’s prices does not mean that such conduct is objectively ‘justified’ as provided for in the case-law cited in paragraph 592 above, in that it produces efficiencies. The mere fact that conduct may be characterised as economically ‘rational’ in the eyes of the undertaking which implements it is not sufficient to justify it objectively for the purposes of competition law.

600 As regards charging prices above AVC, it should be recalled that, according to the case-law arising from the judgment of 3 July 1991, AKZO v Commission (C‑62/86, EU:C:1991:286, paragraphs 71 and 72), an undertaking in a dominant position, even if it prices below ATC but above AVC, abuses a dominant position if it acts in that way as part of a plan for eliminating a competitor. Where there is such a strategy to eliminate, as in the present case, it is therefore irrelevant that the prices charged were allegedly ‘profit-maximising and rational’ in that they were above AVC.

601 As regards the price reductions for the MDM8200 chipset granted to Huawei for the purpose of assisting it in depleting its excess and obsolete stock of that chipset, it must be observed that those reductions concern only the prices offered to Huawei for the MDM8200 chipset. In the explanations put forward, the applicant therefore makes no attempt to justify the predatory prices granted to ZTE or the predatory prices granted for the MDM6200 and MDM8200A chipsets.

602 It should also be noted, as regards the price reductions for the MDM8200 chipset granted to Huawei for the purposes of assisting it in depleting its excess and obsolete stock of that chipset, that, as the Commission states in recital 1185 of the contested decision, Huawei submitted further orders for that chipset with Qualcomm in 2010 and 2011, which the applicant confirmed when questioned in that regard at the hearing. The Commission therefore did not err in finding that it was hardly credible that those price reductions could have been granted for the purposes of assisting Huawei to deplete its excess stock of an obsolete chipset.

603 Similarly, it is hardly credible that the purpose of the price reductions for the MDM8200 chipset granted to Huawei by the applicant was to deplete Huawei’s own excess stock and obsolete stock of that chipset. As the Commission correctly points out, if that had been so, the applicant would have granted price reductions for that chipset to all of its customers. It is apparent from Table 74 in the contested decision that only Huawei benefited from such reductions and was offered prices substantially lower than those charged by the applicant to its other main customers. It is also apparent from recital 1185 of that decision that in August 2010 the applicant had to produce new units of the MDM8200 chipset in order to satisfy Huawei’s demand, which the applicant, when questioned in that regard at the hearing, also confirmed. Lastly, the applicant’s argument that the price reductions granted to Huawei had also been made in order to compete with ZTE devices using an Icera chip simply confirms the indirect predation found by the Commission.

604 Consequently, the eleventh plea must in any event be rejected.

 The third plea: a failure ‘to apply the correct legal standard thereby committing manifest errors of law’

605 The third plea comprises three parts. The first alleges a ‘manifest error of law’ and an infringement of the principles of legal certainty and legitimate expectations, on account of the Commission’s failure to apply the ‘correct legal standard’ as defined in the Enforcement Priorities Communication. The second alleges misapplication of the case-law on predatory pricing and on pricing practices more generally. The third part alleges infringement of the principles of legal certainty and foreseeability.

 The first part: failure to apply the ‘correct legal standard’ as defined in the Enforcement Priorities Communication

606 According to the applicant, it is apparent from the Enforcement Priorities Communication that the relevant test when examining a potential eliminatory practice implemented by an undertaking in a dominant position is whether such a practice was likely to lead to the foreclosure of an ‘as-efficient’ competitor. In accordance with that notice, the Commission, when conducting such an examination, is also required to demonstrate that there is evidence, first, that the undertaking in a dominant position, in engaging in that practice, deliberately incurred losses or forewent profits in the short term and, second, that it could hope to recoup such losses after the presumed period of predation. Since the Commission publicly committed to follow the approach in that notice, it created legitimate expectations for the undertakings concerned.

607 Furthermore, the Enforcement Priorities Communication does not allow the Commission to penalise an undertaking which has acted in good faith in a way which, ultimately, has proved to be less profitable than it had initially envisaged. Lastly, in the applicant’s view, it is apparent from point 66 of the Enforcement Priorities Communication that the Commission can rely only exceptionally on documentary evidence, provided that that evidence demonstrates with sufficient clarity the existence of a predatory strategy.

608 In the present case, the Commission did not apply the Enforcement Priorities Communication, as demonstrated by the absence of any reference to that document in either the SO, the SSO or the LoF. Referring to the fourth and eighth to tenth pleas, the applicant furthermore submits that the Commission has in no way demonstrated in the contested decision any profit sacrifice by the applicant, has failed to conduct an ‘as-efficient’ competitor test and has disregarded evidence showing that, in the absence of the conduct in question, Huawei and ZTE would not in any event have obtained supplies from Icera. Referring in addition to the eleventh plea, the applicant submits that the Commission made manifest errors of appraisal and of fact in rejecting the objective justification which it had put forward. Lastly, referring to the tenth and eleventh pleas in law, it submits that the conduct at issue was implemented in good faith, since it could not have foreseen the methodology applied by the Commission to allocate R&D costs in that decision and that it could have reasonably expected its pricing practices to be profitable.

609 The applicant submits that, consequently, the Commission committed a ‘manifest error of law’ and infringed the principles of legal certainty and legitimate expectations, as well as the principle of sound administration and the obligation to state reasons.

610 The Commission disputes the arguments put forward by the applicant.

611 As a preliminary point, it should be noted that the Enforcement Priorities Communication merely sets out the Commission’s approach as to the choice of cases that it intends to pursue as a matter of priority; such that the administrative practice followed by the Commission is not binding on national competition authorities or courts (judgment of 6 October 2015, Post Danmark, C‑23/14, EU:C:2015:651, paragraph 52), and is not intended to constitute a statement of the law (judgment of 13 December 2018, Deutsche Telekom v Commission, T‑827/14, EU:T:2018:930, paragraph 114).

612 It follows that, like any other document which, although not binding, defines the general framework of analysis which the Commission employs in determining whether to intervene, the Enforcement Priorities Communication limits the Commission’s discretion. It follows that the Commission cannot, in a particular case, depart from it without providing reasons. In such a case, the Commission’s respect for the rights guaranteed by the EU legal order in administrative procedures, including the obligation to state reasons, is of even more fundamental importance (see, to that effect and by analogy, judgment of 10 November 2017, Icap and Others v Commission, T‑180/15, EU:T:2017:795, paragraph 289 and the case-law cited).

613 It is therefore necessary to examine whether, as the applicant claims, in the present case the Commission departed, without justification, from the general framework of analysis set out in the Enforcement Priorities Communication.

614 As regards, in the first place, the applicant’s argument that, in essence, the Commission did not examine whether the practice at issue was likely to lead to the foreclosure of an ‘as-efficient’ competitor in accordance with the Enforcement Priorities Communication, it is apparent from the examination of the first part of the ninth plea, and in particular from paragraphs 524 to 527 above, that the analysis by which the Commission compares the prices charged by an undertaking in a dominant position with certain of its costs for the purposes of assessing whether that undertaking charged predatory prices, includes an ‘as-efficient’ competitor analysis and that, since the Commission has proven, as was done in the present case, that the dominant undertaking had charged such prices, it has therefore implicitly conducted such an analysis, which is sufficient to reject that argument.

615 Moreover, it is apparent from paragraph 26 of the Enforcement Priorities Communication that failure to cover LRAIC, which in the present case the Commission demonstrated in the contested decision, indicates that the dominant undertaking is not recovering all the (attributable) fixed costs of producing the good or service in question and that an as-efficient competitor could be foreclosed from the market, which confirms that, in comparing the dominant undertaking’s prices and costs, the Commission implicitly, but necessarily, carried out an ‘as-efficient’ competitor test.

616 As regards, in the second place, the applicant’s argument that the Commission was required, pursuant to the Enforcement Priorities Communication, to adduce evidence that the undertaking in a dominant position had deliberately incurred losses or sacrificed profits in the short term, it is apparent from point 66 of that communication that it is possible in some cases for the Commission, in order to demonstrate the existence of such sacrifice, to rely on direct evidence consisting of documents from the dominant undertaking which clearly show a predatory strategy, such as a detailed plan, in particular, to sacrifice in order to exclude a competitor.

617 As is apparent from the examination of the tenth plea, the Commission did in fact rely, in Section 12.8.1 of the contested decision, on documents from the applicant clearly showing the existence of an intention on the part of the applicant to eliminate Icera, implemented via price reductions granted to Huawei and ZTE, which is sufficient to reject the present argument, which, in essence, alleges that there was no evidence of sacrifice in the contested decision. In that regard, contrary to the applicant’s allegations, it is in no way apparent from paragraph 66 of the Enforcement Priorities Communication that the Commission can rely only exceptionally on documentary evidence demonstrating the existence of a predatory strategy. On the contrary, in that paragraph, the Commission uses the expression ‘in some cases’, and not the expression ‘in some exceptional cases’ or ‘exceptionally’, or even the expression ‘in exceptional circumstances’, used in footnote 50 to the Enforcement Priorities Communication. In footnote 44 to that communication, which relates to point 66 of that communication, the Commission refers, moreover, to two judgments in which the Court confirmed that the Commission was entitled to rely on such documentary evidence, namely the judgments of 6 October 1994, Tetra Pak v Commission (T‑83/91, EU:T:1994:246, paragraphs 151 and 171), and of 30 January 2007, France Télécom v Commission (T‑340/03, EU:T:2007:22, paragraphs 198 to 215).

618 As regards, in the third place, the applicant’s argument that the Commission was required, pursuant to the Enforcement Priorities Communication, to adduce evidence that the undertaking in a dominant position could, after that period, hope to recoup the losses made during the period of predation, it is apparent solely from point 70 of that communication that the dominant undertaking can reasonably expect its market power after the predatory conduct comes to an end to be greater than it would have been had the undertaking not engaged in that conduct in the first place, and that it is therefore likely to be in a position to benefit from the sacrifice. Moreover, according to the case-law, conduct can constitute abuse, there being no need to demonstrate specifically that the undertaking in question had a reasonable prospect of recouping losses so incurred (see, to that effect, judgment of 6 October 1994, Tetra Pak v Commission T‑83/91, EU:T:1994:246, paragraph 150), which is sufficient to dismiss the present argument.

619 As regards, in the fourth place, the applicant’s argument that, since it could not have foreseen the methodology applied by the Commission to allocate R&D costs in the contested decision and, consequently, having acted in good faith, it cannot, in accordance with the Enforcement Priorities Communication, be penalised by the Commission, it should be noted that, in the examination of the tenth plea, it was found that the Commission did not err in concluding that the applicant’s pricing practices were the implementation of a plan to eliminate Icera, which is sufficient to reject the present argument.

620 As regards, in the fifth place, the applicant’s argument that the SO, the SSO and the LoF did not refer to the Enforcement Priorities Communication, the applicant provides no explanation whatsoever as to how that means that the Commission failed to have regard to the general framework of analysis set out in that communication used by it in determining whether to intervene. Moreover, the Enforcement Priorities Communication in no way requires the Commission to make an express reference to it in its decisions on abuse of a dominant position. Consequently, the present argument must be rejected.

621 As regards, in the sixth place, the applicant’s argument that the Commission did not take account of the fact that Huawei and ZTE would not, in any event, have obtained more supplies from Icera in the absence of the conduct at issue and that it erred in rejecting the objective reasons put forward in order to justify its conduct, it must be observed that the applicant does not explain how the Commission thereby failed to have regard to the general framework of analysis set out in the Enforcement Priorities Communication used by it in determining whether to intervene.

622 It follows from the foregoing that the first part of the third plea must be rejected.

 The second part: misapplication of the case-law on predatory pricing and on pricing practices more generally

623 The applicant alleges that the Commission fails to have regard to the case-law on predatory pricing and on pricing practices more generally, that is to say the case-law establishing the applicable legal standard. In support of that allegation, it sets out a number of extracts from judgments which it regards as relevant and criticises the Commission for having failed to carry out, in the contested decision, an ‘as-efficient’ competitor analysis, even though that analysis is required by the case-law.

624 The Commission disputes the arguments put forward by the applicant.

625 It should be noted that the applicant simply set outs extracts from case-law, without explaining how the contested decision departs from the legal standard laid down in such case-law, except in relation to the ‘as-efficient’ competitor analysis in relation to which, after citing the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632) it complains that in the present case, first, the Commission failed to carry out such an analysis and, second, that it failed to ascertain the size of the share of the market covered by the contested practice.

626 First, it is apparent from the examination of the first part of the ninth plea, and in particular from paragraphs 524 to 527 above, that the analysis by which the Commission compares the prices charged by an undertaking in a dominant position with certain of its costs for the purposes of assessing whether that undertaking charged predatory prices already includes an ‘as-efficient’ competitor analysis and that, since Commission has proven, in the present case, that the dominant undertaking had charged such prices and, therefore, implicitly conducted that analysis, that is sufficient to reject the applicant’s complaint that the contested decision did not contain such an analysis.

627 Second, as is apparent from paragraphs 521 to 523 above, the Commission is not, in accordance with the case-law cited in paragraph 520 above, required, when examining whether an undertaking in a dominant position charged predatory prices, also to ascertain whether the share of the market covered by the contested practice is of sufficient magnitude for that practice to have anticompetitive effects, which is sufficient to reject the applicant’s complaint alleging such failure to ascertain.

628 The second part of the third plea must therefore be rejected.

 The third part: breach of the principles of legal certainty and foreseeability

629 Referring to the seventh and eighth pleas, the applicant submits that, in the contested decision, the Commission applied a contrived and novel price-cost test which differed from such test set out in the SO in relation to two of the chipsets concerned and was unforeseeable in respect of the third. According to the applicant, the Commission thereby infringed the principles of legal certainty and foreseeability.

630 The Commission disputes the arguments put forward by the applicant.

631 It must be stated that, in support of the present part of the plea, the applicant does no more than provide a general reiteration of the allegations made in the first part of the present plea and the seventh and eighth pleas, to which it makes express reference, without developing any specific argument.

632 Consequently, the third part of the third plea must be rejected as being inadmissible.

633 It follows from all of the foregoing that the third plea must be rejected.

 The fourth plea: the ‘“predation theory” is illogical and unsupported by evidence’

634 The fourth plea consists of 13 parts. The first part alleges that the Commission’s ‘predation theory’, which is ‘targeted and selective’, is disconnected from the market definition and was tailored to reach a predetermined conclusion. The second part alleges that that theory is inherently inconsistent and lacks adequate reasoning. The third alleges that the Commission failed to demonstrate that the applicant’s conduct was irrational and failed to establish the alleged predation strategy. The fourth alleges that the applicant’s prices were higher than those of Icera and, therefore, that there was no market foreclosure. The fifth alleges that the Commission disregarded crucial exculpatory evidence. The sixth alleges that the decision fails to show financial predation. The seventh alleges that the ‘predation theory’ is novel, flawed and unsupported. The eighth alleges ‘manifest errors of appraisal’ in relation to direct predation between the third quarter of 2010 and the second quarter of 2011. The ninth alleges that the Commission disregarded commercial reality regarding the MDM8200 chipset. The tenth alleges that that theory is contrary to logic and elementary economics. The eleventh alleges that the Commission failed to show recoupment of the losses sustained during the relevant period or explain when the alleged predation ended. The twelfth alleges failure to apply the ‘as-efficient’ competitor test and that Icera’s good performance during the relevant period disproves the predation theory. The thirteenth part alleges that the Commission erred in punishing genuine competition.

 Preliminary observations

635 It should be observed that, in Section 12 of the contested decision, the Commission sets out its analysis of the abuse committed by the applicant.

636 In particular, in Section 12.1 of the contested decision, the Commission begins by recalling the applicable general principles, in particular in relation to predatory pricing. Next, in Section 12.2 of that decision, it summarises how it applied those general principles in Sections 12.3 to 12.11 of that decision, which the applicant describes as a ‘predation theory’.

637 In that context, the Commission explains in recital 334 of the contested decision that it reached the conclusion that from 1 July 2009 until 30 June 2011, the applicant supplied certain quantities of three of its UMTS chipsets, namely the MDM8200-, MDM6200- and MDM8200A-based chipsets, to two of its key customers, namely Huawei and ZTE, below cost, with the intention of eliminating Icera, its main competitor in the leading edge segment of the UMTS chipset market. According to the Commission, in so doing, the applicant intended to prevent Icera, a small, financially constrained start-up company, from gaining sufficient reputation and scale to be able to challenge its dominance on that market. The Commission states that it concluded, in the light of the applicant having put forward no valid objective justification for such conduct, that the applicant had, during that period, thereby committed abuse of a dominant position contrary to Article 102 TFEU and Article 54 of the EEA Agreement.

 The first part: the Commission’s ‘predation theory’, which is ‘targeted and selective’, is disconnected from the market definition and was tailored to reach a predetermined conclusion

638 The applicant, referring to the tenth plea, claims that the conduct at issue was neither targeted nor selective and that the Commission’s ‘predation theory’ is disconnected from the definition of the relevant market, since, as it relates only to the leading edge segment of that market, which was, moreover, not defined, certain chipsets regarded by the applicant as ‘leading edge’ were excluded from that segment.

639 The applicant also claims that the ‘predation theory’ was developed by the Commission as a theory of convenience, since that theory is unconnected with the complaint. According to the applicant, the MDM8200 and MDM8200A chipsets were sold to customers other than Huawei and ZTE before and after the relevant period, at prices below the LRAIC calculated by the Commission, which calls that theory into question.

640 The Commission disputes the arguments put forward by the applicant.

641 First of all, as regards the applicant’s allegation that the conduct at issue was neither targeted nor selective, it should be recalled that it is apparent from the direct evidence on which the Commission relied, which is examined in more detail in paragraphs 538 and 539 above in the context of the examination of the tenth plea, that the applicant granted price reductions to Huawei and to ZTE, which were clearly identified in those documents, in order to prevent Icera from gaining access to those two crucial customers and, in so doing, to eliminate it from the market. The applicant cannot therefore validly claim that its conduct was neither targeted nor selective.

642 Next, the applicant’s allegation that the Commission’s ‘predation theory’, which relates solely to the leading edge segment, is disconnected from the market definition also cannot succeed. It is apparent from point 64 of the Enforcement Priorities Communication that, as regards predation, conduct entails a sacrifice if the dominant undertaking charges a lower price for all or a particular part of its output. It follows that predatory conduct can involve a limited segment of the market concerned, and not the whole of that market. In that regard, it is clear from the case-law that predatory conduct can even take place on a market other than the market concerned, which has not been strictly defined (see, to that effect, judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 35 and 45).

643 It follows that the Commission was (i) entitled to penalise, in the contested decision, conduct limited to a segment of the relevant market and (ii) not required to define precisely the scope of that segment, as it is obliged to do for the relevant market for the purposes of ascertaining whether there is a dominant position.

644 In addition, as regards the alleged lack of connection between the Commission’s ‘predation theory’ and the complaint, it must be noted that it is apparent from paragraph 55 of the Commission Notice on the handling of complaints by the Commission under Articles [101] and [102 TFEU] (OJ 2004 C 101, p. 65) that, during the first stage following the submission of the complaint, the Commission examines the complaint and may collect further information in order to decide what action it will take on the complaint, that that stage may include an informal exchange of views between the Commission and the complainant with a view to clarifying the factual and legal issues with which the complaint is concerned, and that the Commission may, in this stage, give an initial reaction to the complainant, allowing the complainant an opportunity to expand on its allegations in the light of that initial reaction.

645 In the present case, it is apparent from recitals 11 to 13 of the contested decision that, following the updating and revision of the complaint, a number of exchanges took place between Icera and the Commission, which ultimately led, in June 2012, to Icera making allegations of predation, which the applicant itself acknowledges. In the light of paragraph 55 of the Commission Notice on the handling of complaints by the Commission under Articles [101] and [102] TFEU, it must be held that it is standard practice for a complainant, as was so in the present case, to expand on the allegations made in the complaint in order to take account of the Commission’s initial reaction. Furthermore, it follows from the need to ensure effective enforcement of the competition rules that the Commission cannot be bound by the outline and legal assessments made by a complainant. In any event, it must be observed that the applicant does not explain how the fact that Icera expanded its predation allegations only at a relatively late stage could call into question what it describes as a ‘predation theory’ in that decision, following a thorough investigation by the Commission.

646 Lastly, as regards the applicant’s allegation that it charged prices below LRAIC for the MDM8200 and MDM8200A chipsets to customers other than Huawei and ZTE before and after the relevant period, it must be observed that the applicant provides no explanation whatsoever as to how that could call into question what it describes as a ‘predation theory’ in the contested decision. On the contrary, it rather appears that, if the Commission had discovered evidence of the applicant having charged predatory prices to other customers outside that period, it could have taken that into account in order to strengthen that theory or extend the time during which that theory applied.

647 It follows from the foregoing that the first part of the fourth plea must be rejected.

 The second part: the Commission’s ‘predation theory’ is inherently inconsistent and lacks adequate reasoning

648 The applicant alleges a number of contradictions and inconsistencies which vitiate the contested decision.

649 In the first place, the applicant’s view is that the ‘predation theory’, which is also apparent from recital 993 of the contested decision, according to which the applicant first used the MDM8200 chipset to migrate its customers Huawei and ZTE from those chipsets to the more advanced MDM6200 and MDM8200A chipsets and then focused its pricing strategy on the MDM8200A chipset, in view of the initial lack of success of the MDM6200 chipset, is not compatible with a predatory pricing ‘theory of harm’, since such a finding would merely reflect the existence of vigorous competition from the applicant by launching new products.

650 In the second place, the applicant states that the Commission criticises it in the contested decision for having implemented the conduct at issue in order to prevent Icera from gaining a reputation, even though it is apparent from an internal document, referred to in that decision, that it considered that, several months before the relevant period, Icera had managed to overcome the reputational barrier, which calls into question the Commission’s ‘predation theory’.

651 In the third place, the applicant maintains that the Commission could not, without contradicting itself, (i) state in recitals 411 to 419 of the contested decision that the MDM6200 chipset was not commercially successful during the relevant period and (ii) claim in recital 420 of that decision that the applicant relied on that chipset in its strategy towards ZTE. Similarly, it considers that the Commission cannot (i) claim, in recital 959 of that decision, that the MDM8200 chipset maintained its competitive significance even after the MDM8200A chipset had come commercially available and (ii) state in recital 420 of that decision that the MDM8200A chipset had become the crucial product on which it relied to implement its strategy in 2010 and 2011.

652 The Commission disputes the arguments put forward by the applicant.

653 It must be stated at the outset that the present part of the plea must be rejected as being ineffective, since, even if the three arguments put forward in support of that part of the plea were well founded, that would not affect the lawfulness of the contested decision.

654 The fact that the applicant allegedly competed vigorously by launching new products, that Icera succeeded in gaining a reputation, or that the applicant could not have applied its predatory strategy to the MDM6200 chipset because of its lack of commercial success or to the MDM8200 chipset after the launch of the MDM8200A chipset, does not refute the Commission’s finding that the applicant charged predatory prices with the intention of eliminating its competitor Icera, which is sufficient to demonstrate the existence of abuse by the applicant (judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 70 and 71).

655 The second part of the fourth plea must therefore be rejected.

 The third part: the Commission failed to demonstrate that the applicant’s conduct was irrational and fails to establish the alleged predation strategy

656 The applicant submits that, in the contested decision, the Commission did not demonstrate that its conduct was irrational but for the intention to eliminate Icera, or that its prices were the implementation of a plan to exclude Icera. In its view, the selected documents on which the Commission relies when attempting to demonstrate the existence of such a plan were drafted by low-level employees with no power to make pricing decisions. It adds that the Commission has adduced no evidence of sacrifice on the part of the applicant, even though the Commission was required to demonstrate the existence of such sacrifice, in accordance with the Enforcement Priorities Communication.

657 The Commission disputes the arguments put forward by the applicant.

658 It should be observed that, in support of the present part of the plea, the applicant simply repeats arguments which it put forward in the context of the first part of the third plea, as regards the alleged lack of evidence of a sacrifice having occurred, or in the context of the tenth plea, concerning the alleged lack of evidence of there having been a strategy to eliminate Icera.

659 Those arguments, which were rejected in the context of the examination of the first part of the third plea and in the context of the tenth plea, must be rejected on the same grounds in so far as they are raised in support of the third part of the fourth plea, which is sufficient to reject the third part of the fourth plea.

 The fourth part: the applicant’s prices were higher than those of Icera

660 The applicant submits that, in the contested decision, the Commission acknowledges, first, that the applicant’s prices were higher than Icera’s prices and, second, that Icera’s chips were of superior quality. In its view, if that were the case, it would therefore have been logical for it to reduce the price of its chipsets to compensate for the inferior quality of its products, which the Commission did not take into account.

661 The Commission disputes the arguments put forward by the applicant.

662 It should be recalled that an undertaking in a dominant position abuses that position if it charges prices below AVC, or prices below ATC but above AVC, where such prices are determined as part of a strategy to eliminate a competitor (judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 70 to 72).

663 In the present case, it is apparent from the examination of the fifth to eighth and tenth pleas that the Commission did duly demonstrate that the applicant had charged predatory prices with the intention of eliminating its competitor Icera, which is sufficient to demonstrate the existence of abuse of a dominant position by the applicant.

664 Furthermore, it is apparent from the examination of the eleventh plea that the Commission did not err and did not fail to fulfil its obligation to state reasons when it rejected the objective justifications put forward by the applicant, including the applicant’s allegation that it had charged predatory prices in response to Icera’s aggressive prices.

665 The Commission cannot therefore validly be criticised for having failed to take account of the fact that the predatory prices charged by the applicant had been justified by Icera’s prices being lower or by the applicant’s products being of inferior quality. Moreover, the benchmark for assessing whether the pricing practices were predatory is based not on the prices and costs of the undertaking against which such practices are put in place or on the quality of the products concerned, but rather on the prices and costs of the dominant undertaking.

666 The fourth part of the fourth plea must therefore be rejected.

 The fifth part: the Commission disregarded crucial exculpatory evidence

667 The applicant claims that the Commission disregarded certain ‘crucial’ exculpatory evidence, namely, in the first place, the fact that Huawei and ZTE had stated that they would not have obtained more supplies from Icera, absent the alleged predation, in the second place, the fact that Icera’s chips were technically outperformed and that Icera would not in any event have been able to develop an LTE chip before it and, in the third place, the fact that Icera had not been foreclosed from the market during the relevant period, since Icera had succeeded in attracting significant external investments and had been acquired by Nvidia.

668 The Commission disputes the arguments put forward by the applicant.

669 It should be recalled that, according to the case-law, an undertaking in a dominant position abuses that position if it charges prices below AVC, or prices below ATC but above AVC, where such prices are determined as part of a strategy to eliminate a competitor (judgment of 3 July 1991, AKZO v Commission, C‑62/86, EU:C:1991:286, paragraphs 70 to 72). In the present case, it is apparent in particular from the examination of the fifth to eighth and tenth pleas that the Commission did duly demonstrate that the applicant had charged such predatory prices with the intention of eliminating its competitor Icera, which is sufficient to demonstrate the existence of abuse of a dominant position by the applicant.

670 Therefore, the fact that the two customers concerned by the conduct at issue would not in any event have obtained more supplies from Icera in the absence of such conduct, that Icera’s technology was technically outperformed or that Icera attracted external financing and was acquired during the relevant period, even if they were proven, could not have influenced the Commission’s conclusion that there was abuse, based on the finding that the applicant priced below its costs with the intention of eliminating its competitor Icera. That is all the more true given that those three factors relied on by the applicant are factors external to the applicant, over which it had no influence and those factors cannot therefore mean that the applicant is exempt from any penalty for the infringement which it committed.

671 In any event, it must also be stated that the applicant does no more than describe the three elements in question as ‘crucial’ without explaining how they could, if the Commission had taken them into account, have called into question what it describes as a ‘predation theory’.

672 The fifth part of the fourth plea must therefore be rejected.

 The sixth part: the decision fails to show financial predation

673 The applicant maintains that the predation relied on in the contested decision is not supported by credible evidence and is contradicted by the fact that Icera was acquired by Nvidia during the relevant period. It adds that it disputed the existence of ‘financial predation’ during the administrative procedure, which led the Commission to abandon it in that decision and merely to rely on ‘traditional predation’.

674 The Commission disputes the arguments put forward by the applicant.

675 It is sufficient, in order to reject the present part of the plea, to observe that, by the applicant’s own admission, the Commission did not base the contested decision on a case of financial predation, but rather on traditional predation, that is to say, an undertaking in a dominant position charging prices below cost with the intention of eliminating a competitor.

676 Therefore, the applicant’s argument alleging that there was no evidence of the existence of financial predation must be declared to be ineffective and the sixth part of the fourth plea must be rejected.

 The seventh part: the ‘predation theory’ is novel, flawed and unsupported

677 The applicant explains that the Commission’s ‘predation theory’, according to which it provided Huawei with MDM8200 chipsets at predatory prices so that Huawei could persuade mobile network operators to favour devices incorporating the applicant’s chipsets, rather than ZTE devices incorporating Icera chips, is untenable.

678 According to the applicant, the Commission imagined the ‘predation theory’ solely because for the period from July 2009 to June 2010 it had discovered predatory pricing evidence regarding sales of MDM8200 chipsets only in relation to Huawei, but not in relation to ZTE.

679 The Commission disputes the applicant’s arguments.

680 It must be recalled that it is apparent in particular from the examination of the fifth to eighth and tenth pleas that the Commission did duly demonstrate that the applicant had charged predatory prices, in particular to Huawei for the MDM8200 chipset, with the intention of eliminating its competitor Icera, which the applicant does not dispute in the context of the present part of the plea.

681 Therefore, it is necessary to declare ineffective the applicant’s argument against the Commission’s finding that the applicant initially offered only predatory prices to Huawei and not, at that stage, to ZTE, so that Huawei’s devices incorporating one of the applicant’s chipsets would be favoured over ZTE devices incorporating an Icera chip, since that argument does not call into question what it describes as a ‘predation theory’.

682 The applicant also claims that the Commission developed the ‘predation theory’ because, for the period from July 2009 to June 2010, it had discovered predatory pricing evidence in relation to sales of MDM8200 chipsets only in relation to Huawei, but not in relation to ZTE. Evidence that the applicant charged predatory prices to Huawei with the intention of eliminating Icera was sufficient to demonstrate that there was abuse. The Commission was, consequently, under no obligation also to demonstrate predatory pricing in respect of ZTE in order to impose penalties on the applicant.

683 The seventh part of the fourth plea must therefore be rejected.

 The eighth part: ‘manifest errors of appraisal’ in relation to direct predation between the third quarter of 2010 and the second quarter of 2011

684 Referring to the sixth and seventh pleas, the applicant argues that the Commission made three errors, two of which concern the allocation of R&D costs, with the remaining error concerning the allocation of the NRE payment to ZTE.

685 The Commission disputes the applicant’s arguments.

686 It must be recalled that those arguments were examined and rejected in the examination of the sixth and seventh pleas, which is sufficient to reject the eighth part of the fourth plea.

 The ninth part: the Commission disregarded commercial reality regarding the MDM8200 chipset

687 The applicant submits that the MDM8200 chipset encountered technical issues, which resulted in Huawei accumulating excess inventories of that chipset, and that it was therefore entirely rational for it to grant price reductions for that chipset. In its view, such price reductions were not granted with an intention to predate.

688 The Commission disputes the applicant’s arguments.

689 As is apparent in particular from the examination of the fifth to eighth and tenth pleas, the Commission did duly demonstrate that the applicant had charged predatory prices with the intention of eliminating its competitor Icera. In addition, as is apparent from the examination of the eleventh plea, the Commission did not err and did not fail to fulfil its obligation to state reasons when it rejected the objective justifications put forward by the applicant, including the applicant’s allegation that its predatory prices had been charged in order to assist Huawei in depleting its excess stock of the MDM8200 chipset. In addition, as the Commission correctly points out in its pleadings, the fact that predatory conduct is potentially rational does not make it lawful.

690 Therefore, the Commission cannot validly be criticised for having failed to take account of commercial reality regarding the MDM8200 chipset, namely the technical issues which that product encountered.

691 The ninth part of the fourth plea must therefore be rejected.

 The tenth part: the ‘predation theory’ is contrary to logic and elementary economics

692 In order to demonstrate that the ‘predation theory’ was illogical and contrary to the rules of economics, the applicant submits that its prices were higher than Icera’s prices and that the volume of chipsets offered at predatory prices was, during the relevant period, negligible compared with the volume of chipsets sold on the relevant market and in the leading edge segment of that market.

693 The Commission disputes the applicant’s arguments.

694 As is apparent in particular from the examination of the fifth to eighth and tenth pleas above, the Commission did duly demonstrate that the applicant had charged predatory prices as referred to in the judgment of 3 July 1991, AKZO v Commission (C‑62/86, EU:C:1991:286), with the intention of eliminating its competitor Icera, which is sufficient to demonstrate the existence of abuse by the applicant. Therefore, as was stated in the examination of the fourth part of the present plea, the Commission cannot validly be criticised for having failed to take account of the fact that the predatory prices charged by the applicant were higher than Icera’s prices.

695 Furthermore, as is apparent from the analysis of the first part of the ninth plea, and in particular from paragraphs 521 to 523 above, the Commission is not required, when examining whether an undertaking in a dominant position charged predatory prices, to ascertain whether the share of the market covered by the practice at issue is of sufficient magnitude for that practice to have anticompetitive effects.

696 Accordingly, the tenth part of the fourth plea must be rejected.

 The eleventh part: the Commission failed to show recoupment of the losses sustained during the relevant period or explain when the alleged predation ended

697 Referring to the third plea, the applicant submits, in the first place, that the Commission has not adduced evidence that it recouped losses sustained during the period following the infringement period, even though the applicant was required to adduce such evidence, in accordance with the Enforcement Priorities Communication. In the second place, it criticises the Commission for not having adduced evidence, in the contested decision, of the date on which its allegedly predatory intention came to an end.

698 The Commission disputes the applicant’s arguments.

699 As is apparent from the examination of the third plea, and in particular from paragraph 618 above, the Commission is not required, in accordance with, inter alia, the Enforcement Priorities Communication, to demonstrate that the undertaking in a dominant position recouped the losses suffered during the period of predation.

700 Furthermore, it is clear that there is nothing to oblige the Commission to identify a final stage marking the end of predation and that, the fact that, in the contested decision, the Commission does not go on to analyse the period after 30 June 2011, simply means that, from that date, it no longer identified sufficient evidence of the existence of predatory conduct on the part of the applicant.

701 The eleventh part of the fourth plea must, consequently, be rejected.

 The twelfth part: failure to apply the ‘as-efficient’ competitor test and allegation that Icera’s good performance during the relevant period disproves the ‘predation theory’

702 Referring to the ninth plea, the applicant submits, first of all, that the Commission did not apply an ‘as-efficient’ competitor test. It then explains that the Commission did not take account of evidence demonstrating that Icera had not been foreclosed from the market during the relevant period and had even achieved very good results, which calls into question the ‘predation theory’.

703 The Commission disputes the applicant’s arguments.

704 It should be recalled, first of all, that it is apparent from the examination of the first part of the ninth plea, and in particular from paragraphs 524 to 527 above, that the analysis by which the Commission compares the prices charged by an undertaking in a dominant position with certain of its costs for the purposes of assessing whether that undertaking charged predatory prices includes an ‘as-efficient’ competitor analysis and that, since the Commission has proven, in the present case, that the dominant undertaking had charged such prices and, therefore, implicitly conducted such an analysis, that is sufficient to reject the applicant’s complaint that the contested decision did not contain an ‘as-efficient’ competitor analysis.

705 Next, it should be recalled, as is apparent from paragraphs 520 and 521 above, that the Commission is not required to prove that predatory conduct had anticompetitive effects, which is sufficient to reject the complaint alleging that Icera was not actually eliminated during the relevant period.

706 The twelfth part of the fourth plea must therefore be rejected.

 The thirteenth part: the Commission erred in punishing genuine competition

707 The applicant submits that it was merely engaged in vigorous, but unremarkable competition on the merits, and that charging low prices, which was penalised in the present case, is in fact pro-competitive. In its view, the present case would represent the first time that the Commission found predation due to a failure to recover R&D costs.

708 The Commission disputes the applicant’s arguments.

709 As is apparent from the examination of the tenth plea, the Commission did duly demonstrate, in the contested decision, that the applicant had charged predatory prices to Huawei and ZTE for the MDM8200, MDM6200 and MDM8200A chipsets with the intention of eliminating Icera, relying both on direct evidence, namely internal Qualcomm documents, and on indirect evidence, namely contextual elements. It follows that the applicant cannot validly claim that, in fact, in so doing its actions were pro-competitive.

710 As regards, in addition, the applicant’s allegation that in the present case the Commission finds, for the first time, predation due to the failure to cover R&D costs, it must be stated that such an argument was examined and rejected in the examination of the seventh plea and must be rejected for the same reasons in so far as it is raised in support of the present part of the present plea.

711 The thirteenth part of the fourth plea must, consequently, be rejected. Therefore, since the other 12 parts of the present plea have also been rejected, the fourth plea must be rejected in its entirety.

 The eighth plea: the price-cost analysis is ‘manifestly incorrect’

712 The eighth plea comprises three parts. The first relates to the starting period of the price-cost test. The second relates to the most fundamental errors made in the price-cost test. The third relates to the failure to correct those errors in the contested decision.

713 Given that the second and third parts of the present plea both concern, in essence, errors which the Commission allegedly made in the price-cost test, and which the Commission did not correct in the contested decision, they will be examined together.

 Preliminary observations

714 In Section 12.7.1 of the contested decision, the Commission conducted a quarterly price-cost test for the three chipsets concerned and in respect of the two customers concerned, for the purposes of ascertaining whether the applicant had charged prices below AVC or LRAIC.

715 In Section 12.7.1.1 of the contested decision, the Commission began by explaining that, at the beginning of a chipset’s commercial cycle, its AVC were unusually high and that it was only once it had reached large-scale production that such AVC stabilised. For that reason, the Commission then stated that it had not taken into account, in its price-cost test, quarters prior to the large-scale production stage and that such an approach was favourable to the applicant, since that test would otherwise have led, for those quarters, to a finding of predatory prices, in the light of those artificially high production costs at the beginning of the cycle (recital 940 of the contested decision). Lastly, it determined the quarter in which each of the three chipsets concerned had, in its view, reached such large-scale production and which formed the starting point for its test, namely the third quarter of 2009 for the MDM8200 chipset and the third quarter of 2010 for the MDM6200 and MDM8200A chipsets (recitals 941 to 943 of the contested decision).

716 The Commission then conducted its price-cost test and concluded that the applicant had charged predatory prices for the three chipsets concerned and in relation to the two customers concerned between 1 July 2009 and 30 June 2011, as is apparent in particular from Tables 55 to 57 of the contested decision.

 The first part: the starting period of the price-cost test

717 The applicant submits that the Commission incorrectly, and without providing adequate reasoning, determined the starting period of the price-cost test.

718 In particular, the applicant criticises the Commission for having concluded, in an arbitrary manner and without adequate reasoning, that a chipset reaches large-scale production at the time when a significant increase in its production volume is observed. In the applicant’s view, the Commission ought rather to have taken account of the point at which a significant and lasting reduction of that chipset’s AVC is observed, since it is only once the AVC has stabilised that that chipset reaches its full production capacity.

719 Under its approach based on a significant reduction in AVC, the applicant submits that the MDM8200 chipset reached large-scale production only in the second quarter of 2010, whereas the MDM8200A and MDM6200 chipsets reached large-scale production only in the first quarter of 2011.

720 The Commission disputes the applicant’s arguments.

721 As a preliminary point, it should be noted that the parties do not dispute that a chipset’s AVC is, in general, very high at the beginning of its commercial life cycle, that it then falls significantly as production increases and that it stabilises during the last part of its life cycle. However, the parties’ views differ as to the point at which in the commercial life cycle of a chip the Commission can legitimately undertake its price-cost test for the purpose of ascertaining whether predatory prices have been charged.

722 In that regard, it must be stated that the applicant does no more than propose another approach, but does not explain how the approach taken by the Commission in the contested decision is incorrect, or why the starting point of the price-cost test ought to be the point at which a significant and sustained reduction of a chipset’s AVC is observed. In addition, the fact that the approach adopted by the Commission is not more favourable to the applicant than the approach which the applicant proposes, or that the two approaches are not similar, does not mean that the Commission’s findings are incorrect (see, to that effect, judgment of 29 March 2012, Telefónica and Telefónica de España v Commission, T‑336/07, EU:T:2012:172, paragraph 225), which is sufficient to reject the first part of the eighth plea.

723 It follows from the foregoing that the first part of the eighth plea must be rejected.

 The other parts: the most fundamental errors made in the price-cost test were not corrected in the contested decision

724 In support of the second part of the plea, the applicant claims that, if the most basic errors affecting it were corrected, the price-cost test would only lead to a finding of prices below LRAIC for the MDM8200 chipset over three quarters as regards sales to Huawei and over a single quarter as regards sales to ZTE. Referring to the fourth, sixth and seventh pleas respectively, it explains that those most basic errors concern the ‘theory of predation’, the calculation of prices and the calculation of costs.

725 In support of the third part of the eighth plea, the applicant submits that the Commission erred, and provided insufficient reasons, when it rejected the corrections which it proposed to make to those most basic errors affecting the price-cost test. In particular, referring (i) to the ninth plea, it criticises the Commission for having failed to conduct an ‘as-efficient’ competitor analysis, which, in its view, biased its price-cost test, and (ii) to the sixth and seventh pleas, it disputes the assertions made in recital 994 et seq. of the contested decision.

726 The Commission disputes the applicant’s arguments.

727 It must be stated that the applicant puts forward no new evidence in support of the second and third parts of the eighth plea, and simply refers, moreover, relatively vaguely, to certain other pleas, in particular as regards the existence of errors affecting the ‘theory of predation’ in the contested decision, the Commission’s calculation of prices and costs and the lack of an ‘as-efficient’ competitor analysis. Those elements have been rejected in the context of the examination of the fourth plea as regards the alleged errors affecting the ‘theory of predation’ in that decision, in the context of the examination of the sixth and seventh pleas as regards the alleged errors affecting the Commission’s calculation of prices and costs, and in the context of the first part of the ninth plea as regards the alleged lack of an ‘as-efficient’ competitor analysis, which is sufficient to reject the second and third parts of the eighth plea.

728 The second and third parts of the eighth plea must therefore be rejected. Therefore, since the first part of that plea has also been rejected (see paragraph 723 above), the eighth plea must be rejected in its entirety.

 The twelfth plea: the contested decision is not adequately reasoned

729 The applicant alleges that Article 41 of the Charter and Article 296 TFEU have been infringed, on the ground that the contested decision’s statement of reasons is inadequate.

730 In particular, the applicant submits, in the first place, that the contested decision contains contradictions, incomplete assessments and vague statements, referring in that regard to Annex A.12 to the application, in which it drew up a list of those inconsistencies.

731 In the second place, the applicant claims that, in the contested decision, the Commission makes unsubstantiated assertions and uncritically repeats assertions made by the complainant.

732 In the third place, the applicant relies on instances in which the Commission failed to address, let alone seriously examine, certain arguments raised during the administrative procedure, namely (i) the argument relating to the unsuitability of the questions in the request for information to define the market concerned, (ii) the argument relating to the development costs of the chipsets in question not being incremental to Huawei and ZTE, (iii) the argument relating to the fact that the documents used by the Commission to demonstrate the applicant’s intention to eliminate contain no reference whatsoever to the MDM8200 chipset, (iv) the argument relating to indirect predation and (v) the argument that the applicant merely aligned its prices with those of Icera.

733 In the fourth place, the applicant asserts that the Commission did not mention the methodology for calculating a cost benchmark which it had proposed during the administrative procedure and, without explanation, failed to take that methodology into account in the contested decision.

734 The Commission disputes the applicant’s arguments.

735 It should be stated that Article 41(1)(c) of the Charter imposes on the administration an obligation to give reasons for its decisions. Similarly, Article 296 TFEU provides that acts of the European Union must state the reasons on which they are based.

736 According to settled case-law, the statement of reasons required under Article 296 TFEU for measures adopted by EU institutions must be appropriate to the measure at issue and must disclose clearly and unequivocally the reasoning followed by the institution which adopted that measure in such a way as to enable the persons concerned to ascertain the reasons for it and to enable the competent court to review its legality. The requirements to be satisfied by the statement of reasons depend on all the circumstances of each case, in particular the content of the measure in question, the nature of the reasons given and the interest which the addressees of the measure, or other parties to whom it is of direct and individual concern, may have in obtaining explanations. It is not necessary for the reasoning to go into all the relevant facts and points of law, since the question whether the statement of reasons meets the requirements of Article 296 TFEU must be assessed with regard not only to its wording but also to its context and to all the legal rules governing the matter in question (see judgment of 10 March 2016, HeidelbergCement v Commission, C‑247/14 P, EU:C:2016:149, paragraph 16 and the case-law cited).

737 It is also important to recall that the existence of a statement of reasons is distinct from the question whether those reasons given are correct. It follows that, in a plea alleging failure to state reasons or that the reasons stated are insufficient, objections and arguments which aim to challenge the merits of the contested decision are ineffective (see, to that effect, judgment of 12 July 2018, Nexans France and Nexans v Commission, T‑449/14, EU:T:2018:456, paragraph 164).

738 As regards the first complaint, alleging contradictions, incomplete assessments and vague statements in the contested decision, it must be stated that, contrary to the applicant’s allegations, the contested decision is not vitiated by any of the contradictions, incomplete assessments or vague statements listed in Annex A.12 to the application, which is sufficient to reject the present complaint.

739 As regards the second complaint, alleging unsubstantiated assertions and uncritical repetition of assertions of the complainant in the contested decision, it must be stated that that complaint seeks to challenge the substance of the contested decision and is therefore ineffective in the context of the present plea.

740 As regards the third complaint, alleging that the Commission failed to address in the contested decision certain arguments raised during the administrative procedure, it must be observed that the applicant raised, in the context of the second plea (as regards the questions in the requests for information to define the market concerned), the fourth plea (as regards indirect predation), the seventh plea (as regards development costs); the tenth plea (as regards intention to eliminate) and the eleventh plea (as regards alignment with Icera’s prices), each of those arguments which were specifically examined by the General Court in the examination of those pleas which led to their rejection, which is sufficient to reject the present complaint.

741 Furthermore, in so far as the applicant complains that the Commission failed to make a serious examination of the arguments in question, that criticism seeks to challenge the substance of the contested decision and is therefore ineffective in the context of the present plea.

742 As regards the fourth complaint, alleging that the Commission did not mention in the contested decision the methodology for calculating a cost benchmark which the Commission had proposed during the administrative procedure and that it failed to take that methodology into account, it must be stated that the Commission did not infringe its obligation to state reasons on the basis that, in the present case, it applied its own methodology for calculating the cost benchmark, without taking into account the methodology proposed by the applicant. It is also clear that, in the present case, the Commission gave a detailed, clear and unequivocal explanation in Section 12.6 of that decision of its methodology for calculating a cost benchmark, as well as the reasons why that methodology appeared to it to be the most appropriate. The present complaint must therefore be rejected.

743 Consequently, the twelfth plea must be rejected.

 The thirteenth plea: ‘manifest errors of assessment’, lack of foundation and failure to state reasons in the contested decision as regards the duration of the infringement

744 In the context of the thirteenth plea, the applicant criticises both the start date of the infringement and the end date of the infringement as determined by the Commission.

745 As regards the start date of the infringement, referring to the arguments put forward in support of the eighth plea, the applicant submits that the Commission was not entitled to set that date at 1 July 2009, given that the MDM8200 chipset reached large-scale production only during the second quarter of 2010, and that the MDM6200 and MDM8200A chipsets reached large-scale production only during the first quarter of 2011.

746 As regards the end date of the infringement, referring to the arguments put forward in support of the fourth and tenth pleas, the applicant maintains that the Commission was not entitled to set that date at 30 June 2011.

747 The Commission disputes the applicant’s arguments.

748 It must be observed that, in the context of the present plea, the applicant simply refers to arguments put forward in support of the fourth, eighth and tenth pleas. When questioned in that regard at the hearing, the applicant expressly confirmed that the present plea was not based on any specific argument which had not been put forward in support of another plea.

749 The arguments put forward in support of the fourth, eighth and tenth pleas to which the applicant refers in the context of the present plea have all been rejected, which is sufficient to reject the thirteenth plea.

 The fourteenth plea: the contested decision ‘manifestly errs’ in the imposition and calculation of the fine

750 The present plea comprises seven parts. The first alleges that there was no intention or negligence in the commission of the infringement. The second alleges errors of assessment and of law, breach of the principle of legal certainty and failure to state reasons regarding the determination of the value of sales. The third alleges ‘manifest errors of law and of assessment’ relating to the failure to take into account the effects of the infringement. The fourth alleges errors of assessment and of law and breach of the obligation to state reasons as regards the gravity of the infringement. The fifth alleges ‘manifest errors of law and of assessment’ and breach of the obligation to state reasons as regards the additional amount of the infringement. The sixth alleges errors of assessment and of law regarding the absence of mitigating circumstances pursuant to the 2006 Guidelines and breach of the obligation to state reasons with regard to intention or negligence. The seventh alleges breach of the principle of proportionality.

 Preliminary observations

751 In the contested decision, the Commission decided to impose a fine on the applicant, calculated on the basis of the principles set out in the 2006 Guidelines.

752 First of all, the Commission determined, on the basis of data provided by the applicant, the value of the applicant’s direct and indirect sales of UMTS chipsets, that is to say the goods concerned by the infringement, in the EEA. In that context, it took as a basis the actual value of affected sales, adding the value of the applicant’s sales of UMTS chipsets during the relevant period, namely the second quarter of 2009, the whole of 2010, and the first quarter of 2011. In order to arrive at the basic amount of the fine, it applied a gravity factor of [confidential], setting out in the contested decision the factors which it had taken into account to arrive at such a percentage.

753 Next, the Commission added to the basic amount of the fine an additional amount for deterrence set at [confidential] of the average value of the applicant’s sales of UMTS chipsets in the EEA during the relevant period, that is to say, [confidential] of half of the actual value of sales affected, as previously determined. It also set out, in recitals 1274 to 1278 of the contested decision, the factors taken into account in order to arrive at that percentage.

754 Lastly, the Commission concluded that, in the absence of any aggravating or mitigating circumstances, the amount of the fine was not to be adjusted.

755 In the light of those considerations, the Commission set the amount of the fine imposed on the applicant at EUR 242 042 000, stating that, in the light of the applicant’s turnover of EUR 19 105 million in the financial year preceding the adoption of the contested decision, that amount was below the 10% ceiling laid down in Article 23(2) of Regulation No 1/2003.

756 In the present case, the Court considers it appropriate to begin by examining the second part of the present plea.

 The second part: errors of assessment and of law, breach of the principle of legal certainty and failure to state reasons regarding the determination of the value of sales

757 In support of the second part, the applicant puts forward three arguments. In the first place, it submits that the Commission, when assessing the value of its indirect sales of UMTS chipsets in the EEA, departed from point 15 of the 2006 Guidelines by using data relating to the calendar year, rather than data relating to the last tax year.

758 In the second place, the applicant claims that the Commission departed from points 13 and 24 of the 2006 Guidelines by using sales made throughout the whole of the relevant period, and not sales made during the last full business year, without the slightest explanation.

759 In the third place, the applicant complains that it did not have the opportunity to provide its comments to the Commission regarding the methodology for estimating the value of its sales intended to be applied by the Commission.

760 The Commission disputes the applicant’s arguments. It counters, in the first place, that, in order to determine the value of the applicant’s sales, it had to use the data for the calendar year provided by the applicant in 2015, that is to say, in tempore non suspecto, at a time when calculation of the fine was not being dealt with, rather than the data relating to the last tax year, which was provided by the applicant in 2019 and subsequently corrected a number of times because of multiple errors, and was therefore considered unreliable. It adds that it was unable to estimate the sales made during the last tax year on the basis of the data provided by calendar year, since the applicant never sent it a breakdown of those sales by quarter, which would have enabled it to do so.

761 In the second place, as regards the use of the value of sales relating to the whole of the relevant period, the Commission explained, in the rejoinder, that it was unable to base its calculations on the last full year of the infringement, because the market reconstruction data submitted by the applicant used figures aggregated for the calendar year. It adds in that regard that, since the first request for information of 5 February 2019, it had been asking the applicant to send to it the value of sales of UMTS chipsets for each full financial year, from 2009 to 2011, that is to say, for the whole of that period and that, consequently, the applicant ought to have understood that it intended to use the whole of that period and not only the last financial year.

762 In the third place, the Commission submits that the applicant had the opportunity to comment on the methodology to be applied by the Commission, since it informed the applicant thereof on 13 June 2019, with the result that the applicant was able to provide the Commission with its comments.

763 As regards the applicant’s first argument, relating to the Commission’s use of data relating to the calendar year rather than data relating to the last tax year, it is apparent from point 13 of the 2006 Guidelines that, in determining the basic amount of the fine to be imposed, the Commission will take the value of the undertaking’s sales of goods or services to which the infringement directly or indirectly relates in the relevant geographic area within the EEA, and that the Commission will normally take the sales made by the undertaking during the last full business year of its participation in the infringement.

764 Point 15 of the 2006 Guidelines states that, in determining the value of sales by an undertaking, the Commission will take that undertaking’s best available figures, and point 16 of those guidelines provides that, where the figures made available by an undertaking are incomplete or not reliable, the Commission may determine the value of its sales on the basis of the partial figures it has obtained or any other information which it regards as relevant and appropriate.

765 It follows that the Commission is not required to use the sales made during the last tax year, since to do so is simply a possibility for it; the use of the word ‘normally’ in point 13 of the 2006 Guidelines confirms that it is a possibility, not an obligation. Furthermore, if there are doubts regarding the reliability of the data provided by the undertaking concerned, point 16 of those guidelines allows the Commission to use any other information which it considers relevant and appropriate in order to determine the value of sales of the undertaking concerned.

766 In the present case, it is common ground that, in order to calculate the basic amount of the fine, the Commission used the applicant’s sales data per calendar year, rather than the sales data for each tax year, and that it provided a detailed explanation in recitals 1246 to 1266 of the contested decision of the reasons which had caused it to do so, namely, in particular, a problem with the reliability of the data provided by the applicant relating to its sales by tax year.

767 It should also be observed that, when the Commission informed the applicant, on 13 June 2019, that it intended to use the sales data for each calendar year, the applicant could very well have provided the Commission with sales data for each calendar year, broken down by quarter, which would have allowed the Commission to calculate, on the basis of that information, the value of sales made by the applicant during the 2010 tax year. The applicant failed to do so and therefore it cannot validly complain that the Commission failed to use data which were not available to it.

768 Therefore, the Commission did not infringe its obligation to state reasons and did not err when it decided to use the data provided by the applicant relating to the value of its sales presented by calendar year and not by tax year.

769 As regards the applicant’s second argument, relating to the Commission’s use of sales made throughout the duration of the infringement, rather than sales made during the last tax year, it is apparent from point 24 of the 2006 Guidelines that, in order to take fully into account the duration of the participation of each undertaking in the infringement, the amount determined on the basis of the value of sales will be multiplied by the number of years of participation in the infringement.

770 In the present case, it is not disputed that the Commission calculated the basic amount of the fine by adding together the value of sales made by the applicant during the relevant period, that is to say during the second half of 2009, in 2010, and in the first half of 2011, instead of multiplying the value of sales made during the last calendar year by the number of years of participation in the infringement, namely two years.

771 However, unlike point 13 of the guidelines, which allows the Commission to use a value of sales other than that relating to the last tax year, point 24 of those guidelines does not provide for a possible departure therefrom.

772 While it is true that point 37 of the 2006 Guidelines allows the Commission, in general terms, to depart from the methodology laid down by those guidelines, the fact remains that, in such a case, the Commission is required to state in particular the reasons which prompted it to depart from that methodology.

773 It is apparent from the case-law, first, that when the Commission decides to depart from the general methodology set out in the 2006 Guidelines, by which it limited the discretion it may itself exercise in setting the amount of fines, and relies on point 37 of those guidelines, that requirement relating to the duty to state reasons must be complied with all the more rigorously and, second, that the guidelines lay down a rule of conduct indicating the approach to be adopted from which the Commission cannot depart, in an individual case, without giving reasons which are compatible with, inter alia, the principle of equal treatment. Those reasons must be all the more specific because point 37 of the guidelines simply makes a vague reference to ‘the particularities of a given case’ and thus leaves the Commission a broad discretion where it decides to make an exceptional adjustment of the basic amount of the fines to be imposed on the undertakings concerned. In such a case, the Commission’s respect for the rights guaranteed by the EU legal order in administrative procedures, including the obligation to state reasons, is of even more fundamental importance (see judgment of 13 December 2016, Printeos and Others v Commission, T‑95/15, EU:T:2016:722, paragraph 48 and the case-law cited).

774 However, while the Commission does explain why it used the value of sales relating to the calendar year 2010 rather than those relating to the last tax year, it fails to justify why it used the value of sales relating to the total duration of the infringement (rather than sales made during the last year of the infringement, subsequently multiplied by two, given the two-year infringement period), since it simply submits that it could not have based its calculations on the last full year of the infringement because only aggregated figures for the calendar year were available to it. The Commission could therefore very well have used the value of sales made during the last calendar year of the infringement, which was available to it and which it regarded as reliable, and multiplied that value by the duration of the infringement in years, that is to say, by two.

775 Furthermore, it is clear from the case-law that the statement of reasons must, in principle, be notified to the person concerned at the same time as the decision adversely affecting him or her. The absence of reasoning cannot be legitimised by the fact that the person concerned becomes aware of the reasons for the decision during the procedure before the Courts of the European Union (judgments of 29 September 2011, Elf Aquitaine v Commission, C‑521/09 P, EU:C:2011:620, paragraph 149; of 19 July 2012, Alliance One International and Standard Commercial Tobacco v Commission and Commission v Alliance One International and Others, C‑628/10 P and C‑14/11 P, EU:C:2012:479, paragraph 74; and of 13 December 2016, Printeos and Others v Commission, T‑95/15, EU:T:2016:722, paragraph 46). It follows that the explanation provided by the Commission, in the context of the present proceedings, and in particular at the stage of the rejoinder, as to the reasons which resulted in it departing from point 24 of the 2006 Guidelines, even if it were convincing, could, in any event, not make up for the lack of reasoning in that regard in the contested decision.

776 It follows that, in the contested decision, the Commission departed, without justification, from the method laid down by the 2006 Guidelines by, in the present proceedings and at the stage of the rejoinder, taking an insufficient justification as a basis.

777 As regards the applicant’s third argument, concerning the lack of opportunity to make known its views on the methodology which the Commission intended to apply in order to set the basic amount of the fine, it is common ground that the applicant was informed on 13 June 2019 of that methodology, and in particular of the fact that the Commission intended to use sales data for each calendar year. The applicant therefore did have the opportunity to submit any comments it might have in that regard.

778 However, contrary to the Commission’s assertions, the applicant could not reasonably expect the Commission to calculate the basic amount of the fine on the basis of the value of sales made during the whole of the relevant period, simply because the Commission had requested the applicant to provide it with the value of sales of UMTS chipsets for each full year, from 2009 to 2011, that is to say, for the whole of that period.

779 It follows from the foregoing that the present part and, consequently, the fourteenth plea, must be upheld in part. Therefore, Article 2 of the contested decision must be annulled, there being no need to examine the other parts of the present plea, all of which seek to have that provision annulled.

 The fifteenth plea: ‘manifest errors of law, fact and assessment’ and failure to state reasons in the contested decision, in establishing the Commission’s jurisdiction and the effect on trade

780 The present plea comprises two parts. The first alleges that the Commission lacked jurisdiction to apply Article 102 TFEU. The second concerns the existence of effects on trade between Member States caused by the conduct at issue.

 The first part: the Commission lacked jurisdiction to apply Article 102 TFEU

781 The applicant submits that the Commission erred in finding that it had jurisdiction to apply Article 102 TFEU in the contested decision, for the reasons set out below.

782 In the first place, the conduct at issue was not implemented in the EEA because there were no direct sales of the chipsets concerned in or into the EEA.

783 In the second place, for the reasons set out in support of the ninth plea, the conduct at issue was not liable to have anticompetitive foreclosure effects within the EEA. And even if that conduct were to have been so liable, those effects would not in any event have been substantial, immediate and foreseeable.

784 The Commission disputes the applicant’s arguments.

785 It should be recalled that, as regards conduct adopted outside the territory of the EEA, the Commission’s jurisdiction under public international law to find and punish an infringement of Article 102 TFEU may be established on the basis of either the implementation test or the qualified effects test (see, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraphs 40 to 47; see also, to that effect and by analogy, judgment of 12 July 2018, Brugg Kabel and Kabelwerke Brugg v Commission, T‑441/14, EU:T:2018:453, paragraphs 95 to 97).

786 In that regard, it must also be pointed out that the implementation test and the qualified effects test are alternatives and are not cumulative (see, to that effect, judgments of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraphs 62 to 64, and of 12 July 2018, Brugg Kabel and Kabelwerke Brugg v Commission, T‑441/14, EU:T:2018:453, paragraph 98).

787 In the present case, it must be stated that, in Section 13 of the contested decision, in order to establish its jurisdiction under public international law to find and punish an infringement of Article 102 TFEU, the Commission relied both on the implementation test (recital 1203 of the contested decision) and on the qualified effects test (recitals 1204 to 1210 of the contested decision).

788 The Commission explained in recital 1203 of the contested decision that the implementation test was satisfied, given that the applicant knew or could not have been unaware that the devices assembled by Huawei and ZTE incorporating one of Qualcomm’s baseband chipsets would also be sold in the EEA, as evidenced by a number of items of evidence contemporaneous with the facts in question, in particular a reply by the applicant made during the administrative procedure, as well as by a number of internal Qualcomm documents.

789 The applicant in no way disputes the fact that Huawei and ZTE, to which it sold chips at predatory prices, incorporated them into devices sold in the EEA, or the fact that it knew or could not have been unaware that that was the reason why the Commission concluded, in the contested decision, that the implementation test was satisfied.

790 In addition, the applicant cannot reasonably rely on there being no direct sales in the EEA, since the Commission does not rely on that factor in the contested decision for the purposes of demonstrating that the implementation test is satisfied.

791 Therefore, the Commission did not err in concluding that the implementation test was satisfied in the present case.

792 Furthermore, in the light of the fact that the implementation test and the qualified effects test are alternatives, as is apparent from the case-law cited in paragraph 786 above, since the applicant has not succeeded in refuting the Commission’s conclusion that the implementation test was satisfied, it is not necessary to examine whether the Commission correctly concluded that the other test, namely the qualified effects test, was also satisfied since the implementation test, in itself, provides a basis for the Commission’s jurisdiction to apply Article 102 TFEU.

793 It follows from the foregoing that the first part of the fifteenth plea must be rejected.

 The second part: the existence of effects on trade between Member States caused by the conduct at issue

794 The applicant submits, referring to its arguments in support of the ninth plea, that the Commission erred in concluding in recital 1216 of the contested decision that the conduct at issue had had an appreciable effect on trade between Member States.

795 The Commission disputes the applicant’s arguments.

796 It must be borne in mind that, according to settled case-law, in order to determine whether trade between Member States is capable of being affected by the abuse of a dominant position, account must be taken of the consequences for the effective competitive structure in the internal market (see judgment of 4 May 1988, Bodson, 30/87, EU:C:1988:225, paragraph 24 and the case-law cited).

797 Consequently, practices whose objective is to eliminate from the market their main established competitor in the internal market are inherently capable of affecting the structure of competition in that market and thereby of affecting trade between Member States within the meaning of Article 102 TFEU (judgment of 8 October 1996, Compagnie maritime belge transports and Others v Commission, T‑24/93 to T‑26/93 and T‑28/93, EU:T:1996:139, paragraph 203).

798 It is also apparent from the case-law that an anticompetitive practice relating to an intermediate product is capable of affecting trade within the internal market, even if there is no trade in that intermediate product between the Member States, where the product constitutes the raw material for another product marketed elsewhere in the European Union (see, to that effect, judgment of 30 January 1985, Clair, 123/83, EU:C:1985:33, paragraph 29).

799 In the present case, it is apparent from the examination of the tenth plea that the Commission has demonstrated to the requisite legal standard that the conduct at issue had the objective of eliminating Icera, the applicant’s main competitor, established at the material time within the European Union, which the intervener expressly confirmed at the hearing. In addition, that conduct concerned UMTS chipsets, that is to say, intermediate products, intended for MBB devices sold, in particular, in the European Union, which the applicant does not legitimately dispute.

800 Consequently, the Commission did not infringe its obligation to state reasons and did not err when it concluded, in the contested decision, that the conduct at issue was capable of affecting trade between Member States. The second part of the fifteenth plea must therefore be rejected.

801 It follows from all of the foregoing that the fifteenth plea must be rejected.

 The request to annul or reduce the amount of the fine

802 Unlimited jurisdiction empowers the General Court, in addition to carrying out a mere review of the lawfulness of the penalty, which merely permits dismissal of the action for annulment or annulment of the contested measure, to substitute its own appraisal for the Commission’s and, therefore, to vary the contested measure, even without annulling it, by taking account of all the factual circumstances, so as, in particular, to amend the amount of the fine, to reduce that amount as well as to increase it (see, to that effect, judgments of 8 February 2007, Groupe Danone v Commission, C‑3/06 P, EU:C:2007:88, paragraphs 61 and 62, and of 3 September 2009, Prym and Prym Consumer v Commission, C‑534/07 P, EU:C:2009:505, paragraph 86).

803 In the exercise of its unlimited jurisdiction, it is for the Court to determine the amount of the fine, taking into account all of the circumstances of the present case. That exercise involves, in accordance with Article 23(3) of Regulation No 1/2003, taking into consideration the seriousness and duration of the infringement committed, in compliance with the principles of, inter alia, proportionality and the individualisation of penalties (see judgment of 21 January 2016, Galp Energía España and Others v Commission, C‑603/13 P, EU:C:2016:38, paragraph 90 and the case-law cited).

804 In the context of its obligation to state reasons, the Court is also required to set out in detail the factors which it took into account when setting the amount of the fine (see, to that effect, judgment of 14 September 2016, Trafilerie Meridionali v Commission, C‑519/15 P, EU:C:2016:682, paragraph 52).

805 In the present case, in order to determine the amount of the fine intended to penalise the applicant for having implemented selective predatory pricing for two years in order to eliminate its main competitor, the Court considers it appropriate, in order to take account of its examination of the second part of the fourteenth plea, to proceed in the manner set out below.

 Preliminary observation

806 While it is not bound by the Commission’s guidelines on setting the amount of fines (see, to that effect, judgment of 14 September 2016, Trafilerie Meridionali v Commission, C‑519/15 P, EU:C:2016:682, paragraphs 52 to 55), the Court considers it appropriate in the present case to apply the methodology laid down by the 2006 Guidelines, which the applicant does not dispute.

 Determination of the basic amount of the fine

807 In the light of the fact that point 13 of the 2006 Guidelines does not lay down any obligation to use the value of sales made during the last business year, as is apparent from paragraphs 763 to 765 above, account is to be taken in the present case of the value of the applicant’s sales in the EEA for UMTS chipsets during the last calendar year of the infringement, that is to say, the calendar year 2010. That choice is justified by the fact that, in the present case, the data submitted by the applicant for each business year are not reliable, for the reasons set out in recitals 1246 to 1266 of the contested decision.

808 In that regard, it is apparent from Table 75 of the contested decision, which the parties do not dispute, that, during the calendar year 2010, the value of UMTS chipsets sales made in the EEA by the applicant was [confidential].

 Degree of gravity of the infringement

809 As regards the proportion of the value of sales to be taken into consideration, all the relevant circumstances of the case must be taken into consideration and, in particular, the economic importance of the worldwide market for UMTS chipsets, the nature of the infringement committed by the applicant, the fact that the only competitor capable of challenging the applicant in the leading edge segment of that market was the subject of the practice at issue, the fact that that conduct might also have deterred other potential entrants, and the geographic scale of the infringement, which was worldwide and included the whole of the EEA.

810 Consequently, it is appropriate to apply a gravity factor of 11%, which gives an amount of [confidential].

 Number of years of infringement

811 It is apparent from point 24 of the 2006 Guidelines that, in order to take fully into account the duration of the participation of each undertaking in the infringement, the amount determined on the basis of the value of sales will be multiplied by the number of years of participation in the infringement.

812 In the present case, since the infringement lasted two years (recital 1295 of the contested decision), the amount determined on the basis of the value of sales must be multiplied by two, which gives an amount of [confidential].

 Additional amount

813 It is apparent from point 25 of the 2006 Guidelines that, in order to deter undertakings of the same size and having the same resources from engaging in anticompetitive practices similar to those made the subject of penalties in the present case, a sum of between 15% and 25% of the value of sales may be included in the basic amount of the fine.

814 In the present case, the Court considers that the objective of deterrence may be guaranteed by applying a rate lower than the purely indicative rate set out in point 25 of the 2006 Guidelines.

815 For those reasons, it is appropriate to include in the basic amount of the fine a sum corresponding to 11% of the value of the applicant’s sales of UMTS chipsets in the EEA during the calendar year 2010, as found in Table 75 of the contested decision, which is not disputed by the parties, which gives an amount of [confidential].

 Mitigating or aggravating circumstances

816 In the present case, none of the factors to which the applicant makes general references can lead to a reduction in the amount of the fine by virtue of point 29 of those guidelines and, in the light of all the relevant circumstances, no adjustment to the basic amount of the fine is appropriate.

817 The Court also considers, as did the Commission in the contested decision, that it is not appropriate in the present case to increase the amount of the fine on account of aggravating circumstances.

818 It follows that the amount of the fine is not to be adjusted on account of mitigating or aggravating circumstances.

 Conclusion as regards the amount of the fine

819 It follows from the foregoing that the amount of the fine imposed on the applicant is set at EUR 238 732 659.33.

 Costs

820 Under Article 134(1) of the Rules of Procedure, the unsuccessful party is to be ordered to pay the costs if they have been applied for in the successful party’s pleadings. Furthermore, under Article 134(3) of the Rules of Procedure, where each party succeeds on some and fails on other heads, the parties are to bear their own costs. However, if it appears justified in the circumstances of the case, the Court may order that one party, in addition to bearing its own costs, pay a proportion of the costs of the other party.

821 In the present case, both the applicant and the Commission have been partially unsuccessful.

822 In those circumstances, the applicant must be ordered to bear nine tenths of its own costs and to pay nine tenths of the Commission’s costs and all of the intervener’s costs. The Commission is to bear one tenth of its own costs and to pay one tenth of the applicant’s costs.

On those grounds,

THE GENERAL COURT (First Chamber, Extended Composition)

hereby:

1. Annuls Article 2 of Commission Decision C(2019) 5361 final of 18 July 2019 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.39711 – Qualcomm (predation));

2. Sets the amount of the fine imposed on Qualcomm Inc. in Article 2 of that decision, for the infringement it committed, as results from Article 1 of that decision, at EUR 238 732 659.33;

3. Dismisses the action as to the remainder;

4. Orders Qualcomm to bear nine tenths of its own costs and to pay nine tenths of the European Commission’s costs and to pay all the costs of Nvidia Corp;

5. Orders the Commission to bear one tenth of its own costs and to pay one tenth of Qualcomm’s costs.