OPINION OF ADVOCATE GENERAL
WAHL
delivered on 20 October 2016 (1)
Case C‑413/14 P
Intel Corporation Inc.
v
European Commission
(Appeal — Article 102 TFEU — Abuse of dominant position — Loyalty rebates — Classification as abuse — Applicable legal test — Single and continuous infringement — Rights of the defence — Article 19 of Regulation (EC) No 1/2003 — Interview relating to the subject matter of an investigation — Jurisdiction of the Commission — Implementation — Effects)
Table of contents
I – Legal framework
II – Background to the proceedings
III – Procedure before the General Court and the judgment under appeal
IV – Procedure before the Court and forms of order sought
V – Assessment of the grounds of appeal
A – Introductory remarks
B – The first ground of appeal: the legal test to be applied to so-called ‘exclusivity rebates’
1. The main arguments of the parties
2. Analysis
(a) The principal assessment of the appellant’s rebates and payments carried out by the General Court
i) The basic tenets of the Court’s case-law regarding rebates
ii) The circumstances of the case as a means to determine whether the impugned conduct has a likely effect on competition
iii) There are only two categories of rebates according to the case-law
– An assumption of unlawfulness by virtue of form cannot be rebutted
– Loyalty rebates are not always harmful
– The effects of loyalty rebates are context-dependent
– Related practices require consideration of all the circumstances
iv) Intermediate conclusion
(b) The alternative assessment of capability carried out by the General Court
i) Capability and/or likelihood
ii) The factors considered by the General Court to support a finding of abuse
iii) Other circumstances
– Market coverage
– Duration
– Market performance of the competitor and declining prices
– The AEC test
(c) Conclusion
C – The second ground of appeal: market coverage in determining whether an undertaking has abused its dominant position
1. The main arguments of the parties
2. Analysis
D – The third ground of appeal: classification of certain rebates as ‘exclusivity rebates’
1. The main arguments of the parties
2. Analysis
E – The fourth ground of appeal: rights of the defence
1. The main arguments of the parties
2. Analysis
(a) The meeting in question was an interview within the meaning of Article 19 of Regulation No 1/2003
(b) The note to the file did not cure the procedural error
(c) The consequence of the failure to record the meeting in question
F – The fifth ground of appeal: jurisdiction
1. The main arguments of the parties
2. Analysis
(a) General remarks: implementation and/or effects?
(b) Assessment of the application of the relevant jurisdictional criteria by the General Court
i) Implementation
ii) ‘Qualified’ effects
G – The sixth ground of appeal: the amount of the fine
1. The main arguments of the parties
2. Analysis
VI – Consequences of the assessment
VII – Conclusion
1. By this appeal, Intel Corporation (‘Intel’ or ‘the appellant’) asks the Court to set aside the judgment of 12 June 2014 in Intel v Commission, (2) by which the General Court dismissed its action for annulment of Commission Decision C(2009) 3726 final of 13 May 2009 relating to a proceeding under Article 82 [EC] (now Article 102 TFEU) and Article 54 of the EEA Agreement (Case COMP/C‑3/37.990 — Intel) (‘the decision at issue’). (3)
2. The case raises a number of important questions of principle. Those questions include the application of the concept of ‘single and continuous infringement’ in the context of what is now Article 102 TFEU, the discretion the Commission ought to have in recording the interviews it conducts during its investigations and the scope of the Commission’s jurisdiction to conduct investigations of infringements originating abroad.
3. In addition, the case offers the Court an opportunity to refine its case-law relating to the abuse of a dominant position under Article 102 TFEU. More specifically, the question arises as to whether, in the light of the line of authority devolving from Hoffmann-La Roche, (4) it is warranted to draw a distinction between different types of rebates. Taking account of that case-law, the Court must determine the correct legal test to be applied in relation to a particular category of rebates, which the General Court termed as ‘exclusivity rebates’ in the judgment under appeal.
4. In particular, the Court is called upon to decide whether the General Court was right to consider that rebates of the kind at issue in the present case are inherently anticompetitive. The inherently anticompetitive nature of those rebates would render it unnecessary to consider all the circumstances of the case in order to determine whether the conduct in question is, in actual fact, capable of restricting competition in a particular market.
I – Legal framework
5. Recital 25 of Regulation (EC) No 1/2003 (5) explains that the Commission should in particular be empowered to interview any persons who may be in possession of useful information and to record the statements made.
6. Article 19 of the regulation concerns the power of the Commission to take statements. Paragraph 1 thereof states:
‘In order to carry out the duties assigned to it by this Regulation, the Commission may 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.’
7. Article 27(2) of that regulation provides:
‘The rights of defence of the parties concerned shall be fully respected in the proceedings. They shall be entitled to have access to the Commission's file, subject to the legitimate interest of undertakings in the protection of their business secrets. The right of access to the file shall not extend to confidential information and internal documents of the Commission or the competition authorities of the Member States...’
8. Recital 3 of Regulation (EC) No 773/2004 (6) explains that, when the Commission conducts interviews, the persons interviewed ought to be informed of the purpose of the interview and of any record which may be made.
9. Article 3 of that regulation concerns the Commission’s power to take statements. It provides:
‘1. Where the Commission interviews a person with his consent in accordance with Article 19 of Regulation [No 1/2003], it shall, at the beginning of the interview, state the legal basis and the purpose of the interview, and recall its voluntary nature. It shall also inform the person interviewed of its intention to make a record of the interview.
2. The interview may be conducted by any means including by telephone or electronic means.
3. The Commission may record the statements made by the persons interviewed in any form. A copy of any recording shall be made available to the person interviewed for approval. Where necessary, the Commission shall set a time-limit within which the person interviewed may communicate to it any correction to be made to the statement.’
II – Background to the proceedings
10. The background to the dispute, as set out in the judgment under appeal, may be summarised as follows.
11. Intel is a US-based company that designs, develops, manufactures and markets central processing units (‘CPUs’), ‘chipsets’, and other semiconductor components, as well as platform solutions for data processing and communications devices.
12. On 18 October 2000, Advanced Micro Devices (‘AMD’) submitted a formal complaint to the Commission pursuant to Regulation No 17, (7) which it supplemented with new facts and allegations, in the context of a supplementary complaint of 26 November 2003.
13. In May 2004, the Commission launched investigations focusing on allegations contained in AMD’s supplementary complaint.
14. On 17 July 2006, AMD lodged a complaint with the Bundeskartellamt (Federal Cartel Office, Germany), in which it claimed that Intel had engaged inter alia in exclusionary marketing arrangements with Media-Saturn-Holding GmbH (‘MSH’), a European retailer of microelectronic devices and the largest desktop computer distributor in Europe. The Federal Cartel Office exchanged information with the Commission on that subject.
15. On 23 August 2006, the Commission held a meeting with a senior executive, Mr D1, of Dell Inc., a customer of Intel. (8) The Commission did not place the indicative list of topics for the meeting in the case file and did not take minutes of it. A member of the Commission’s team responsible for the file drafted a note sometime after the meeting, which was described as internal by the Commission. On 19 December 2008, the Commission provided the appellant with a non-confidential version of that note.
16. On 26 July 2007, the Commission notified to the appellant a statement of objections (‘the Statement of Objections of 2007’) concerning its conduct vis-à-vis five major original equipment manufacturers (‘OEMs’), namely Dell, Hewlett-Packard Company (HP), Acer Inc., NEC Corp. and International Business Machines Corp. (IBM).
17. On 17 July 2008, the Commission sent the appellant a supplementary statement of objections concerning its conduct in respect of MSH and Lenovo Group Ltd (‘Lenovo’). It included new evidence on Intel’s conduct vis-à-vis some of the OEMs covered by the Statement of Objections of 2007, which had been obtained by the Commission after its publication.
18. After various procedural steps, on 13 May 2009, the Commission adopted the decision at issue in which it took the view that Intel had infringed Article 82 EC and Article 54 of the Agreement on the European Economic Area (EEA), from October 2002 until December 2007, by implementing a strategy aimed at foreclosing a competitor, AMD, from the market for x86 CPU microprocessors (‘x86 CPUs’).
19. That decision included the following considerations.
20. The products concerned are CPUs. x86 architecture is a standard designed by Intel for its CPUs. It can run both the Windows and Linux operating systems. Windows is primarily linked to the x86 instruction set. Prior to 2000, there were several manufacturers of x86 CPUs. However, most of these manufacturers have since exited the market. According to the decision at issue, Intel and AMD are essentially the only two companies still manufacturing x86 CPUs.
21. In addition, the Commission concluded in the decision at issue that the relevant product market was not wider than the market for x86 CPUs. Nevertheless, it left open the question whether there is a single x86 CPU market for all computers or whether it is necessary to draw a distinction between three separate x86 CPU markets, namely the market for desktop computers, the market for notebook computers and the market for servers. According to the decision at issue, in view of Intel’s market shares for each of those segments, the conclusion on dominance remains the same.
22. In the decision at issue, the geographical market was defined as worldwide.
23. Regarding dominance, in the 10-year period examined by the Commission (1997-2007), Intel held market shares in excess of or around 70%. Moreover, the Commission identified significant barriers to entry to and expansion within the x86 CPU market. Those barriers arise, in particular, from sunk investments in research and development, intellectual property and the production facilities that are necessary to produce x86 CPUs. On the basis of Intel’s market shares and the barriers to entry and expansion in the relevant market, the Commission concluded that Intel held a dominant position in that market at least in the period covered by the decision, that is, from October 2002 to December 2007.
24. In the decision at issue, the Commission identified two types of conduct by Intel vis-à-vis its trading partners, namely conditional rebates and so-called ‘naked restrictions’.
25. As regards the first type, Intel awarded four OEMs, namely Dell, Lenovo, HP and NEC, rebates which were conditioned on these OEMs purchasing all or almost all of their x86 CPUs from Intel. Intel also granted payments to MSH, which were conditional upon MSH selling exclusively computers containing Intel’s x86 CPUs.
26. According to the decision at issue, the conditional rebates granted by Intel are described as fidelity rebates. Regarding Intel’s conditional payments to MSH, the Commission concluded that the economic mechanism of those payments is equivalent to that of the conditional rebates to OEMs.
27. The decision at issue also proceeds to analyse in economic terms the capability of the rebates and the payments to MSH to foreclose a hypothetical competitor as efficient as Intel (as-efficient-competitor test; ‘the AEC test’). (9)
28. In light of those considerations, the Commission concluded that Intel’s conditional rebates and payments induced the loyalty of the key OEMs and of MSH. The effects of these practices were complementary, significantly diminishing competitors’ ability to compete on the merits of their x86 CPUs. As a consequence, Intel’s anticompetitive conduct resulted in a reduction of consumer choice and in lower incentives to innovate.
29. As regards the second type of conduct identified in the decision at issue, namely ‘naked restrictions’, the Commission considered that Intel granted three OEMs, namely HP, Acer and Lenovo, payments which were conditional upon those OEMs postponing or cancelling the launch of AMD CPU-based products and/or putting restrictions on the distribution of those products. The Commission concluded that Intel’s conduct regarding naked restrictions harmed competition, because customers were deprived of a choice which they would otherwise have had. In the Commission’s view, that did not constitute normal competition on the merits.
30. The Commission concluded in the decision at issue that, in each instance, Intel’s conduct vis-à-vis the OEMs mentioned above and MSH constitutes an abuse under Article 102 TFEU, but that each individual abuse is also part of a single strategy aimed at foreclosing AMD, Intel’s only significant competitor, from the market for x86 CPUs. Those individual abuses are therefore part of a single infringement of Article 102 TFEU.
31. By applying the Guidelines on the method of setting fines imposed pursuant to Article 23(2)(a) of Regulation No 1/2003 (‘the 2006 Guidelines’), (10) the Commission imposed on the appellant a fine of EUR 1.06 billion.
32. The decision at issue provides:
‘Article 1
Intel … has committed a single and continuous infringement of Article [102 TFEU] and Article 54 of the EEA Agreement from October 2002 until December 2007 by implementing a strategy aimed at foreclosing competitors from the market of x86 CPUs which consisted of the following elements:
(a) Granting rebates to Dell between December 2002 and December 2005 at a level that was conditional on Dell obtaining all of its x86 CPU supplies from Intel;
(b) Granting rebates to HP between November 2002 and May 2005 at a level that was conditional on HP obtaining at least 95% of its corporate desktop x86 CPU supplies from Intel;
(c) Granting rebates to NEC between October 2002 and November 2005 at a level that was conditional on NEC obtaining at least 80% of its client PC x86 CPU supplies from Intel;
(d) Granting rebates to Lenovo between January 2007 and December 2007 at a level that was conditional on Lenovo obtaining all of its notebook x86 CPU supplies from Intel;
(e) Granting payments to [MSH] between October 2002 and December 2007 at a level that was conditional on [MSH] selling only computers incorporating Intel x86 CPUs;
(f) Granting payments to HP between November 2002 and May 2005 conditional on: (i) HP directing HP’s AMD-based x86 CPU business desktops to Small and Medium Business and Government, and Educational and Medical customers rather than to enterprise business customers; (ii) precluding HP’s channel partners from stocking HP’s AMD-based x86 CPU business desktops such that such desktops would only be available to customers by ordering them from HP (either directly or via HP channel partners acting as sales agent); and (iii) HP delaying the launch of its AMD-based x86 CPU business desktop in the [Europe, Middle East and Africa] region by six months;
(g) Granting payments to Acer between September 2003 and January 2004 conditional on Acer delaying an AMD-based x86 CPU notebook;
(h) Granting payments to Lenovo between June 2006 and December 2006 conditional on Lenovo delaying and finally cancelling its AMD-based x86 CPU notebooks.’
III – Procedure before the General Court and the judgment under appeal
33. By application lodged on 22 July 2009, the appellant requested the General Court to annul the decision at issue. The Association for Competitive Technology, Inc. (‘ACT’) intervened in support of Intel.
34. In the judgment under appeal, the General Court dismissed the action in its entirety.
IV – Procedure before the Court and forms of order sought
35. By its appeal, lodged with the Court on 26 August 2014, Intel claims that the Court should:
– set aside in whole or in part the judgment under appeal;
– annul in whole or in part the decision at issue;
– cancel or substantially reduce the fine imposed;
– in the alternative, refer the case back to the General Court for determination in accordance with the judgment of the Court of Justice;
– order the Commission to pay the costs of these proceedings and of the proceedings before the General Court.
36. ACT has submitted a response in support of the form of order sought by the appellant.
37. The Commission contends that the Court should dismiss the appeal and order the appellant to pay the costs.
38. Intel, ACT and the Commission presented oral arguments at the hearing held on 21 June 2016.
V – Assessment of the grounds of appeal
39. The appellant relies on six grounds in support of its appeal. The first ground of appeal alleges errors in law in the legal characterisation of rebates termed as ‘exclusivity rebates’ by the General Court. The second ground of appeal alleges an error in law in the finding of an infringement in 2006 and 2007 and in the assessment of the relevance of market coverage. The third ground concerns an error in law as regards the classification as ‘exclusivity rebates’ of certain rebate arrangements that merely covered a minority of a customer’s purchases. The fourth ground of appeal alleges a procedural error resulting from an incorrect interpretation of Article 19 of Regulation No 1/2003, read in conjunction with Article 3 of Regulation No 773/2004, concerning the absence of an obligation to record an interview. The fifth ground of appeal relies on an infringement of Article 102 TFEU and concerns the Commission’s jurisdiction to apply Article 102 TFEU to the appellant’s arrangements in China with Lenovo in 2006 and 2007. Finally, the sixth ground concerns the amount of the fine arising from an error in law concerning the retroactive application of the 2006 Guidelines on the setting of fines.
40. I shall deal with all those issues in turn. Before doing so, however, I consider it useful to make some preliminary observations as regards the structure and rationale of Article 102 TFEU. Those observations constitute the starting point for the ensuing assessment of the first three grounds of appeal.
A – Introductory remarks
41. From the outset, EU competition rules have aimed to put in place a system of undistorted competition, as part of the internal market established by the EU. (11) In that regard, it cannot be overemphasised that protection under EU competition rules is afforded to the competitive process as such, and not, for example, to competitors. (12) In the same vein, competitors that are forced to exit the market due to fierce competition, rather than anticompetitive behaviour, are not protected. Therefore, not every exit from the market is necessarily a sign of abusive conduct, but rather a sign of aggressive, yet healthy and permissible, competition. (13) This is because, given its economic character, competition law aims, in the final analysis, to enhance efficiency. The importance placed on efficiency is also in my view clearly reflected in the case-law of the EU Courts.
42. From that emphasis it naturally follows that dominance as such is not considered to be at variance with Article 102 TFEU. Rather, only behaviour which constitutes an expression of market power to the detriment of competition and, thus, to consumers is prohibited and accordingly sanctioned as an abuse of dominance.
43. A logical corollary to the objective of enhanced efficiency is that the anticompetitive effects of a particular practice assume crucial importance. Irrespective of whether we are dealing with an enforcement shortcut such as that offered by the concept of ‘restriction by object’ in the context of Article 101 TFEU, (14) or with single firm conduct falling within the scope of Article 102 TFEU, EU competition rules seek to capture behaviour that has anticompetitive effects. To date, the form of a particular practice has not been deemed important.
44. In the judgment under appeal, the General Court distinguished between three categories of rebates: volume-based rebates, ‘exclusivity rebates’ and rebates that are based on a mechanism that may have a fidelity-inducing effect. Unlike a rebate scheme based purely on the volume of purchases (category 1), which reflect gains in efficiency and economies of scale, exclusivity rebate schemes (category 2), according to the taxonomy employed by the General Court, are incompatible with the objective of undistorted competition within the internal market. Such rebates are conditional on the customer obtaining all or most of its requirements from the undertaking in a dominant position. (15)
45. In addition to the two previously mentioned categories of rebates, the judgment under appeal refers to a residual category of rebates which possesses a fidelity-building mechanism without being directly linked to exclusive or quasi-exclusive supply (category 3). That category includes rebates such as retroactive rebates. (16) The General Court considered that rebates belonging to category 3 should be distinguished from ‘exclusivity rebates’ on the ground that they are not directly conditional on exclusivity. For that reason, the General Court accepted that it was necessary to consider all the circumstances in order to determine whether such rebates are capable of restricting competition. (17)
46. Due to the conditional character of the rebates, the General Court classified the rebates and payments offered by the appellant as ‘exclusivity rebates’. Relying on the case-law devolving from Hoffmann-La Roche, the General Court considered that, in order to determine whether the undertaking in question has abused its dominant position, it was sufficient that the rebates were ‘exclusivity rebates’ belonging to category 2. Once that was established, it was no longer necessary to consider ‘all the circumstances’ in order to verify that the conduct was capable of restricting competition. Solely on the basis of the form of the conduct, such capability could be assumed. This was so, according to the General Court, because such rebates are designed, as a rule, to remove or restrict the purchaser’s freedom to choose his sources of supply and, in that sense, to prevent customers from obtaining their supplies from competing producers. (18)
47. In accordance with that methodology, the assumption that ‘exclusivity rebates’ offered by a dominant undertaking result always, and without exception, in anticompetitive foreclosure permeates the entire judgment under appeal. It is on the basis of that assumption that the relevance of context and, by way of extension, the need to consider the capability of the conduct to have anticompetitive effects was rejected by the General Court.
48. Bearing that in mind, the fate of the first, second and third grounds of appeal depends, ultimately, on whether the Court considers that assumption to be correct.
B – The first ground of appeal: the legal test to be applied to so-called ‘exclusivity rebates’
1. The main arguments of the parties
49. As its primary contention Intel, supported by ACT, claims that the General Court erred in its legal characterisation of what it termed as ‘exclusivity rebates’, that is, ‘fidelity rebates within the meaning of Hoffmann-La Roche’. (19) In its view, the General Court erred in concluding that, unlike other rebates and pricing practices, such rebates are inherently capable of restricting competition and thus are anticompetitive without any need to consider either the relevant circumstances of the rebates in question or the likelihood that the rebates might restrict competition. (20) In that context, the appellant claims that the General Court erred in upholding the finding of an abuse without considering the likelihood of anticompetitive harm. Moreover, Intel submits that, in any event, the General Court erred in its alternative finding, according to which the rebates in question in this case were capable of restricting competition. (21)
50. The Commission contends that the first ground of appeal should be dismissed. It maintains, in essence, that the present ground of appeal is premised on a fallacy: that ‘exclusivity rebates’ constitute mere pricing practices. The Commission argues that ‘exclusivity rebates’ are inherently different from other pricing practices. In its view, rebates conditional on exclusivity exhibit features that render it unnecessary to verify whether they are capable of restricting competition in a specific case. The Commission considers, in particular, that the case-law of the Court regarding rebates does not lend support to the appellant’s view that no distinction ought to be drawn between ‘exclusivity rebates’ and other rebates that have a fidelity-inducing effect or, for that matter, pricing practices.
51. Regarding the alternative assessment of all the circumstances, the Commission considers that the appellant has not put forward any argument that calls into question the alternative assessment of capability carried out by the General Court in the judgment under appeal.
2. Analysis
52. The gist of the approach to the first ground lies in determining the correct legal test to be applied to so-called ‘exclusivity rebates’. In other words, the question is whether the General Court was justified in finding that there is no need to consider ‘all the circumstances’ to verify that those rebates are capable of having an anticompetitive effect. Put simply: was the General Court correct in holding that, owing to their form, not even context can save ‘exclusivity rebates’ from condemnation.
53. At the outset, contrary to what the Commission suggested in its written pleadings, I see no reason why the Court should not examine the first ground of appeal in its entirety. Under that ground, the appellant is clearly seeking to challenge errors in law which in its view vitiated the classification of Intel’s rebates and payments as ‘exclusivity rebates’ distinct from other rebates inducing loyalty. More particularly, it challenges the fact that the General Court considered it unnecessary to assess all the circumstances before coming to the conclusion that the impugned conduct amounted to an abuse of a dominant position under Article 102 TFEU. Intel also challenges the alternative assessment of ‘capability’ (22) which the General Court carried out. In its view, that assessment does not properly take into consideration several circumstances that are relevant for the purposes of determining whether the impugned conduct is capable of restricting competition. Although closely linked to a review of the facts, that question cannot escape judicial review since the Court of Justice has jurisdiction under Article 256 TFEU to review the General Court’s legal characterisation of those facts and the legal conclusions it has drawn from them.
54. As for the substance of the first ground of appeal, I shall begin by examining whether the General Court was right to consider that in assessing the ‘exclusivity rebates’ offered by Intel to the OEMs in question and the marketing arrangements agreed upon with MSH, it was not necessary to look at ‘all the circumstances’ to determine whether the impugned conduct amounts to an abuse of a dominant position contrary to Article 102 TFEU. In that context, the basic tenets of the relevant case-law will be set out to illustrate that the case-law requires an assessment of all the circumstances. As a logical corollary to that conclusion, I shall move on to examine the alternative assessment carried out by the General Court concerning the capability of the rebates offered by the appellant to restrict competition.
(a) The principal assessment of the appellant’s rebates and payments carried out by the General Court
55. As mentioned above (points 44 to 46), on the basis of the Court’s case-law, the General Court distinguished between three categories of rebates: volume-based rebates (category 1), ‘exclusivity rebates’ conditional on the customer obtaining all or most of its requirements from the dominant undertaking (category 2), and other types of rebates where the grant of a financial incentive is not directly linked to exclusive or quasi-exclusive supply (category 3). (23)
56. Specifically, the General Court found that the rebates granted to Dell, HP, NEC and Lenovo, to which the Commission referred, in particular, in Article 1(a) to (d) of the contested decision, are ‘exclusivity rebates’ falling within category 2. This was so because those rebates were conditional on the fact that those companies purchased from Intel, at least in a certain segment, either all their x86 CPU requirements (in the case of Dell and Lenovo) or most of their requirements (namely 95% in the case of HP and 80% in the case of NEC). (24) In relation to the payments made to MSH, the General Court found that the Commission was not required to examine the circumstances of the case, but was required only to demonstrate that the appellant had granted a financial incentive which was subject to an exclusivity condition. (25)
57. In particular, relying on the statement of the Court in Hoffmann-La Roche, (26) the General Court held that whether an ‘exclusivity rebate’ can be categorised as abusive does not depend on an analysis of the capability of the rebates to restrict competition in light of the circumstances of the case. (27)
58. The appellant claims that that conclusion is wrong in law. In particular, it claims that the General Court wrongly disregarded the relevance of the statements of the Court in other cases pertaining to rebates under Article 102 TFEU, and also those concerning other pricing practices.
59. In the following, I shall explain why I agree with the appellant.
i) The basic tenets of the Court’s case-law regarding rebates
60. On a general note, the case-law of the Court displays a mistrust of a variety of different rebate mechanisms offered by dominant undertakings. This may be explained by the fact that it is generally considered that dominant undertakings have a special responsibility not to allow their conduct to impair competition on the internal market. (28) From this special responsibility it follows that mechanisms which in one way or another tie customers to source supplies from the dominant undertaking are regarded as loyalty-inducing and, thus, presumptively abusive.
61. It follows from the line of authority devolving from the seminal Hoffmann-La Roche judgment that rebates that are conditional on the customer purchasing all or most of its requirements from the dominant undertaking are presumptively unlawful. The same presumption of unlawfulness applies to other rebates which also induce loyalty, even though they are not formally based on exclusivity. Rebates, be they retroactive and individualised, as in the cases of Michelin I, (29) British Airways (30) and Tomra (31) or market-share based and individualised as in Hoffmann-La Roche, (32) have been regarded as anticompetitive by the Court. So far, the only type of rebate that has escaped the presumption of unlawfulness is one which is volume-based. Those rebates are linked solely to the volume of purchases made from an undertaking in a dominant position. (33)
62. The rebates and payments offered by Intel can be described as market-share-based loyalty rebates. (34) To be eligible for a discount, the customer must source a certain percentage of its requirements from the dominant undertaking. As explained, relying on the statement of the Court in Hoffmann-La Roche, the General Court considered that, where a rebate is an exclusivity rebate falling under category 2, there is no need to consider the capability of such a rebate to restrict competition on the basis of the circumstances of the case. (35)
63. Hoffmann-La Roche dealt with a market-share-based rebate scheme which was conditional on the customer purchasing a certain percentage of its requirements from the dominant undertaking. More specifically, the rebates were incremental depending on the percentage of the turnover reached by the purchases. (36) In that case, the Court held that, save in exceptional circumstances, loyalty rebates are not based on an economic transaction which justifies this burden or benefit. Instead, they are, in the Court’s view, designed to remove or restrict the purchaser’s freedom to choose its sources of supply and to deny other producers access to the market. (37) According to the Court, therefore, ‘an undertaking which is in a dominant position on a market and ties purchasers … by an obligation or promise on their part to obtain all or most of their requirements exclusively from that undertaking abuses its dominant position within the meaning of Article [102 TFEU], whether the obligation in question is stipulated without further qualification or whether it is undertaken in consideration of the grant of a rebate’. (38) The Court went on to hold that the ‘same applies if the said undertaking, without tying the purchasers by a formal obligation, applies, either under the terms of agreements concluded with these purchasers or unilaterally, a system of loyalty rebates, that is to say, discounts conditional on the customer’s obtaining — whether the quantity of its purchases is large or small — all or most of its requirements from the undertaking in a dominant position’. (39)
64. In making that seminal statement, the Court made no mention of the need to consider all the circumstances when determining whether an abuse of a dominant position had been established to the requisite legal standard.
65. Seen in that light, it is perhaps not surprising that the General Court concluded as it did.
66. However, it should be pointed out already here that in Hoffmann-La Roche the conclusion concerning the unlawfulness of the rebates in question was, nevertheless, based on a thorough analysis of, inter alia, the conditions surrounding the grant of the rebates and the market coverage thereof. (40) It was on the basis of that assessment that the Court held that the loyalty rebates in question were, in that case, intended, by granting a financial advantage, to prevent customers from obtaining their supplies from competing producers.
67. Since Hoffmann-La Roche, as the General Court rightly pointed out, (41) the case-law has mainly centred on devising the appropriate criteria for determining whether an undertaking has abused its dominant position by making use of rebate schemes that are not directly linked to exclusive or quasi-exclusive supply. These are the rebates that, in the taxonomy employed in the judgment under appeal, fall within category 3.
68. In that subsequent case-law, the Court has constantly reiterated the statement of principle stemming from Hoffmann-La Roche concerning the presumptive abusiveness of loyalty rebates. Yet, as ACT correctly pointed out at the hearing, in practice it has consistently taken into account ‘all the circumstances’ in ascertaining whether the impugned conduct amounts to an abuse of a dominant position contrary to Article 102 TFEU.
69. In Michelin I, a target rebate case, the Court held that it is necessary to consider all the circumstances when dealing with a discount system which is not based on exclusive dealing or on an obligation to obtain a specific portion of supplies from the dominant undertaking. (42) In further subsequent cases relating to rebates that are not directly conditional on exclusivity, the Court held that in determining whether an undertaking has abused its dominant position it is useful to consider the criteria and the rules for the grant of the discount and whether, in providing an advantage not based on any economic service justifying it, the discount tends to remove or restrict the buyer’s freedom to choose his sources of supply, to bar competitors from access to the market, or to strengthen the dominant position by distorting competition. (43)
70. Reiterating a statement of principle concerning a presumptive abusiveness is, as shown in the Court’s case-law, however, not the same thing as failing to consider the circumstances in a concrete case. In fact, the judgment under appeal constitutes one of the very few cases where the Court’s statement in Hoffmann-La Roche has been applied verbatim, without examining the circumstances of the case, before concluding that an undertaking has abused its dominant position. (44) To justify that strict approach to ‘exclusivity rebates’, the General Court considered in the judgment under appeal that the rebates and payments made by Intel were conditional on exclusivity (in a similar, yet not identical manner as in Hoffmann-La Roche, given the lack of a formal exclusivity obligation). That circumstance served to distinguish the present case from those mentioned in the previous point.
71. It could, therefore, easily be concluded that, on its face, the judgment under appeal simply reaffirms the existing case-law and applies it to Intel’s conduct.
72. However, such a conclusion would overlook the importance of the legal and economic context in accordance with that same case-law.
ii) The circumstances of the case as a means to determine whether the impugned conduct has a likely effect on competition
73. In this section, I shall explain why an abuse of dominance is never established in the abstract: even in the case of presumptively unlawful practices, the Court has consistently examined the legal and economic context of the impugned conduct. In that sense, the assessment of the context of the conduct scrutinised constitutes a necessary corollary to determining whether an abuse of dominance has taken place. That is not surprising. The conduct scrutinised must, at the very least, be able to foreclose competitors from the market in order to fall under the prohibition laid down in Article 102 TFEU. (45)
74. Even a brief perusal of the cases discussed above (points 66 to 69), shows that the case-law does not omit to look at the legal and economic context of the conduct — or, to employ the standard formula in Article 102 TFEU cases, ‘all the circumstances’ — in order to determine whether an undertaking has abused its dominant position. That is the case for both rebates conditioned on exclusivity and other types of loyalty inducing arrangements.
75. Accordingly, in my view, the General Court’s interpretation of Hoffmann-La Roche misses an important point.Contrary to what was held in the judgment under appeal, (46) in Hoffmann-La Roche the Court considered several circumstances relating to the legal and economic context of the rebates in finding that the undertaking in question had abused its dominant position.True, that judgment does not explicitly state that an analysis of all the circumstances is crucial for determining whether the impugned conduct amounts to an abuse of a dominant position. Nevertheless, as noted above (point 66), a closer look at the judgment shows that the Court examined in commendable detail the particularities of the pharmaceutical market in question, the market coverage of the rebates, as well as the terms and conditions of the contracts between the dominant undertaking and its customers. (47) On the basis of that detailed analysis of the legal and economic context of the rebates, namely the conditions for the grant of the rebates, the market coverage thereof, as well as the duration of the rebate arrangements, the Court reached the conclusion that loyalty rebates are unlawful, save in exceptional circumstances. (48)
76. As the General Court acknowledged in the judgment under appeal, (49) apart from Hoffmann-La Roche, the case-law of the Court pertaining to rebate schemes (other than those purely based on volume) has consistently and explicitly held it to be of particular importance to consider all the circumstances in order to determine whether the impugned conduct amounts to an abuse of a dominant position contrary to Article 102 TFEU. (50) That is, as such, not surprising: apart from Hoffmann-La Roche, I am not aware that the Court has dealt with other cases pertaining to exclusive supply obligations similar to those at issue there. Unsurprisingly, therefore, the need to consider all the circumstances was yet again reiterated in Post Danmark II, a preliminary ruling handed down after the judgment under appeal was delivered, in relation to retroactive rebates that are not tied to an exclusivity obligation. (51)
77. But what does an assessment of ‘all the circumstances’ entail?
78. As I see it, the analysis of ‘context’ — or ‘all the circumstances’, as it is termed in the Court’s case-law — aims simply but crucially to ascertain that it has been established, to the requisite legal standard, that an undertaking has abused its dominant position. (52) Even in the case of seemingly evident exclusionary behaviour, such as pricing below cost, context cannot be overlooked. (53) Otherwise, conduct which, on occasion, is simply not capable of restricting competition would be caught by a blanket prohibition. Such a blanket prohibition would also risk catching and penalising pro-competitive conduct.
79. That is why context is essential.
iii) There are only two categories of rebates according to the case-law
80. For the purposes of Article 102 TFEU, loyalty rebates constitute — as I see it — a near-equivalent to a restriction by object under Article 101 TFEU. This is because loyalty rebates, like restrictions by object, are presumptively unlawful. However, as already noted above (point 61), loyalty rebates must be understood as encompassing not only those conditional on the customer sourcing all or most of its requirements with the dominant undertaking, but also other pricing structures conditional upon the customer reaching a particular target.
81. Contrary to what the General Court held in the judgment under appeal, the case-law distinguishes between two, not three, categories of rebates. On the one hand, some rebates are presumptively lawful, such as volume-based rebates. (54) Any investigation into the (un)lawfulness of such rebates necessarily requires a full examination of their actual or potential effects. Those rebates are not at issue here.
82. On the other hand, for loyalty rebates (which are presumptively unlawful), whether those rebates are directly conditional upon exclusivity or not, the Court employs an approach that displays certain similarities with the approach to restrictions by object under Article 101 TFEU. That is because, under that provision too, in order to ascertain whether certain conduct constitutes a restriction by object, the legal and economic context of the impugned conduct must first be examined so as to exclude any other plausible explanation for that conduct. In other words, the particular context of the impugned conduct is never ignored.
83. As already mentioned above, in Hoffmann-La Roche the Court did consider all the circumstances. A requirement to that effect was later explicitly formulated by the Court in Michelin I in relation to rebates not directly connected to exclusivity. That requirement has subsequently been fine-tuned in cases such as British Airways, Michelin II and Tomra.The purpose of an analysis of all the circumstances is to ascertain whether an abuse of dominance has been established to the requisite legal standard and whether, as a consequence, the rebates are capable of anticompetitive foreclosure.
84. In the judgment under appeal, however, the General Court went one step further. By applying the statement of the Court in Hoffmann-La Roche to the letter, without placing that statement in its proper context, it distinguished one sub-type of loyalty rebate, which it termed ‘exclusivity rebates’, from other types of rebates that induce loyalty. (55) In doing so, it created a ‘super category’ of rebates for which consideration of all the circumstances is not required in order to conclude that the impugned conduct amounts to an abuse of dominance contrary to Article 102 TFEU. More importantly, the abusiveness of such rebates is assumed in the abstract, based purely on their form.
85. That is by no means a methodologically self-evident step to take. The following four reasons explain why.
– An assumption of unlawfulness by virtue of form cannot be rebutted
86. Firstly, if it were accepted that ‘exclusivity rebates’ constitute a distinct category of rebate that ought to be distinguished from other types of rebate schemes that induce loyalty, the underlying assumption of unlawfulness would no longer be open to rebuttal. (56) This is so because such a prohibition is based on the form of the conduct instead of its effects.
87. In fact, the judgment under appeal seems to adopt the starting point that an ‘exclusivity rebate’, when offered by a dominant undertaking, can under no circumstances have beneficial effects on competition. That is because, according to the General Court, competition is restricted by the mere existence of a dominant position itself. (57) That viewpoint amounts to negating the possibility, already accepted in Hoffmann-La Roche, (58) and reiterated in the judgment under appeal, (59) of invoking an objective (pro-competitive) justification for the use of the rebates in question.
88. Contrary to what the Commission suggested at the hearing, there is no efficiency or other consideration that can assist an undertaking in justifying the use of ‘exclusivity rebates’ where the prohibition is concerned with form, rather than effects. (60) Indeed, irrespective of the effects, the form remains the same. That is problematic. The dominant undertaking should be able, as the General Court rightly pointed out in the judgment under appeal, (61) and as the Commission itself accepted in its written pleadings, to justify the use of a rebate scheme by providing evidence that the exclusionary effect produced may be counterbalanced or even outweighed by advantages in terms of efficiency. (62)
– Loyalty rebates are not always harmful
89. Secondly, creating a ‘super-category’ of rebates is warranted only if it is considered that there can be no redeeming features relating to arrangements that are conditioned on exclusivity, irrespective of the individual circumstances of the particular case. Paradoxically, however, the General Court itself admitted that exclusivity conditions can also have beneficial effects. Yet, it rejected any need to look into those effects on the ground that, owing to the dominant position of the undertaking on the market, competition is already and irrefutably restricted. (63)
90. Experience and economic analysis do not unequivocally suggest that loyalty rebates are, as a rule, harmful or anticompetitive, even when offered by dominant undertakings. (64) That is because rebates enhance rivalry, the very essence of competition.
91. It is true, however, that the greatest competitive concern in relation to rebates is said to arise where the customers of a dominant undertaking must carry a percentage of its products and/or where the discount is conditional on the customer purchasing all (or a substantial part) of its requirements from it. That could be seen as an argument in favour of treating ‘exclusivity rebates’ more strictly. Yet, other types of rebates can have a similar distorting effect. This is so even where the scheme is not explicitly linked to exclusivity. (65)
92. Indeed, as the case-law clearly illustrates, a mechanism that induces loyalty may take different forms. As was the case in Hoffmann-La Roche (66) and Tomra, (67) the loyalty mechanism can be inherent in the requirement that the customer purchase all or most of its material requirements from the dominant undertaking. It can also take the form of individualised sales targets (68) or bonuses, (69) which are not necessarily tied to a particular proportion of requirements or sales.
93. Seen in that light, there is no objective reason why category 2 rebates should receive a stricter treatment than those falling under category 3.
– The effects of loyalty rebates are context-dependent
94. Thirdly, contemporary economic literature commonly emphasises that the effects of exclusivity are context-dependent. (70) Conversely, few commentators would deny that loyalty rebates in particular can — depending on the circumstances — have an anticompetitive foreclosure effect.
95. Awareness of context-dependency may also help explain why the Court has, in its recent Article 102 TFEU case-law, highlighted the importance of considering all the circumstances. It did so, inter alia, in Tomra. Admittedly, as the General Court observed, (71) the rebates under scrutiny on appeal in Tomra concerned individualised retroactive rebates, that is, rebates belonging to category 3 in the taxonomy employed in the judgment under appeal. However, in Tomra the Court did not explicitly draw a distinction between rebates belonging to categories 2 and 3. It simply observed that where rebates are conditional on the customer obtaining all or most of its requirements from the dominant undertaking, it is necessary to consider all the circumstances in order to determine whether the impugned conduct amounts to an abuse of a dominant position. (72)
96. The parties draw opposing conclusions from the Court’s statement: both parties argue that it settles the controversy surrounding the legal test to be applied in relation to ‘exclusivity rebates’ once and for all. Yet they continue to disagree on what that test should be.
97. To my mind, the Court’s statement in Tomra does not assist us here. As the parties’ diametrically opposed interpretations illustrate, the language employed in that judgment regarding the type of rebates that should be subject to an assessment of all the circumstances is simply too ambiguous.
98. Rather, it is open to question whether the distinction drawn by the General Court in the judgment under appeal between Tomra and Hoffmann-La Roche is warranted. In that regard, I would draw attention to two points.
99. On the one hand, similarly to Tomra, in Hoffmann-La Roche, the rebates considered there displayed certain characteristics of individualised retroactive rebates. In fact, several contracts examined in Hoffmann-La Roche not only contained a rebate clause relating to the majority of the purchaser’s requirements. They also included rebate clauses providing for a discount, the percentage of which was to increase depending on whether the percentage of the purchaser’s estimated requirements had been met during the reference period. (73) On the other hand, even assuming that a distinction would be warranted due to a supposed intrinsic difference between the rebates involved in the two cases — which I consider not to be the case — the Court by no means overlooked the context of the impugned rebates in Hoffmann-La Roche. Why should it have done so in Tomra, more than three decades later?
100. If anything, the difference between the rebates at issue in Tomra and Hoffmann-La Roche is one of degree rather than kind. The same can be said about Post Danmark II, a case in which the relevance of context and all the circumstances in determining whether the impugned conduct amounts to an abuse of a dominant position was recently confirmed. (74)
– Related practices require consideration of all the circumstances
101. Fourthly, and lastly, the case-law relating to pricing and margin squeeze practices requires, as the appellant rightly points out, consideration of all the circumstances in order to determine whether the undertaking in question has abused its dominant position. (75)
102. The General Court dismissed the relevance of that case-law on the ground that, unlike an incentive to exclusive supply, a particular price cannot be abusive in and of itself. (76) Yet, the judgment under appeal considers Intel’s rebates to be anticompetitive on account of price. (77) To my mind, dismissing the relevance of that case-law is problematic: it results in an unwarranted distinction between different types of pricing practices. Indeed, loyalty rebates, margin squeeze practices as well as predatory pricing possess the common feature of constituting ‘price based exclusion’. (78)
103. It goes without saying that it is of the utmost importance that legal tests applied to one category of conduct are coherent with those applied to comparable practices. Sound and coherent legal categorisation benefits not only undertakings in terms of increased legal certainty, but also assists competition authorities in the enforcement of competition law. Arbitrary categorisation does not.
104. The Court seems to agree. Most recently in Post Danmark II, the Court applied the case-law relating to pricing and margin squeeze practices to support its findings in relation to a rebate scheme offered by a dominant undertaking. (79) It is true, however, that Post Danmark II could also be read as supporting the view that, in so far as ‘exclusivity rebates’ are concerned, there might be no need to consider all the circumstances. (80) The reason for this is that, in that case, the Court distinguished the rebates in question from those based on an exclusivity obligation before holding that all the circumstances must be considered to determine whether the dominant undertaking has abused its dominant position. Indeed, if anything, the retroactive rebates considered by the Court in that case resembled those that were considered, in the judgment under appeal, to be fidelity-inducing rebates belonging to category 3. (81)
105. As explained above, such a distinction is a distinction without a difference (given that the difference lies in form rather than effects). More fundamentally, however, such a reading of that judgment would stand in contrast to the approach of the Court — sitting in Grand Chamber — in Post Danmark I, where it held that in so far as pricing practices are concerned, all the circumstances must be considered. (82) Tellingly in fact, the Court reiterated, at a different point in its judgment in Post Danmark II — and there without distinguishing between different types of rebates schemes — that ‘the assessment of whether a rebate scheme is capable of restricting competition must be carried out in the light of all relevant circumstances’. (83) Assuredly it did so in order to ensure a coherent jurisprudential approach to the assessment of conduct falling under the purview of Article 102 TFEU.
iv) Intermediate conclusion
106. Having regard to the above, ‘exclusivity rebates’ should not be regarded as a separate and unique category of rebates that requires no consideration of all the circumstances in order to determine whether the impugned conduct amounts to an abuse of a dominant position. Accordingly, in my opinion, the General Court erred in law in considering that ‘exclusivity rebates’ can be categorised as abusive without an analysis of the capacity of the rebates to restrict competition depending on the circumstances of the case.
107. That said, the General Court proceeded, in the alternative, to assess in detail whether the rebates and payments offered by the appellant were capable of restricting competition. That is to say, it investigated ‘all the circumstances’. That is why the finding regarding an error in law made in the previous point does not necessarily entail the setting aside of the judgment under appeal. Rather, such a conclusion can only be reached in so far as the alternative assessment performed by the General Court discloses an error in law.
108. Therefore, it is necessary to move on to consider that alternative assessment.
(b) The alternative assessment of capability carried out by the General Court
109. The appellant puts forward, in essence, three sets of arguments that call into question the General Court’s alternative assessment. First, it argues that the General Court erred in law in upholding the finding of an abuse by the Commission without considering the likelihood of anticompetitive effects. Second, the factors which the General Court took into account were either irrelevant or assessed incorrectly. Third, the General Court failed to assess correctly several other factors that are crucial for a finding of an abuse.
110. The Commission maintains that there is no higher ‘likelihood’ threshold that needs to be met (in comparison to the ‘capability’ standard) to establish the existence of an abuse of dominance: capability is enough. In its view, the arguments put forward by the appellant do not cast doubt on the findings made in the judgment under appeal regarding the capability of Intel’s conduct to restrict competition.
111. By its arguments, the appellant calls into question the legal test that was applied in the judgment under appeal to ascertain that the impugned conduct was capable of restricting competition. First: what level of probability is required in a capability assessment? Second: what are the relevant circumstances to be taken into consideration in determining whether the conduct in question is capable of restricting competition? I shall examine each of those questions in turn.
i) Capability and/or likelihood
112. The appellant argues that the General Court erred in law in upholding the finding of an abuse without considering the likelihood of anticompetitive harm from the impugned conduct.
113. In its alternative assessment, the General Court explained that the Commission may confine itself to determining that the conduct scrutinised was capable of restricting competition. In addition, it pointed out that the Commission does not need to prove that an actual foreclosure effect has materialised, even where all the circumstances are considered. (84)
114. Certainly, evidence of actual effects does not need to be presented. This is because it is sufficient, in relation to conduct that is presumptively unlawful, that the impugned conduct be capable of restricting competition. Importantly, however, that capability cannot merely be hypothetical or theoretically possible. Otherwise, there would be no need to consider all the circumstances in the first place.
115. True, there is some discrepancy in the case-law regarding the terminology employed. The case-law refers to capability and likelihood, sometimes even interchangeably. (85) It is my understanding that those terms designate one and the same compulsory step in an analysis seeking to determine whether the use of loyalty rebates amounts to an abuse of a dominant position.
116. But what degree of probability of anticompetitive foreclosure is required? That question is at the core of the disagreement between the appellant and the Commission regarding the adequacy of the General Court’s assessment of capability: whereas the Commission believes that the assessment was adequate, Intel claims that the General Court failed to verify that the appellant’s conduct could, in the circumstances of the case, restrict competition.
117. The aim of the assessment of capability is to ascertain whether, in all likelihood, the impugned conduct has an anticompetitive foreclosure effect. For that reason, likelihood must be considerably more than a mere possibility that certain behaviour may restrict competition. (86) Contrariwise, the fact that an exclusionary effect appears more likely than not is simply not enough. (87)
118. Although it is certainly true that in its case-law the Court has consistently emphasised the special responsibility of dominant undertakings, that responsibility cannot be taken to mean that the threshold for the application of the prohibition of abuse laid down in Article 102 TFEU can be lowered to such an extent as to become virtually non-existent. That would be the case if the degree of likelihood required for ascertaining that the impugned conduct amounts to an abuse of a dominant position was nothing more than the mere theoretical possibility of an exclusionary effect, as seems to be suggested by the Commission. If such a low level of likelihood were accepted, one would have to accept that EU competition law sanctions form, not anticompetitive effects.
119. Clearly, that would considerably hamper the attainment of the objectives of EU competition law. To assume the existence of an abuse on the basis that, on balance, anticompetitive foreclosure seems more likely than not risks capturing not only isolated instances of practices, but a non-negligible number of practices that may, in reality, be pro-competitive. The cost of error of such an approach would be unacceptably high due to over-inclusion.
120. To avoid over-inclusion, the assessment of capability as concerns presumptively unlawful behaviour must be understood as seeking to ascertain that, having regard to all circumstances, the behaviour in question does not just have ambivalent effects on the market or only produce ancillary restrictive effects necessary for the performance of something which is pro-competitive, but that its presumed restrictive effects are in fact confirmed. Absent such a confirmation, a fully-fledged analysis has to be performed.
121. Consequently, the question that arises here is whether the General Court’s assessment of capability is conclusive because, on the basis of that assessment, it is possible to confirm that the appellant had abused its dominant position contrary to Article 102 TFEU. In particular, it must be ascertained whether, as required by the case-law, that assessment can confirm that the rebates remove or restrict the customer’s freedom to choose its sources of supply, bar competitors from access to the market, or strengthen the dominant position by distorting competition. (88)
ii) The factors considered by the General Court to support a finding of abuse
122. In the judgment under appeal, the General Court based the finding that the rebates and payments offered by the appellant were capable of restricting competition on the following factors: (i) the appellant was an unavoidable trading partner for the customers concerned; (ii) the OEMs’ low operating margins made the rebates attractive and reinforced their incentive to comply with the exclusivity condition; (iii) the appellant’s rebates were taken into consideration by its customers in deciding to obtain all or almost all of their requirements from that company; (iv) the appellant’s two different practices complemented and reinforced each other; (v) the appellant targeted undertakings which were of particular strategic importance for market access; and finally (vi) the appellant’s rebates were part of a long-term strategy aimed at barring AMD’s access to the most important sales channels. (89)
123. Intel considers that those factors cannot be relied upon to establish, to the requisite legal standard, that Intel’s rebates and payments were capable of producing an anticompetitive foreclosure effect. More specifically, the appellant claims that the factors relied upon by the General Court boil down to two points: that the OEMs took Intel’s rebates into account because the rebates were attractive, and that Intel used two complementary infringements to exclude AMD from important customers.
124. First, therefore, the appellant contests the relevance of the circumstance that the rebates and payments in question were actually taken into account in the commercial decisions of those benefiting from them. (90)
125. I agree with the appellant.
126. An attractive offer, which translates into a financial incentive to stay with the supplier making that offer, may be a factor that points to a loyalty-inducing effect at the level of an individual customer. However, it is unhelpful in establishing that the rebates have a likely anticompetitive foreclosure effect. Indeed, as the appellant rightly points out, it is the essence of competition that customers take lower prices into account in making purchasing decisions. Put differently, the fact that a lower price is in fact taken into account makes a foreclosure effect possible, but, on the other hand, it does not rule out the contrary either. In other words, that factor is simply inconclusive for the purposes of determining whether the impugned conduct is capable of restricting competition to the requisite degree of likelihood.
127. Second, the appellant contends that the existence of an overall strategy comprising two types of infringement (rebates and payments as well as naked restrictions), which were considered to complement and strengthen one another, cannot establish capability to restrict competition. (91)
128. While a strategy of foreclosure may certainly provide an indication of a subjective intent to foreclose competitors, the mere will to do so does not translate into capability to restrict competition. A more fundamental problem can, however, be gleaned from the reasoning of the General Court. In fact, a closer look at the judgment under appeal reveals that the General Court has put the cart before the horse: it relied on the existence of an overall strategy based on two complementary infringements to determine the capability of the impugned conduct to restrict competition. In doing so, the General Court began its reasoning from the assumption that the strategy under consideration was abusive, rather than assessing all the circumstances to determine whether an infringement had been established to the requisite legal standard.
129. Having first addressed those two specific criticisms, I will now move on to examine the appellant’s more general criticism regarding the assessment of capability. The appellant claims that the factors considered as relevant are not sufficient to ascertain that the impugned conduct is capable of anticompetitive foreclosure. In particular, the General Court failed to attribute relevance to other factors that are of fundamental importance in such an assessment.
130. To recall, the assessment of all the circumstances aims to ascertain that the impugned conduct results, in all likelihood, in anticompetitive foreclosure. Keeping that in mind, the question that arises is the following: Are the findings made in the judgment under appeal — that Intel was an unavoidable trading partner and that the impugned rebates and payments targeted undertakings which were of particular strategic importance for market access — sufficient in law to establish Intel’s liability? The answer to that question hinges upon whether the circumstances argued to be crucial by Intel, and regarded as irrelevant by the General Court, cast doubt on the presumed anticompetitiveness of Intel’s conduct.
131. I shall address that issue next.
iii) Other circumstances
132. The appellant argues that the General Court erred in its analysis of the circumstances of the case by failing to consider the following circumstances: (i) the insufficient market coverage of the challenged rebates and payments; (ii) the short duration of the challenged rebates; (iii) market performance of the competitor and declining prices; and (iv) the AEC test carried out by the Commission.
133. For its part, the Commission considers that the judgment under appeal established, to the requisite legal standard, that the rebates and payments offered by Intel were capable of anticompetitive foreclosure. The unchallenged factors suffice to uphold the conclusion that Intel’s rebates and payments were capable of restricting competition.
134. I disagree with the Commission.
135. As already explained, in a somewhat similar fashion to the enforcement shortcut concerning restrictions by object under Article 101 TFEU, the assessment of all the circumstances under Article 102 TFEU involves examining the context of the impugned conduct to ascertain whether it can be confirmed to have an anticompetitive effect. If any of the circumstances thus examined casts doubt on the anticompetitive nature of the behaviour, a more thorough effects analysis becomes necessary.
136. As I will explain in the following points, the assessment of all the circumstances should have led the General Court to conclude that, in order to determine whether the impugned conduct amounted to an abuse of a dominant position contrary to Article 102 TFEU, an analysis of the actual or potential effects of that conduct was necessary.
– Market coverage
137. The appellant claims that consideration of a likely effect on competition must take account of the market coverage of the rebates in question. In its view, loyalty rebates are unlikely to restrict competition when their market coverage is low, because larger portions of the market may be reached by competitors without them having to match those discounts. The appellant also points out that, in its case, the tied share of the market was, in comparison, on average considerably lower than in cases such as Tomra and Van den Bergh Foods. (92) For example, in Tomra, the tied share of the market was (on average) 39%. (93) The Commission, on the other hand, considers the issue of market coverage not to be relevant in determining whether the impugned conduct is capable of anticompetitive foreclosure.
138. In the context of the alternative assessment of capability, the General Court stated that the market coverage of the rebates and payments offered by the appellant was approximately 14% on average during the total period of the infringement (if the calculation is not limited solely to the contestable share of demand). (94) It deemed that to be significant. (95) According to the judgment under appeal, the rebates and payments offered by Intel can be distinguished from the circumstances underlying Van den Bergh Foods given that the form of the scheme under consideration was different from that at issue in the present proceedings. (96)
139. I, for one, am not convinced that the case-law referred to by the appellant is irrelevant, as stated by the General Court. The exclusivity mechanism in Van den Bergh Foods did indeed function through the supply of a free freezer. But that is a distinction without a difference. The supply of that freezer was conditional on its exclusive use for storing the dominant undertaking’s ice cream. As a result, 40% of all retailers of ice cream were subject to product exclusivity. (97)
140. As explained, the focus of EU competition rules has consistently been on effects, not form. Viewed in that light, the size of the tied part of the market is equally relevant, irrespective of the form of the scheme. That is why it is generally accepted that the likelihood of negative effects on competition increases in line with the size of the tied market share. (98)
141. That said,defining the level of market coverage that may cause anticompetitive effects is by no means an arithmetic exercise. Unsurprisingly, therefore, the Court has rejected the idea that a precise threshold of foreclosure of the market must be determined, beyond which the practices at issue can be regarded as abusive for the purposes of applying Article 102 TFEU. This was confirmed by the Court in Tomra. (99)
142. It is certainly true that thresholds may prove problematic due to the specificities of different markets and the circumstances of each individual case. For example, where loyalty rebates target customers that are of particular importance for competitors to enter or expand their share of the market, even modest market coverage can certainly result in anticompetitive foreclosure. Whether that is the case will depend on a number of factors specific thereto.
143. Seen in that light, a market coverage of 14% may or may not have an anticompetitive foreclosure effect. What is certain, however, is that such market coverage cannot rule out that the rebates in question do not have an anticompetitive foreclosure effect. This is so even assuming that the rebates and payments in question target key customers. (100) Quite simply, 14% is inconclusive.
144. That inconclusiveness is not remedied by the reliance, in the judgment under appeal, on the established fact that the appellant was an unavoidable trading partner on the CPU market. It must be noted that, according to the General Court, the fact that an undertaking constitutes an unavoidable trading partner indicates, at the very least, that an ‘exclusivity rebate’ or payment offered by such an undertaking is capable of restricting competition. (101)
145. That conclusion is only correct if it is accepted that the degree of likelihood required does not amount to anything more than the mere possibility that certain behaviour has anticompetitive effects. As previously explained, however, the assessment of all the circumstances seeks to ascertain that the impugned conduct has, in all likelihood, an anticompetitive effect.
146. On that basis, I conclude that the assessment of market coverage carried out in the judgment under appeal is inconclusive. Importantly, it cannot establish, to the requisite legal standard, that the share of the market affected by the rebates and payments was sufficient to bring about anticompetitive foreclosure.
– Duration
147. The appellant considers that the duration of a rebate arrangement is critical to an analysis of capability to restrict competition. In particular, it calls into question the assessment of duration in the judgment under appeal, which was based on the cumulation of multiple short-term agreements.
148. The Commission, on the other hand, argues that Intel is mistaken in considering that the potential exclusionary effects of loyalty rebates can only derive from a contractual obligation: on the contrary, the dominant undertaking’s market power renders such contractual obligations unnecessary. In short: duration is irrelevant.
149. Specifically, the General Court held that the relevant criterion is not the duration of the notice period for terminating a contract or the fixed duration of an individual contract which forms part of a series of successive contracts. Instead, the relevant criterion is, in its view, the overall period during which the appellant applies exclusivity rebates and payments vis-à-vis a customer. (102) That period amounted to approximately five years in the case of MSH, three years in the case of Dell and NEC, over two years in the case of HP and approximately one year in the case of Lenovo. The grant of ‘exclusivity rebates’ and payments during such periods was deemed generally capable of restricting competition. In the view of the General Court, that was in particular true of the CPU market, which is highly dynamic and characterised by short product-life cycles. (103)
150. At the outset, I would observe that the short duration of an arrangement does not exclude that the arrangement is capable of having anticompetitive effects. Similarly, the question of whether the overall period is short or long in the abstract is irrelevant.
151. Where, as here, exclusivity hinges, in the final analysis, on the choice of the customer to source the majority of supplies from the dominant undertaking, it cannot simply be assumed — ex post facto — that the cumulation of short-term agreements shows that those rebates are capable of restricting competition.
152. There are at least two reasons for this.
153. First, unlike in a situation of exclusive dealing, there is no penalty involved in switching suppliers. This is true provided that a competitor can at least in principle match the lost rebate. If, however, the competitor cannot sell the goods in question without a loss, the customer is de facto tied to the dominant undertaking. Viewed from that perspective, the size of the rebate cannot be regarded as wholly insignificant either.
154. More specifically, in an ex post analysis of duration, as here, it is necessary to determine whether another supplier could have compensated for the loss of the rebates. Otherwise, customers’ choice to stay with the dominant undertaking would automatically be considered to indicate an abuse, despite the fact that customers could walk away from their agreements with the dominant undertaking without constraints.
155. Plainly, it cannot simply be assumed, on the basis of a customer’s choice to stay with the dominant undertaking, that that choice constitutes an expression of abusive behaviour. That is because there may be other plausible explanations for that choice. Those include, but are not limited to, quality concerns, the security of supply, and the preference of end-users.
156. Second, a long overall duration of the arrangement can certainly point to a loyalty-inducing effect of the rebate mechanism at the level of the individual customer. However, unless further compelling evidence to that effect has been presented, the fact that a customer has decided to stay with the dominant undertaking cannot suffice to establish that the rebates offered are capable of restricting competition to the requisite standard. Indeed, it should be borne in mind that where the customer has the option of switching suppliers on a regular basis, even where that option has not been exercised, loyalty rebates will also enhance rivalry. Thus, they can also have a pro-competitive effect.
157. Therefore, I consider that the assessment of duration undertaken in the judgment under appeal — which was limited to considering the overall duration of the arrangements scrutinised — is inconclusive. Quite simply, that assessment does not assist in ascertaining that that conduct has, in all likelihood, an anticompetitive effect.
– Market performance of the competitor and declining prices
158. The appellant contests the General Court’s rejection of AMD’s market performance and evidence concerning lack of foreclosure (declining prices of x86 CPUs) as irrelevant in the context of an assessment of capability.
159. According to the General Court, the competitor’s success and falling prices could not exclude that the appellant’s practices had no effect. It observed that in the absence of those practices, it may legitimately be considered that the increase in the competitor’s market share and in its research and development investment and the decrease in the price of x86 CPUs might have been greater. (104)
160. In my view, the General Court was correct in considering that AMD’s market performance and the decrease in the price of x86 CPUs are inconclusive for the purposes of ascertaining whether the impugned conduct is capable of restricting competition. However, the same conclusion would apply even if competitor performance was poor. As I see it, the consideration of such factual elements can only serve a meaningful purpose as a part of a detailed assessment of actual or potential effects on competition. It does not assist in determining whether a presumptively unlawful rebate scheme is capable of restricting competition.
– The AEC test
161. The appellant considers that when, as in the present case, the Commission has carried out a substantial analysis of the economic circumstances in relation to allegedly abusive conduct it is wrong in law to ignore that analysis simply because it does not help to establish an infringement.
162. The Commission argues that the AEC test is not relevant for establishing that the impugned conduct is capable of restricting competition. In its view, the case-law of the Court does not support the appellant’s claim that the AEC test should be part of an assessment of all the circumstances.
163. In the judgment under appeal, the General Court dismissed the relevance of the AEC test for the purposes of determining, in the context of an assessment of all the circumstances, whether the impugned conduct is capable of restricting competition. Accordingly, it did not review the test applied by the Commission in the decision at issue. First, it considered that the AEC test was irrelevant because, in view of their form, the Commission is not required to demonstrate the foreclosure capability of ‘exclusivity rebates’ on a case-by-case basis. The judgment under appeal dismissed the relevance of the AEC test essentially because it only serves to verify that the impugned conduct does not make market access impossible. According to the judgment under appeal, ‘exclusivity rebates’ may hamper market access for competitors of the undertaking in a dominant position, even if access to the market is not economically speaking entirely impossible, but only made more difficult. (105) Second, it noted that, even in relation to rebates falling under category 3, the case-law does not require the use of an AEC test. Third, it observed that the AEC test has been deemed relevant by the Court only in pricing and margin squeeze practice cases, which are inherently different from ‘exclusivity rebate’ cases. (106)
164. First, as I have demonstrated above (points 122 to 160), it is necessary to consider all the circumstances to ascertain whether the conduct in question is capable of foreclosing competitors also in so far as ‘exclusivity rebates’ are concerned. In other words, the foreclosure capability must be demonstrated in each individual case. Certainly, the AEC test can be dismissed as irrelevant if one accepts that a mere hypothetical or theoretical possibility that the impugned conduct has an anticompetitive foreclosure effect suffices to establish an abuse. Indeed, in theory, any rebate offered by a dominant undertaking can, in some circumstances, have an anticompetitive effect.
165. However, given that an exclusionary effect is required, the AEC test cannot be ignored. As the General Court noted, the test serves to identify conduct which makes it economically impossible for an as-efficient competitor to secure the contestable share of a customer’s demand. In other words, it can help identify conduct that has, in all likelihood, an anticompetitive effect. By contrast, where the test shows that an as-efficient competitor is able to cover its costs, the likelihood of an anticompetitive effect significantly decreases. That is why, from the perspective of capturing conduct that has an anticompetitive foreclosure effect, the AEC test is particularly useful.
166. As regards the second and the third considerations, I have explained above (points 101 to 105) why the case-law relating to pricing practices and margin squeeze practices should not be disregarded. In any event, any remaining uncertainty in that regard has disappeared with Post Danmark II. That case demonstrates that the case-law pertaining to other types of price-based exclusion cannot simply be disregarded in the context of rebate cases. As the Court confirmed, by reference inter alia to that case-law, the AEC test may prove useful in the context of assessing a rebate scheme too. (107)
167. However, it is also worth observing that in Post Danmark II the Court was careful to qualify its position on the AEC test. It specifically observed that although the AEC test can prove helpful in certain situations, there is no legal obligation to make use of that test. (108) This is in line with the Court’s statement to the same effect in Tomra. In that case, the Court considered that the Commission is not required to demonstrate that loyalty rebates force the competitors of the dominant undertaking to charge prices below cost in order to compete for the contestable share of the market. Rather, the Court held that the Commission could establish that the rebates in question were capable of restricting competition on the basis of qualitative elements indicating their anticompetitive nature. (109)
168. In that sense, it would certainly seem tempting to conclude that, in the present case, there is no need to have recourse to an AEC test. Following that logic, as argued by the Commission, the assessment of capability carried out by the General Court would not contain an error in law in having disregarded the AEC test as irrelevant.
169. However, that standpoint overlooks two issues. Unlike in Tomra, the Commission did in fact carry out an extensive AEC analysis in the decision at issue. More fundamentally still, the other circumstances assessed by the General Court do not unequivocally support a finding of an effect on competition. In those circumstances, it is clear to me that the AEC test cannot simply be ignored as an irrelevant circumstance.
170. Therefore, the General Court erred in law by not considering the AEC analysis carried out by the Commission in the decision at issue as a part of the assessment of all the circumstances.
171. To conclude my analysis of the alternative assessment of capability carried out by the General Court, I observe the following.
172. The circumstances considered in that assessment cannot confirm an effect on competition. At most, that assessment shows that an anticompetitive foreclosure effect of the impugned conduct is theoretically possible, but the effect as such has not been confirmed. As a matter of principle, an assessment of all the circumstances must, at the very least, take into account the market coverage and duration of the impugned conduct. In addition, it may be necessary to consider other circumstances that may differ from case to case. In the present case, the AEC test, precisely because that test was carried out by the Commission in the decision at issue, cannot be ignored in ascertaining whether the impugned conduct is capable of having an anticompetitive foreclosure effect. The assessment of the relevant circumstances should, taken as a whole, allow us to ascertain, to the requisite degree of likelihood, that the undertaking in question has abused its dominant position contrary to Article 102 TFEU. Absent such a confirmation due to, for example, low market coverage, short duration of the impugned arrangements or a positive result of an AEC test, a more thorough economic assessment of the actual or potential effects on competition is necessary for the purposes of establishing an abuse.
(c) Conclusion
173. I have concluded that the General Court erred in law, first, in finding that ‘exclusivity rebates’ constitute a separate and unique category of rebates that require no consideration of all the circumstances in order to establish an abuse of a dominant position contrary to Article 102 TFEU. Second, it erred in law in its alternative assessment of capability by failing to establish, on the basis of all the circumstances, that the rebates and payments offered by the appellant had, in all likelihood, an anticompetitive foreclosure effect.
174. It follows that the first ground ought to be upheld.
C – The second ground of appeal: market coverage in determining whether an undertaking has abused its dominant position
1. The main arguments of the parties
175. In the second ground of appeal, the appellant claims that, irrespective of the conclusion reached concerning the first ground, the market coverage of Intel’s conduct did not, in any event, allow it to restrict competition during 2006 and 2007. In that period, the infringement only concerned MSH and Lenovo. In the appellant’s view, the General Court erred in law in considering that, because the Commission had made a finding of a single and continuous infringement in the decision at issue for the years 2002 to 2007, the finding of an infringement for 2006 and 2007 could be based on the average market coverage across the entire period from 2002 to 2007 (rather than on the market coverage of the conduct in the two years in question). (110)
176. The Commission points out that the second ground of appeal is a mere addition to the first ground. It is based entirely on the same premises as the first ground. That institution considers that market coverage is irrelevant for the purposes of determining whether Intel’s rebates were capable of restricting competition: the market coverage of Intel’s practices only relates to the extent to which those practices actually restricted competition. Owing to the strategic importance of the OEMs targeted in 2006 and 2007, the significance of Intel’s practices cannot be measured simply by reference to market coverage. In that context, the Commission argues that market coverage during those two years should be seen in the light of a single and continuous infringement which relates to the existence of an overall strategy to foreclose AMD from the worldwide market for CPUs.
2. Analysis
177. I have concluded above that the General Court erred in its alternative assessment of capability in accordance with all the circumstances. In particular, it erred in its assessment of market coverage by failing to acknowledge that a tied market share of 14% cannot establish, to the requisite legal standard, that the impugned conduct is capable of restricting competition. On that basis alone, the second ground ought to be upheld too.
178. Nevertheless, I believe this ground merits a brief discussion of its own. This is because the findings of the General Court regarding the existence of a single and continuous infringement constitute the basis for a finding of an infringement for the years 2006 and 2007. In fact, the General Court considered that, in the context of a single and continuous infringement based on an overall strategy of foreclosure, a global assessment of the average share of the tied market was sufficient for the purposes of ascertaining that the conduct in question was capable of anticompetitive foreclosure. (111)
179. Hence, the crux of the present ground of appeal lies in defining the role of the concept of single and continuous infringement in assessing the capability of a single undertaking’s conduct to restrict competition. More specifically, the question arises as to whether recourse to that concept can remedy the fact that market coverage is too small to establish on its own that the impugned conduct was capable of restricting competition during a specific period of time.
180. In the case-law of the Court, the concept of single and continuous infringement has been employed, in particular, in the context of Article 101 TFEU to capture several elements of anticompetitive conduct under the umbrella of one single and continuous infringement for the purposes of enforcement. In that regard, the underlying rationale is to ensure effective enforcement in cases where infringements are composed of a complex of anticompetitive practices that can take different forms and even evolve over time. (112)
181. In other words, the aim is to avoid the unfortunate enforcement outcome where various agreements and concerted practices under Article 101 TFEU, which in reality form part of an overall plan to restrict competition, are treated separately. For that reason, recourse to the concept of single and continuous infringement tempers the burden of proof generally weighing on enforcement authorities regarding the need to prove the continuous nature of the anticompetitive practices scrutinised. More particularly, where a complex of agreements and practices have been implemented over a long period of time, it is not unusual that changes in the scope, form and participants to those agreements and/or practices have taken place during the relevant time period. Without the assistance of the concept of single and continuous infringement, the Commission would have to meet a higher evidentiary threshold. It would need to identify and prove the existence of several distinct anticompetitive agreements and/or concerted practices as well as identify the parties involved in each of them separately. Treating the impugned practices separately could also in some cases result in a time-bar of older agreements and/or concerted practices. That would make enforcement less efficient.
182. The concept of single and continuous infringement thus constitutes a procedural rule.
183. By alleviating the evidentiary burden of the competition authorities, the concept is of particular relevance in the context of fining. More specifically, the fact that evidence has not been produced in relation to certain specific periods does not preclude the infringement from being regarded as established during a longer overall period. This requires, however, that that finding be supported by objective and consistent indications to that effect. Typically, in the context of an infringement extending over several years, the fact that an agreement is shown to have been applied during different periods, which may be separated by longer or shorter intervals, does not have an impact on the existence of the agreement as such. That is so provided that the various actions which form part of the infringement pursue a single purpose and fall within the framework of a single and continuous infringement. (113) Indeed, of particular significance is the fact that the Commission has been able to prove an overall plan to restrict competition. (114)
184. By contrast, recourse to the concept of a single and continuous infringement does not — and cannot — extend the ambit of the prohibitions under the Treaties.
185. In the present instance, the concept of a single and continuous infringement has been inserted into a wholly different context. (115) In the judgment under appeal, it was employed to find an infringement concerning a single undertaking’s conduct, in relation to which it had not been verified that that conduct alone was capable of restricting competition within the internal market.
186. I must express my doubts as regards that exercise.
187. Fundamentally, as the appellant observes, recourse to the concept of a single and continuous infringement cannot convert lawful conduct into an infringement.
188. However, since the Commission did conclude that a single and continuous infringement had taken place, the General Court considered it to be sufficient to make a global assessment of the part of the market which was foreclosed on average during the period running from 2002 to 2007. (116) On that basis, it considered it irrelevant that the market coverage was considerably smaller during the years 2006 and 2007 than the tied share of the market on average (14%).
189. In other words, the General Court replaced a material criterion with a procedural one. It abandoned the criterion of sufficient market coverage, which it paradoxically held to be relevant for ascertaining whether the impugned conduct was capable of anticompetitive foreclosure, and replaced it with the criterion of a single and continuous infringement. That, quite simply, cannot stand. Either it is accepted that market coverage does not matter at all and that EU competition rules sanction form rather than effects (I have explained above why that is an untenable solution), or it has to be taken seriously in an assessment of all the circumstances.
190. In doing the above, the General Court failed to ascertain that the behaviour in question was capable of restricting competition during the whole period in question.
191. In any event, had it not failed to ascertain whether that was the case, it would have had to conclude that such a small tied market share is inconclusive for the purposes of establishing that the impugned conduct was capable of restricting competition.
192. Similarly to what was noted in point 143 above in relation to a market share of 14%, it cannot be excluded that a tied share of less than 5% of the market may also, in certain circumstances, be sufficient to foreclose competitors. Be that as it may, in the context of an assessment of capability such a market share is simply inconclusive. As explained, it cannot be assumed (on the basis of the form of the conduct) that certain arrangements are caught by the prohibition laid down in Article 102 TFEU without having due regard to the share of the tied market. In circumstances where the tied market share does not provide conclusive evidence of an effect on competition, determining that abuse has occurred requires consideration of the actual or potential effects of the conduct in question.
193. To reiterate, where market coverage is considered inconclusive for the purposes of establishing an effect on competition during a certain period of time, that problem cannot be remedied by applying the concept of a single and continuous infringement. Rather, as that concept itself suggests, for several instances of behaviour to constitute a single and continuous instance of infringement, each instance must also constitute in and of itself an infringement. Put differently, that behaviour must constitute an infringement throughout the whole period in question.
194. Therefore, the second ground of appeal ought to be upheld as well.
D – The third ground of appeal: classification of certain rebates as ‘exclusivity rebates’
1. The main arguments of the parties
195. The appellant, supported by ACT, submits that the General Court erred in law in classifying the rebate arrangements with HP and Lenovo as ‘exclusivity rebates’. Although those rebates covered 95% of HP’s corporate desktops and 80% of Lenovo’s notebooks, they constitute a minority of the CPU purchases of those two undertakings taken as a whole. Intel contends, in essence, that, because the exclusivity requirement attached to those rebates related to certain segments of the requirements for CPUs of those OEMs, the classification of the rebates in question as ‘exclusivity rebates’ is wrong in law. In the appellant’s view, the General Court erred in considering that that has the same effect as the supply of ‘all or most’ of the customer’s total requirements. More specifically, that approach effectively deprives the ‘all or most’ requirement of any rigour: it leads to an unwarranted broadening of the scope of the concept of ‘exclusivity rebates’ that would be automatically condemned under the General Court’s approach to Article 102 TFEU.
196. The Commission contends that this ground of appeal should be dismissed for two reasons. First, it argues that the OEMs’ procurement freedom in certain segments cannot neutralise the restriction on the OEMs’ freedom to choose their source of supply in one segment of the CPU market. Second, the Commission maintains that Intel has misconstrued the relevant case-law of the Court which requires that the dominant undertaking’s competitors must be able to compete on the merits for the entire market.
2. Analysis
197. Like the second ground of appeal, the third ground of appeal is intimately linked to the first. In substance, it raises the issue of whether the General Court was right to consider that the rebates offered by the appellant to HP and Lenovo could be categorised as ‘exclusivity rebates’. (117)
198. I have explained above why there is no separate category of ‘exclusivity rebates’. A presumption of unlawfulness applies to loyalty rebates, including (but not limited to) those termed by the General Court as ‘exclusivity rebates’. One of the possible reasons why a rebate is considered a loyalty rebate is that it is based on the requirement that the customer purchases ‘all or most’ of its requirements from the dominant undertaking. (118) However, form alone does not determine the fate of such rebates. That is because all the circumstances must be considered before it can be concluded that the impugned conduct amounts to an abuse of a dominant position. Therefore, if the Court allows the appellant’s appeal on the first ground as I suggest, it is not necessary to consider the third.
199. However, the third ground remains significant if the Court were to dismiss the first ground and consider that ‘exclusivity rebates’ must be distinguished from other types of loyalty rebates.
200. If the Court were to reach that conclusion, the ‘all or most’ requirement assumes a crucial role in the assessment of those rebates. That is because only rebates which are conditional upon the customer purchasing ‘all or most’ of the customer’s requirements from the dominant undertaking would fall under the umbrella of ‘exclusivity rebates’.
201. Keeping that in mind, I would observe as follows.
202. Regarding HP, for instance, the exclusivity condition related to the requirement that HP purchase 95% of its x86 CPU supplies for corporate desktops from the appellant. That certainly amounts to ‘all or most’ of CPU requirements in that segment. However, the picture is blurred by the fact that those 95% appear to correspond to approximately 28% of HP’s total CPU requirements. (119) As the appellant argues, that can hardly be claimed to constitute ‘all or most’ of those total requirements.
203. In that regard, the General Court held in the judgment under appeal that it is irrelevant whether the condition that the customer purchases ‘all or most’ of its requirements from the dominant undertaking relates to the whole market or a particular segment thereof. (120) To justify that approach, the General Court made reference to Tomra. According to the statement of the Court in that case, competitors should be able to compete on the merits for the entire market and not just a part of it. However, that statement says nothing about how the criterion of ‘all or most’ should be construed. Rather, it relates to whether the foreclosure by a dominant undertaking of a substantial part of the market can nevertheless be justified if the contestable part of the market suffices to accommodate a limited number of competitors.(121)
204. That is not the issue here. The issue here is whether the ‘all or most’ requirement can also relate to a specific part of the relevant product market.
205. In the decision at issue, in defining the relevant product market, no distinction was drawn between CPUs used in corporate computers and those used in consumer computers. That was because, for a specific type of computer, the same CPUs can be used in the business/commercial segment and the private/consumer segment. (122) The possibility of substitution among segments seems to suggest that the market cannot be divided.
206. On that point, the General Court observed in the judgment under appeal that the question whether the CPUs used in the business segment are different from the x86 CPUs used for consumer computers is irrelevant in the present context. In its view, even if those CPUs were interchangeable, the purchasing choice of the OEMs in question would have been restricted considerably in the segment concerned. (123)
207. That argument is at first sight convincing.
208. However, it overlooks an important point: the reasoning in the judgment under appeal takes as a starting point HP’s (and Lenovo’s) viewpoint, rather than that of AMD. From AMD’s perspective, it is wholly irrelevant whether HP’s and Lenovo’s freedom to choose is considerably restricted in one segment or not, bearing in mind that HP and Lenovo are customers of Intel, rather than its competitors.
209. Indeed, a point that must be emphasised is that we are dealing with exclusionary conduct vis-à-vis the appellant’s competitor AMD, not exploitation of the appellant’s customers. What counts, from the perspective of AMD (and hence, for the purposes of determining whether the impugned conduct constitutes an exclusionary abuse of a dominant position contrary to Article 102 TFEU), is the overall percentage of requirements that are tied as a result of Intel’s rebates and payments.
210. As ACT points out, it does not matter whether some requirements are purchased for one particular segment. What matters is whether the OEMs in question can still purchase significant quantities from Intel’s competitors. Here, that appears to be the case: HP and Lenovo could still purchase significant quantities of x86 CPUs from AMD. The question of whether an undertaking has abused its dominant position by excluding a competitor cannot depend on a seemingly arbitrary segmentation of the market.
211. Seen in that light, it seems difficult to argue that, as regards HP, the requirement pertaining to exclusivity on 95% of corporate desktops would amount to anything more than 28% of HP’s overall requirements. By the same logic, exclusivity in Lenovo’s notebooks does not equal exclusivity overall. Quite simply, the ‘all or most’ requirement cannot be satisfied in such circumstances.
212. At the risk of stating the obvious, the approach adopted in the judgment under appeal leads to a result which is hardly warranted: even an ‘exclusivity rebate’ concerning a segment of the relevant market that covers an insignificant part of the customer’s overall requirements (let us say, for the sake of argument, 3%) could be condemned automatically.
213. Therefore, I conclude that the General Court erred in law regarding its classification of the rebates offered by the appellant to HP and Lenovo.
214. Independently of whether the Court agrees with me on the first and second grounds of appeal, the third ground of appeal ought therefore to be upheld.
E – The fourth ground of appeal: rights of the defence
1. The main arguments of the parties
215. The fourth ground of appeal concerns the appellant’s rights of defence embodied in Article 47 of the Charter of Fundamental Rights of the European Union (the ‘Charter’). Intel claims that the General Court erred in law in concluding that no procedural error had taken place regarding a meeting held with a Dell executive, Mr D1, in 2006 during the investigation leading to the adoption of the decision at issue (‘the meeting in question’).
216. In that regard, the appellant claims, first, that the General Court erred in holding that the meeting in question was not an interview within the meaning of Article 19 of Regulation No 1/2003. Second, it contends that the General Court erred in considering that although the Commission ought to have recorded the meeting in question due to its importance, the breach of the principle of good administration that that failure entailed was cured by the inclusion of a non-confidential version of the note to the file (the Commission’s internal aide mémoire; ‘the note to the file’) to which the appellant was given access. Third, it argues that the General Court erred in its assessment in the alternative as to whether a procedural error of the kind identified in the judgment under appeal constitutes a basis for the annulment of the decision at issue so far as concerns the appellant’s conduct vis-à-vis Dell.
217. As its primary contention, the Commission argues that the fourth ground of appeal is ineffective since Intel has not contested the finding in the judgment under appeal that the rebates granted to Dell were ‘exclusivity rebates’. That ground is, in the Commission’s view, also inadmissible, as the question of whether the breach of the principle of good administration could be cured by giving Intel access to the non-confidential version of the note to the file depends on an assessment of the importance of the meeting in question and of the adequacy of the note that was made. Those are issues of fact, not amenable to review on appeal.
218. In the alternative, Intel’s arguments are unfounded: Intel has not put forward any relevant argument that would call into question the assessment made in the judgment under appeal concerning the note to the file. The Commission also maintains that the decision at issue was primarily based on documentary evidence on which the meeting in question could under no circumstances have shed a different light.
2. Analysis
219. From the outset, it is necessary to emphasise that the fourth ground of appeal is by no means ineffective, or, for that matter, inadmissible, as argued by the Commission.
220. In this ground of appeal, Intel specifically contends that the finding of an infringement must be annulled in relation to Dell because the findings of fact on which that infringement is based are vitiated by a breach of its rights of defence. That is a matter of law on which the Court can and should adjudicate. The classification in the judgment under appeal that Intel’s rebates to Dell were conditioned on exclusivity, whether contested or not, is irrelevant in that regard. I have explained above that a finding that the rebates at issue (however they are ‘labelled’) are unlawful cannot be made without looking into all relevant circumstances. The Commission itself admitted, at the hearing, that, in principle, even ‘exclusivity rebates’ could be justified by the undertaking in question. Equally, the fact that the parties agree that information obtained during the meeting in question was not relied on by the Commission to inculpate Intel is beside the point: that has no bearing on the possible exculpatory value of the meeting. (124) More fundamentally still, the question of whether a breach of the rights of the defence has taken place is entirely independent of whether that (potential) breach had an impact on the material content of the decision at issue.
221. Quite simply, it matters little, if at all, how Intel’s rebates were classified or what evidence was employed to inculpate the appellant if its rights of defence have been breached. The only thing on which the Court needs to satisfy itself is whether the appellant has shown that it would have been better able to ensure its defence if it had had access to a record of the meeting in question. To address that issue, the Court must also consider, inter alia, whether the note to the file — which was disclosed to the appellant belatedly only at first instance — might have ‘cured’ any prior procedural irregularity arising from the Commission’s decision not to record the meeting in question. That is why I am not convinced by the Commission’s argument that the appellant’s arguments concerning the adequacy of the note to the file contest, in reality, findings of fact.
222. As I will explain in the following, the fourth ground should, in fact, be upheld.
(a) The meeting in question was an interview within the meaning of Article 19 of Regulation No 1/2003
223. The appellant claims that the General Court erred in law by concluding that there was no breach of Article 19 of Regulation No 1/2003, read in conjunction with Article 3 of Regulation No 773/2004. (125) In that regard, Intel argues that the distinction operated in the judgment under appeal between ‘formal’ and ‘informal’ interviews is wrong in law. The same is in its view true for the finding that the Commission is not under an obligation to record ‘informal’ interviews. (126)
224. Before moving on to consider that distinction, it is useful to recall briefly the (procedural) steps that led to the disclosure to the appellant of the note to the file concerning the meeting in question.
225. It is apparent from the judgment under appeal that, during the administrative procedure, the Commission first denied that a meeting with Mr D1 had taken place. It admitted that a meeting had taken place only after Intel had demonstrated the existence of an indicative list of topics concerning the meeting in question. At that point in time, the Commission continued to deny that a record had been made of it. Some months later, however, the Hearing Officer acknowledged that a note to the file existed but pointed out that the appellant did not have a right of access to that note. Nevertheless, the Commission sent a copy of the non-confidential version of the note to the file to Intel in December 2008 ‘as a matter of courtesy’. That copy was heavily redacted. After a request was made to that effect by the General Court, the confidential version of that note was finally communicated to the appellant during the proceedings before the General Court in January 2013. (127)
226. Now, turning to the interpretation of Article 19 of Regulation No 1/2003, the power to conduct interviews constitutes a logical corollary to the wide investigative powers granted to the Commission under Regulation No 1/2003. The question that arises here is whether there are, nevertheless, limits to those powers.
227. Those limits can be clearly identified on the basis of the wording of the relevant provisions. To begin with, Article 19 of Regulation No 1/2003 provides that the Commission can interview any person (natural or legal) who consents to being interviewed for the purposes of collecting information relating to the subject matter of an investigation. While Article 3(1) of Regulation No 773/2004 embodies a legal obligation to record interviews, Article 3(3) thereof provides that the Commission may choose how to record the statements made by the persons interviewed.
228. In that light, it seems to me rather obvious that, where the Commission decides to conduct an interview, it cannot omit to record the substance of that interview. By contrast, it remains free to choose how (the medium to be employed) to record it.
229. That is not, as such, contradicted by the judgment under appeal. (128)
230. Rather, the problem lies in the fact that the General Court distinguished, in the judgment under appeal, between informal and formal interviews. No such distinction exists in the legislative framework laid down by Regulation No 1/2003.
231. That distinction is in my view highly problematic. Devising, by way of judicial construction, a new tool for the Commission to conduct its investigations would allow it to circumvent the rules that the legislator has put in place specifically to regulate the powers granted to that institution in the context of investigations relating to infringements of competition rules.
232. One of those rules is, as follows clearly from Article 19 of Regulation No 1/2003, read in conjunction with Article 3 of Regulation No 773/2004, that information gathered in interviews that relate to the subject matter of an investigation is to be recorded. As I see it, any meeting with a third party which is specifically arranged to collect substantive information to be employed in the assessment of a case must fall within the scope of Article 19 of Regulation No 1/2003.
233. By contrast, that does not imply that the Commission could never contact third parties informally. As is clear from the very wording of Article 19 of Regulation No 1/2003, only exchanges relating to the subject matter of an investigation fall within the scope of that provision. Where exchanges between the Commission and third parties do not relate to the subject matter of a particular (usually ongoing) investigation, no obligation to record such exchanges exists.
234. In the present case, however, I cannot see how the meeting in question could be conceived as anything else than an interview within the meaning of Article 19 of Regulation No 1/2003.
235. The meeting was not only linked to the subject matter of the Commission’s ongoing investigation into Intel’s practices. As the note to the file indicates, the topics addressed during the meeting, which appears to have lasted five hours, went to the very heart of the matter investigated (namely whether Intel’s rebates to Dell were conditional on exclusivity). More importantly still, the person interviewed was one of Dell’s most senior executives. (129)
236. In that respect, it is irrelevant whether the purpose of the meeting was to collect evidence in the form of countersigned minutes or statements or, as maintained by the Commission, was not. (130)
237. If it were accepted that only such contacts with third parties fall within the purview of Article 19 of Regulation No 1/2003, it would considerably broaden the Commission’s discretion to conduct interviews without any obligation to record them. It would also enable the Commission to be selective in terms of the evidence to be disclosed to undertakings suspected of having infringed EU competition rules: the Commission staff convening the person to be interviewed or the staff attending that meeting could, on the basis of their subjective views, decide what became part of the file and what did not.
238. However, that is not how the right of ‘access to the file’ was conceived by the EU legislature. Disclosure of all evidence constitutes the rule, and non-disclosure of specific pieces of evidence the exception, as indicated by Article 27(2) of Regulation No 1/2003. The interpretation of Article 19 proposed by the Commission would risk rendering Article 27(2) ineffective.
239. At the hearing, the Commission was at pains to explain which contacts with third parties it must record and which ones it need not. Strikingly, in attempting to explain its position, the Commission seemed to suggest that it could decide to have recourse to Article 19 of Regulation No 1/2003 entirely at its own discretion. The fact that the Commission was unable to provide a clear answer to the Court on this point is understandable: it appears very difficult to identify a criterion apt to distinguish between formal and informal interviews which differs from that enshrined in the law, namely whether the interview relates to the subject matter of the investigation.
240. Of equal importance is the fact that the decision whether to record an interview would also escape any possible judicial review. Since no written record exists, how are the EU Courts to check whether the Commission complied with the provisions of Regulation No 1/2003 and, more generally, whether the rights of the undertakings and natural persons involved in an investigation were fully respected?
241. Indeed, in the final analysis, the requirement laid down in Article 3 of Regulation No 773/2004 that interviews are to be recorded exists for at least two interlocking reasons. That requirement ensures, on the one hand, that the undertakings suspected of infringing EU competition rules can organise their defence and, on the other hand, that the EU Courts can review, ex post, whether the Commission employed its investigative powers within the confines of the law.
242. For those reasons, it is my firm conviction that the General Court erred in law in holding that the Commission had not infringed Article 19 of Regulation No 1/2003 by not organising the meeting in question as an interview within the meaning of that provision and by failing to record it adequately.
(b) The note to the file did not cure the procedural error
243. As mentioned above (point 216), the General Court did not find an infringement of Article 19 of Regulation No 1/2003 in the judgment under appeal. However, it found that — considering the content and importance of the information received during the meeting in question — the Commission ought to have recorded it. That amounted in the General Court’s view to a breach of the principle of good administration. In that regard, the General Court considered that in the circumstances of the present case, at least a succinct note containing the names of the participants as well as a brief summary of the topics addressed should have been placed in the file. The appellant could then have requested access to that document. (131)
244. Nonetheless, in the General Court’s view, that procedural irregularity was cured by the fact that, during the administrative procedure, a non-confidential version of the note to the file was made available to Intel and that it was offered the opportunity to submit its observations on that document. That note, which was intended as an internal summary of the topics discussed for the members of the Commission’s services dealing with the case, contained the names of the participants and ‘a brief summary of the subjects addressed’. (132)
245. The appellant argues that this was an error in law not only because the Commission was required to record the substance of the meeting in question, but also because the note did not, contrary to what was held by the General Court, record a ‘brief summary of the subjects addressed’.
246. I agree.
247. As a matter of principle, a note of the kind described in the judgment under appeal cannot, under any circumstances, remedy a breach of an essential procedural requirement. Crucially, as the General Court admitted in the judgment under appeal, that note amounts to a succinct summary of the topics discussed during that meeting. (133) By contrast, it does not spell out the substance of the interview. The Commission itself acknowledges that fact. Most fundamentally, however, that note is silent on the content of the information provided by Mr D1 during the meeting about the issues referred to therein.
248. In my view, such a note cannot cure a breach of Article 19 of Regulation No 1/2003, read in conjunction with Article 3 of Regulation No 773/2004.
249. It cannot be overemphasised that the information contained in the file about an interview must be sufficient to ensure that the rights of defence of the undertakings accused of infringing EU competition rules are respected. That is manifestly not the case here. I shall elaborate further on that issue in point 257 et seq. below.
250. Consequently the question arises whether the procedural error resulting from an infringement of Article 19 of Regulation No 1/2003, read in conjunction with Article 3 of Regulation No 773/2004, may lead to the illegality of the decision at issue in relation to the findings made concerning Dell. Contrary to what the General Court ruled, (134) the appellant thinks it should. ACT also shares that view. True, the appellant’s arguments relate to reasoning that is included in the judgment under appeal for the sake of completeness. It could therefore be argued that those arguments are ineffective and cannot lead to that judgment being set aside. (135) However, to the extent that the Court agrees with me that the General Court erred in holding that (1) the meeting in question was not an interview within the meaning of Article 19 of Regulation No 1/2003 and that (2) the note to the file cured any procedural error stemming from the Commission’s decision not to record that meeting, the Court must also consider the reasoning relating to the consequences of a possible procedural irregularity in the judgment under appeal.
(c) The consequence of the failure to record the meeting in question
251. According to the judgment under appeal, the present situation can be distinguished from that underlying Solvay, (136) a case on which the appellant primarily relies. In that case, the Commission had lost several documents after the close of the administrative procedure. The undertaking in question had not had access to those documents during the procedure before the Commission. In those circumstances, the Court ruled that a procedural error of that kind justifies the annulment of the Commission decision. The yardstick for such an annulment was expressed in the following terms: a procedural error constitutes a basis for annulment where it cannot be excluded that the (lost) material would have enabled the undertaking concerned to offer an interpretation of the facts different from that adopted by the Commission that could have been useful for its defence. (137)
252. In the General Court’s view, however, the statement of the Court in Solvay could not be transposed to the present set of circumstances. This was so because, unlike in Solvay, the content of the meeting in question could be reconstituted. (138) That is why, pursuant to the case-law dealing with access to the file, (139) the General Court required Intel to adduce prima facie evidence that the Commission had ‘failed to record exculpatory evidence which was at variance with the thrust of the direct documentary evidence on which the Commission relied in the [decision at issue] or, at the very least, sheds different light on it’. A mere hypothesis concerning the relevance of the information provided during the meeting in question was not regarded as sufficient. (140)
253. Indeed, it is settled law that where access to a part of the file has been withheld during the administrative procedure, but access has nevertheless been granted during judicial proceedings, the relevant yardstick is, as a matter of principle, whether the information withheld could have been useful, in one way or another, for the undertaking’s defence. It is not required that that information would have led to a decision different in content. (141) Rather, it must be shown that the undertaking would have been better able to ensure its defence had there been no error. (142)
254. That rule only applies, however, where no direct documentary evidence has been used as evidence. Where the Commission has relied on direct documentary evidence in the contested decision, the undertaking concerned must show that the Commission has failed to record exculpatory evidence which is at variance with the thrust of the direct documentary evidence, or at least sheds a different light on it. (143) In other words, in so far as direct documentary evidence has been employed by the Commission to inculpate the undertaking concerned, the burden of proof is particularly difficult to satisfy.
255. The question of whether such an approach is, generally speaking, warranted is beyond the scope of this Opinion. Yet, placing such a requirement on the appellant in the present case is in my view clearly wrong in law. That is because the burden of proof thus placed on the undertaking concerned is quite simply impossible to fulfil. The correct approach is to ask, as the Court’s statement in Solvay requires, whether it can from the outset be excluded that the information to which the undertaking concerned has not had access could have been useful for that undertaking’s defence.
256. In the present case, that question must be answered in the negative.
257. In Solvay, there was no conceivable way of reconstructing the content of the missing files from other sources. Moreover, the Commission had itself admitted that the missing files in all likelihood contained relevant information for the undertaking’s defence (more specifically replies to requests for information). (144)
258. In the present case, the meeting in question was not recorded adequately, as I have explained above. Nevertheless, the appellant was given access to the non-confidential version of the note to the file and the so-called follow-up document during the administrative procedure. That document contained Dell’s written responses to the questions put to Mr D1 during the meeting in question. Later, during the proceedings before the General Court, access to the confidential version of the note was given to the appellant. Those two documents provided — in the General Court’s view — sufficient indication of what was discussed during the meeting. On the basis of those documents, it concluded that the meeting disclosed no new exculpatory evidence that the appellant could have employed in its defence. (145)
259. However, the information that can be deduced from those documents regarding what took place during the meeting in question remains mere speculation, as the judgment under appeal amply shows. (146) As the analysis of the information available in the judgment under appeal illustrates, where no adequate record of the meeting exists, it is not possible to tell with certainty what was discussed and to what extent that might have been exculpatory, inculpatory, or indeed neutral. (147)
260. Judicial review cannot be based on assumptions about evidence.
261. It is certainly true, as the Commission points out, that an analysis of whether a breach of the rights of the defence ought to lead to the annulment of the Commission’s decision begins with the objections raised against the undertaking in question and the evidence which had been disclosed in support of those objections. (148) Otherwise, it would always be possible to argue that information not contained in the file could have been useful for the undertaking concerned. (149)
262. Bearing in mind the Commission’s objections vis-à-vis Intel in the case at hand, there is little doubt as to the relevance of the meeting in question. Indeed, as noted by the General Court, the note to the file as well as the follow-up document show that issues relevant for establishing whether Dell received anticompetitive loyalty rebates from the appellant were discussed during the meeting. (150)
263. In situations like those, the burden of proof remains — generally speaking — on the undertaking concerned. (151) As the Commission observes, the undertaking must present the facts and adduce the evidence to prove that it could have used in its defence documents to which it was denied access during the administrative procedure. However, that is so where documents have been withheld during the administrative procedure and where the content of those documents could later be established and reviewed by the Court. (152) As Advocate General Kokott observed in Solvay, the reason for this is that the undertaking concerned can in such circumstances indicate the authors and the nature of the document that has been withheld from it. But not only. More fundamentally still, in such circumstances, the undertaking concerned is also able to describe the content of those documents. (153)
264. The situation here is different. The identity of the author and the nature of the meeting are known on the basis of the note to the file. Nonetheless, the content of the answers given by Mr D1 to the questions put to him by the Commission remains obscure. Admittedly, as the General Court observed, the note to the file as well as the follow-up document shed some light on the specific topics touched upon during the meeting in question. However, those documents are not sufficient to reconstruct ex post the evidence given, namely what was actually said during that meeting.
265. Although the judgment under appeal does not address the issue explicitly, it is only possible to come to the opposite conclusion if one assumes that Mr D1 and Dell are one and the same and that he could only have reiterated Dell’s position on the topics discussed. Account taken of his status as a senior executive within Dell, that assumption may certainly be correct.
266. However, that assumption may also be incorrect.
267. Contrary to what the Commission appeared to suggest at the hearing, it is just as likely that Mr D1 expressed his personal view on the topics addressed during the meeting in question. (154) We simply do not know. That is why it cannot be ruled out that the meeting shed a different, or indeed new, light on the conditionality of the rebates offered to Dell. Instead of acknowledging that possibility, the General Court placed on the appellant the arguably impossible task of proving that the unrecorded meeting disclosed exculpatory evidence that could have shed a different light on the evidence provided by the Commission in support of its objections. For obvious reasons, it concluded that the appellant had not been able to succeed in that task.
268. On that basis, I must conclude that the fourth ground of appeal ought therefore also to be upheld.
269. If the Court were to disagree with me, I would nevertheless caution against rejecting the fourth ground of appeal, for the following reasons.
270. Let us assume, for the sake of argument, that the evidence in question could be reconstituted to the requisite legal standard ex post, as the General Court held in the judgment under appeal. According to the General Court, the appellant had therefore to show that the evidence in question could call into question the ‘direct documentary evidence’ that had already been determined as sufficient to condemn Intel for an abuse of a dominant position regarding the rebates offered to Dell. (155) That approach rests on a fallacy. It wrongly assumes that the evidence that has been withheld during the administrative procedure must necessarily possess a weaker probative value than the evidence that the Commission presented in support of its finding of abuse. More specifically, the problem stems from the excessively broad interpretation of the concept of ‘direct documentary evidence’ in the judgment under appeal.
271. To my knowledge, that concept has not been defined explicitly by the Court. Nonetheless, the case-law provides helpful indication as to its scope.
272. Generally speaking, the concept of direct documentary evidence is employed in the case-law in the context of Article 101 TFEU to describe certain types of evidence (as opposed to, for example, circumstantial or economic evidence) that the Commission may use to establish that an infringement has been committed, for instance that specific undertakings have taken part in a cartel or a related practice contrary to Article 101 TFEU. (156)
273. Unlike circumstantial evidence, (157) direct documentary evidence emanates, as a rule, from the undertaking(s) suspected of having infringed EU competition rules and Article 101 TFEU in particular. Typically, such evidence takes the form of a document that in and of itself points to the existence of a cartel or a related practice (or the participation of specific undertakings in such a practice). That would be the case of, for example, a memorandum on an agreement between participants, exchange of emails between participants concerning pricing or, indeed, minutes of meetings concerning such practices. (158) Where the Commission has relied on such evidence to establish an infringement or the participation of undertakings in an infringement, the undertakings must — for the purposes of annulment of the decision in question — prove that the evidence that was inaccessible to them during the administrative procedure was at variance with the thrust of the direct documentary evidence presented. (159)
274. The evidence relied upon by the Commission in the decision at issue to establish the conditionality of the rebates granted to Dell can, at best, be described as circumstantial or presumptive. (160) In fact, a point that should not be overlooked is that the ‘exclusivity rebates’ at issue in the present case (including those granted to Dell) were considered to be de facto conditional on exclusivity. That is because the rebates were not based on a formal exclusive supply obligation. (161) Rather, the conditionality of the rebates offered to Dell was inferred (indirectly) from the level of the rebates. (162) Special relevance was also given to Dell’s impression of the risks involved in switching part of its supplies to a competitor. (163) At the risk of stating the obvious, such evidence can hardly be described as ‘direct documentary evidence’ regarding the conditionality of the rebates in question.
275. In the absence of any written document attesting to the existence of an exclusive supply obligation, accepting any piece of written evidence as ‘direct documentary evidence’ of an abuse of a dominant position contrary to Article 102 TFEU would in my view gravely undermine the rights of defence of the undertaking concerned: it would not suffice for an undertaking to show that a piece of evidence to which it had not been given access during the administrative procedure could have been useful for its defence. In addition, that undertaking would have to show (as the General Court required in the judgment under appeal) that the withheld evidence was at variance with the thrust of the evidence presented by the Commission to support its finding of abuse.
276. Bearing that in mind, I am firmly convinced that circumstantial evidence of the kind referred to in the decision at issue must be assessed as a whole (before it can be decided whether the bulk of evidence presented suffices to prove an abuse of dominance). For the purposes of annulment of the decision at issue, the undertaking concerned must in such circumstances simply show that it could have used, in one way or another, the withheld evidence for its defence, and not that that evidence is at variance with the thrust of the evidence presented by the Commission to establish the infringement. (164)
277. Therefore, I conclude that, even under this alternative, the fourth ground of appeal ought to be upheld.
F – The fifth ground of appeal: jurisdiction
1. The main arguments of the parties
278. By its fifth ground of appeal, Intel, supported by ACT, claims that the General Court erred in law in holding that the Commission had jurisdiction to apply Article 102 TFEU to Intel’s 2006 and 2007 agreements with Lenovo (respectively ‘the 2006 agreement’ and ‘the 2007 agreement’ or jointly ‘the Lenovo agreements’). On the one hand, the 2006 agreement encouraged, by way of the grant of a financial incentive, Lenovo to postpone (and finally cancel) the launch of two AMD-based products on the worldwide market. (165) On the other hand, the 2007 agreement related to the rebates that Intel was to offer if Lenovo decided to source exclusively from Intel CPUs for its notebook. (166) The appellant argues that the naked restrictions and rebates, as regards Lenovo, were neither implemented in the EEA, nor had they any foreseeable, immediate or substantial effect in that area.
279. The Commission contends that the fifth ground is unfounded: the General Court did not err in holding that the Commission had the power to apply Article 102 TFEU with regard to the Lenovo agreements. The Commission argues that, under public international law, jurisdiction can be based on several factors, provided that a sufficient link exists between the behaviour complained of and the applicable rules of the territory concerned. In that regard, the implementation criterion and the ‘qualified’ effects criterion are only two possible ways of establishing such a link. In applying those criteria — the Commission contends — the judgment under appeal discloses no error in law.
2. Analysis
280. The present ground is by no means of lesser importance than those examined so far. It provides the Court with the welcome opportunity of clarifying the line of case-law devolving from ICI, and subsequently developed in Wood Pulp, (167) concerning the territorial application of EU competition law. It will enable the Court to fine-tune that line of case-law and adjust it to present-day conditions, characterised by global economies, integrated marketplaces and elaborate patterns of trade.
281. Here it is important to bear in mind the broader ramifications that the Court’s ruling is likely to have. Indeed, an overly generous interpretation of the rules on territorial jurisdiction is not uncontroversial from the viewpoint of public international law, with which EU law is to be interpreted in conformity. (168) It is therefore worth framing this ground of appeal in a broader context.
282. Generally speaking, jurisdiction takes (at least) three different forms: prescriptive jurisdiction, jurisdiction to enforce, and adjudicative (or ‘curial’) jurisdiction. Intel calls into question the Commission’s jurisdiction to apply EU competition law to unilateral conduct stemming from agreements arguably producing their effects outside the European Union. Therefore the present proceedings do not concern physical enforcement outside EU territory, an issue that raises a wide array of difficulties from the perspective of public international law.
283. I would also remark that public international law allows States to exercise jurisdiction extraterritorially in certain instances. However, although admittedly not binding per se, (169) mutual regard to the spheres of jurisdiction both of the European Union and of concerned third States, (170) or comity, suggests that restraint is called for in asserting extraterritorial jurisdiction. Unsurprisingly, the European Union itself opposes the extraterritorial application of the laws of third States when it considers doing so to be unlawful. (171)
284. That having been said, a survey of the case-law of the Court reveals that the application of EU law presupposes an adequate link to the EU territory. (172) That way, the basic principle of territoriality under public international law is observed. Still, it is not unusual for a State or an international organisation also to take into account, in the exercise of its sovereignty, circumstances that occur or have occurred outside its territorial jurisdiction. (173)
285. It follows from the existing case-law of the Court that EU competition law operates with a requirement that there be an adequate link to the EU territory, be it in the form of the presence of a subsidiary, or the implementation of anticompetitive conduct within that territory. However, in earlier cases, that link was much more discernable than in the matter currently under consideration.
286. In the present case the General Court considered that two alternative criteria may be applied to assert jurisdiction: the criterion of implementation and that of ‘qualified’ effects of the practices within the EEA. (174) In its view, the application of those criteria led to the same conclusion: the Commission had jurisdiction over the Lenovo agreements. (175)
287. In the following, I shall first explain my position on the issue of jurisdiction with regard to the public enforcement of EU competition rules. (176) I shall then illustrate why I consider this ground of appeal to be well founded.
(a) General remarks: implementation and/or effects?
288. The first point I would make is a simple and self-evident one. To determine whether the Commission can apply EU competition rules to specific conduct, the starting point must be the wording of Articles 101 and 102 TFEU. Far from affording the Commission carte blanche to apply EU competition law to behaviour wherever it occurs and no matter whether it has any clear link to the EU territory, those provisions are concerned with collective or unilateral anticompetitive conduct within the internal market: Article 101 TFEU prohibits agreements or practices ‘which have as their object or effect the prevention, restriction or distortion of competition within the internal market’; and Article 102 TFEU, for its part, prohibits ‘any abuse … within the internal market’.
289. The jurisdictional rule for the application of EU competition rules is thus clearly embedded in those provisions. Although Article 102 TFEU is somewhat less clear, Article 101 TFEU is very explicit in that it applies to any conduct that has anticompetitive effects on the internal market.
290. Moreover, like the Commission, I do not read the judgment of the Court in Woodpulp as implying that implementation is the only valid jurisdictional criterion. I am rather of the view that when anticompetitive conduct is implemented in the European Union, the applicability of Articles 101 and 102 TFEU is beyond doubt. In other words, it cannot be disputed that conduct implemented in the European Union may have effects within the internal market and, consequently, cannot escape review under EU competition rules. In that context, a point that should not be overlooked is that the criterion of implementation is firmly rooted in the principle of territoriality and, accordingly, if met, it is a decisive factor for establishing the Commission’s jurisdiction to apply those rules to a specific conduct. (177)
291. The fact that only part of the relevant conduct takes place in the European Union is, in that regard, immaterial. (178) In Woodpulp, the Court dealt with a series of practices to fix wood pulp prices — which the Commission had considered to be contrary to what is now Article 101 TFEU — that had been adopted outside (what is now) the European Union by foreign wood pulp producers. In that context, the Court explained why the implementation, rather than the conclusion or formation, of an agreement or related practice was relevant to establish jurisdiction. If the prohibitions laid down in the Treaties were applied only where the agreement, decision or concerted practice was formed or adopted within the EU territory, that would provide undertakings with an easy way to fend off the application of EU competition rules. In that case, the criterion of implementation was met through the direct sale of the cartelised products: the undertakings concerned had sold wood pulp directly to purchasers in the European Union. (179)
292. Unlike Intel, however, I do not consider that only direct sales into the European Union by the undertaking in question can be regarded as meeting the implementation criterion for the purposes of the Woodpulp case-law. The ordinary meaning of ‘implementation’ is to carry out or to put into effect. Therefore, in order to meet that criterion, one of the essential constituent elements of the anticompetitive conduct must take place in the European Union. Whether that is so depends mainly on the nature, form and scope of the conduct in question. A case-by-case assessment of the unlawful conduct is required to verify whether that conduct is implemented within the European Union. For example, I am not convinced that indirect sales of the relevant product can never be regarded as implementation. (180) To my mind, that depends on the circumstances of each case. Among the elements which should be considered in that context is, for instance, whether one of the undertakings that cartelised a product and the undertaking which incorporates it into another product which is then sold in the Union form part of a single economic entity or, failing that, whether there are other corporate or structural links between the undertakings in question.
293. To conclude on this point, collective or unilateral conduct is implemented within the internal market — and thus unquestionably triggers the application of Articles 101 and 102 TFEU — when there is an element of intra-territorial conduct. (181) In other words, when part of the unlawful conduct is executed, applied or put into effect within the internal market because one of its essential constituent elements takes place there.
294. However, were implementation to be considered the only jurisdictional criterion triggering the application of EU competition rules, various types of conduct that may well have the object or effect of preventing, restricting or distorting competition within the internal market would fall beyond the reach of those rules. Here I have in mind conduct that is characterised by an unlawful omission, such as refusal to deal or boycotts. As mentioned in points 288 and 289 above, such an interpretation of Articles 101 and 102 TFEU would be contrary to the wording of those provisions.
295. In fact, several Advocates General have already advised the Court to adopt an effects-based approach to jurisdiction in the field of competition law. (182) The Court has not, to date, either endorsed or expressly rejected that approach. (183)
296. Against that background, I believe that the Court should explicitly address that issue here and, in line with what has been suggested by the Advocates General mentioned in the previous point, adopt an effects-based approach to the application of Articles 101 and 102 TFEU.
297. Whether such an approach is anchored in a (broad) concept of territoriality, or instead involves some extra territorial application of the EU rules is not determinative. (184) What is crucial is that, under certain conditions, effects is a jurisdictional criterion which, as concerns this kind of legislation, is generally acceptable under the rules of public international law (185) and has been embraced by many jurisdictions worldwide. (186) Indeed, many legal scholars take the view that any controversy regarding its acceptability is something that, by now, belongs to the past. (187)
298. In that context, it is worth noting that several other provisions of EU law regulate the foreign conduct of entities which are neither nationals of an EU Member State nor physically or legally present in the Union, because of the effect produced by that conduct in the internal market. That is the case, for example, of a number of provisions governing transactions in financial instruments or other types of economic conduct. (188)
299. That does not mean, however, that any effect, no matter how weak or indirect, could trigger the application of EU competition rules. In a globalised economy, conduct that takes place anywhere in the world, for example in China, will almost inevitably have some sort of effect in the European Union. Yet, the application of Articles 101 and 102 TFEU cannot be based on a link or effect that is too remote or purely hypothetical.
300. I consider it to be particularly important that jurisdiction is asserted with restraint in relation to behaviour that has not, strictly speaking, taken place within the territory of the European Union. Indeed, to comply with a certain form of comity and, by the same token, to ensure that undertakings can operate in a foreseeable legal environment, it is only with a great deal of caution that the effect of the conduct complained of can be used as the yardstick for asserting jurisdiction. That is all the more important today. There are over 100 national or supranational authorities worldwide that claim jurisdiction over anticompetitive practices.
301. As the General Court held in Gencor, the application of EU competition rules to some specific conduct could be justified only when that conduct has foreseeable, immediate and substantial effects in the internal market. (189) Here obvious parallels can be drawn to the competition rules applicable in the United States of America (‘the US’): Section 1 of the Sherman Act lays down a general prohibition on restraints of trade without any geographical limitations. That is why, in 1982, US Congress enacted the Foreign Trade Antitrust Improvement Act (‘FTAIA’), (190) to clarify (and possibly restrict) the extraterritorial application of the Sherman Act. Inter alia, the FTAIA states that, in essence, US antitrust rules do not apply to foreign conduct unless such conduct has a direct, substantial, and reasonably foreseeable effect in the US.In Empagran, the US Supreme Court — interpreting the Sherman Act and the FTAIA — found that it was unreasonable to apply US laws to foreign conduct where the resulting foreign injury complained of was independent of any domestic injury. (191)
302. Similar principles should guide this Court in interpreting and applying Articles 101 and 102 TFEU to undertakings’ collective or unilateral conduct that takes place entirely outside the Union’s borders. As I see it, such conduct is caught by those provisions only to the extent that a direct (or immediate), substantial and foreseeable anticompetitive effect within the internal market can be detected. That criterion of ‘qualified’ effects (meaning, as I understand it, that the effects are sufficiently significant to justify asserting jurisdiction), is not satisfied where, for example, the effect in the European Union is merely hypothetical or, in any event, of minor significance. It is also not satisfied where the distortion of competition within the internal market cannot be imputed to the undertaking in question, since those harmful effects were not foreseeable to it.
303. The wording of Articles 101 and 102 TFEU does not justify the application of EU rules by the Commission with respect to conduct that has no ‘qualified’ effect in the territory of the European Union. To hold the contrary would also be problematic under the rules of public international law. An over-reach of EU competition rules would risk encroaching upon the sovereign interests of other States and be legally and practically difficult to enforce. (192) It would also considerably increase the overlaps in the jurisdictions of different States or polities and thereby create great uncertainty for undertakings and increased risks of conflicting rules (or judgments) applying to the same conduct. Lastly, but no less importantly, it may raise questions under the principle of good administration: what would be the interest in enforcing EU rules with respect to conduct that has no significant effect in the European Union? Would that be a valid and effective use of the European Union’s limited resources?
304. In the light of the above, I take the view that the General Court cannot be criticised, as Intel argues, for having examined the Commission’s jurisdiction to apply Article 102 TFEU under both the implementation and the ‘qualified’ effects criteria. To be sure, it would have been more logical to first assess whether Intel’s conduct was implemented in the Union and, if not, only then whether that conduct was nonetheless producing ‘qualified’ effects in the internal market.
305. However, the fact that Intel did not question the Commission’s jurisdiction during the administrative procedure — which the General Court emphasises in paragraph 246 of the judgment under appeal — is irrelevant. As the Court has consistently stated, the scope of judicial review provided for in Article 263 TFEU extends to all the elements of Commission decisions relating to proceedings applying Articles 101 and 102 TFEU which are subject to in-depth review by the General Court, in law and in fact, in the light of the pleas raised by the appellants, and taking into account all the elements submitted by the latter, whether those elements pre-date or post-date the contested decision, whether they were submitted previously in the context of the administrative procedure or, for the first time, in the context of the proceedings before the General Court, in so far as those elements are relevant to the review of the legality of the Commission decision. (193)
306. By way of conclusion, the legal framework applied by the General Court cannot be criticised. Nonetheless the application of those jurisdictional criteria to the alleged abuses originating from the Lenovo agreements in the judgment under appeal calls for the following critical remarks on my part.
(b) Assessment of the application of the relevant jurisdictional criteria by the General Court
307. I will begin by examining the General Court’s findings as regards the implementation in the EEA of the naked restrictions and exclusivity rebates originating from the Lenovo agreements.
i) Implementation
308. In the judgment under appeal, (194) the General Court found that the Lenovo agreements were intended to be implemented worldwide by Lenovo, including in the EEA. In the light of those agreements, Intel could not claim that it had no influence on the use made by Lenovo of the Intel CPUs. Intel was also aware that Lenovo was present in the internal market and sold notebooks in that market.
309. I consider this reasoning to be vitiated by an error in law. Had the Commission concluded that Intel together with Lenovo had infringed Article 101 TFEU, it would have been correct for the General Court to look at whether their agreements were intended to be implemented by either party in the EEA. However, the decision at issue deals with conduct which the Commission challenged under Article 102 TFEU: unilateral conduct on the part of Intel. It is consequently that unilateral conduct — the alleged abuse — that must be implemented in the EEA.
310. Yet, nowhere does the judgment under appeal refer to conduct initiated or carried out by Intel in the EEA territory to implement what had been agreed in the Lenovo agreements. That is not surprising. Those agreements, entered into by a US company and a Chinese company, concerned sales of CPUs manufactured and sold outside the Union, for incorporation into computers manufactured in China. They only limited the possibility for AMD, another US-based company, of selling CPUs on the Chinese market.
311. Instead of focusing on a possible implementation by Intel, the General Court concentrated on customer behaviour in a downstream market in order to establish a link to the EEA territory. The mere fact that Lenovo refrained, for a certain period of time, from selling a certain computer model on a worldwide level, possibly including the EEA, constituted for the General Court an instance of implementation of Intel’s abuses.
312. That reasoning fails to convince. By tying implementation to the behaviour of the customer of the undertaking accused of having breached Article 102 TFEU, almost any conduct — no matter how remotely related to the EU territory — could be construed as falling under the Commission’s jurisdiction on the basis of the criterion of implementation. The other elements taken into consideration by the General Court are equally unpersuasive. First, the mere fact that Intel had influence on the use made by Lenovo of Intel CPUs does not seem to me to have any bearing in that regard. Had Lenovo had some corporate or structural links to Intel, the conclusion could have been different. Second, the fact that Intel was aware that Lenovo was present in the internal market and sold notebooks in that market is, to my mind, also of little relevance. I must emphasise once more that the unlawful conduct does not relate to the sale of notebooks: it relates to the exclusion of AMD from the CPU market. Simple awareness of a customer’s presence in the EEA cannot be regarded as an instance of implementation of the abuses in an upstream market.
313. In the light of the elements set out in the judgment under appeal, I am therefore not convinced that Intel’s alleged abuse can be considered to have been implemented in the EEA within the meaning of Woodpulp. Nothing in the conduct in question, as far as I see, could be characterised as having been implemented, executed or put into effect in the internal market.
314. That does not exclude, however, that Intel’s conduct may have had anticompetitive effects in the internal market that are caught by Article 102 TFEU. Therefore, I shall now turn to the General Court’s findings with respect to the effects of Intel’s abuse in the EEA.
ii) ‘Qualified’ effects
315. In the judgment under appeal, (195) the General Court first explained that the test to be applied was whether Intel’s conduct could have immediate, substantial and foreseeable effects in the internal market. In its view, that does not mean that the effect on the market must be actual, but only that it must be sufficiently probable that the conduct at issue is capable of having an appreciable and not negligible influence there. It then examined the effects of the two types of conduct separately.
316. As regards the naked restrictions, the General Court noted that, in respect of the fourth quarter of 2006, the two notebooks concerned by the postponement of the launch had projected sales figures of 5 400 and 4 250 units in the whole EMEA region (Europe, the Middle East and Africa). The General Court added that the EEA is an important part of that region. Since Intel did not provide any evidence in support of its assertion that it was possible that all those computers were intended for areas outside the EEA, the General Court took the view that the effects in the EEA were at least potential. It went on to acknowledge that the numbers concerned in the EMEA region were modest but it added that Intel’s conduct was part of a single and continuous infringement. (196) It also took the view that Intel’s conduct was intended to produce immediate effects in the EEA (where, during a certain time, an AMD-based computer was not available) and direct (Intel’s conduct directly concerned the sales of computers by Lenovo). (197)
317. As concerns the exclusivity rebates, the General Court found the effects to be immediate because no Lenovo notebook incorporating an x86 CPU produced by an Intel competitor was available anywhere in the world, including in the EEA. It then added that the anticompetitive effect was foreseeable and even intended by Intel. With regard to the substantial nature of the effect, the General Court stated that the exclusivity rebates formed part of a single and continuous infringement. (198)
318. The General Court’s reasoning is not just succinct. More importantly, it is vitiated by an error in law.
319. For both types of conduct, the General Court’s only argument regarding the substantial nature of the effect in the internal market is that they formed part of a single and continuous infringement. However, as explained in point 179 et seq. above, the concept of single and continuous infringement is merely a procedural rule aimed at alleviating the evidentiary burden of competition authorities. That concept does not — and cannot — extend the ambit of the prohibitions under the Treaties.
320. Yet, that is precisely what the General Court did in the judgment under appeal. Instead of looking at whether the exclusivity rebates and the naked restrictions were each capable of producing an appreciable anticompetitive effect in the internal market, which would have triggered the application of Article 102 TFEU, it merely bundled them together with conduct which took place in the European Union into a single and continuous infringement the effect of which was, in its view, significant. Therefore, two distinct types of foreign conduct that, in principle, may have both fallen outside the scope of Article 102 TFEU were suddenly caught by that provision because they were examined together with other conduct as parts of an overall plan to restrict competition.
321. Had the General Court properly applied the test of ‘qualified’ effects (in assessing whether each type of conduct fell within the jurisdiction of the Commission), the outcome of its analysis may well have been different. For example, the General Court itself stated that the number of computers affected by the naked restrictions was ‘modest’ and that it was unclear whether all or some of those were supposed to be sold in the EEA. As regards the latter element, I must also point to another error in law made by the General Court: it is clearly incumbent upon the Commission to prove that the effects of challenged conduct within the internal market may be appreciable. Indeed, according to settled case-law, it is for the Commission to prove that all the conditions required for the application of Articles 101 and 102 TFEU in a specific case are met. (199) Thus it was wrong to require Intel to disprove an assumption made by the Commission regarding the possible sale in the EEA of computers intended for a much wider region.
322. To be sure, the Lenovo agreements had an immediate and direct effect if those terms are intended to mean that those agreements influenced Lenovo’s conduct as regards the purchase of CPUs and the subsequent sale of notebooks with a x86 CPU requirement. However, the key question here is whether the anticompetitive effects stemming from those agreements were immediate and direct in the EEA. In other words, the General Court should have asked: could those agreements immediately or directly diminish Intel’s competitors’ ability to compete for x86 CPUs within the internal market? The General Court does not examine that aspect at all. It simply stated that those agreements had an impact on Lenovo’s business choices. That is something that any commercial agreement is arguably meant to have.
323. The same flawed reasoning was applied by the General Court in relation to the foreseeability of the effects produced by the Lenovo agreements. Yet again, the General Court focused on the effect those agreements had (or were intended to have) on Lenovo’s commercial choices. The judgment under appeal does not address the foreseeability of the anticompetitive effect those agreements (allegedly) produced in the internal market.
324. On the basis of the elements referred to in the judgment under appeal, far from being immediate, substantial and foreseeable, any anticompetitive effect resulting from the Lenovo agreements appears rather hypothetical, speculative and unsubstantiated. That does not mean, however, that the Lenovo agreements did not have, or could not have, any such ‘qualified’ effect in the internal market.
325. On the one hand, doubts may legitimately arise as to whether, for example, conduct which affected the sale in the EEA of a few thousand computers, representing an extremely limited percentage of the worldwide market for CPUs, over a particularly short time-frame, could be found to have any immediate, substantial and foreseeable effect in the EEA. On the other hand, it cannot be ruled out that the Lenovo agreements could have had a significant impact on AMD’s continuous capacity to develop, manufacture and market CPUs worldwide, including in the EEA. Seen from Intel’s perspective, the exclusion of the only viable competitor in the market for CPUs can be achieved regardless of whether it chooses to target customers that have operations in the EEA or elsewhere. The desired effect remains the same.
326. Regrettably, no such analysis was carried out by the General Court. The fundamental question of whether the Lenovo agreements had the capacity to produce any immediate, substantial and foreseeable anticompetitive effect in the EEA remains therefore unanswered. That is so despite its crucial importance to rule on the application of Article 102 TFEU to the alleged abuse originating from those agreements.
327. On the above basis I conclude that the General Court erred in law in applying both the implementation criterion and the ‘qualified’ effect criterion to dismiss Intel’s (and ACT’s) arguments regarding the Commission’s lack of jurisdiction to apply Article 102 TFEU with respect to the abuses originating from the Lenovo agreements. Therefore the fifth ground of appeal ought to be upheld.
G – The sixth ground of appeal: the amount of the fine
1. The main arguments of the parties
328. The sixth ground of appeal deals with the amount of the fine imposed at first instance. It is divided in two parts. First, Intel claims that the fine is disproportionate, irrespective of any further reduction of the fine as a consequence of the errors in law committed by the General Court. Second, Intel claims that the General Court erred in applying the 2006 Guidelines to conduct that pre-dated them. The retroactive application of the 2006 guidelines to justify a fine more than 50 times larger than that specified under the law in force at the time most of the conduct occurred is, in the appellant’s submission, at variance with the fundamental principles of EU law. In particular, the appellant calls into question the conformity of that approach with Article 7 ECHR and Article 49 of the Charter.
329. The Commission takes the view that the present ground of appeal should be dismissed as partly inadmissible and partly ineffective or, alternatively, as unfounded.
2. Analysis
330. The appellant raises the present ground of appeal as a self-standing ground for annulling the judgment under appeal. It alleges a breach of the principle of proportionality and an erroneous (retroactive) application of the Commission’s fining guidelines by the General Court in determining the fine.
331. The first part of the sixth ground of appeal deals with the (dis)proportionality of the fine imposed on the appellant in the decision at issue, and subsequently confirmed by the General Court. In substance, it asks: what are the appropriate parameters for assessing the proportionality of a fine imposed by the Commission in the context of its investigations?
332. That is by no means an uninteresting question. As a matter of principle, it touches upon the very heart of the powers afforded to the Commission to investigate and sanction infringements of EU competition rules. Additionally, it has implications for the way the EU Courts exercise their unlimited jurisdiction in the context of fining.
333. A detailed answer to that question would require consideration of a wide array of sensitive issues. I have in mind, in particular, the interrelationship between deterrence and the level of fines, the relevant point of reference for measuring proportionality (that is: proportionality in relation to what?) and the limits that Article 49(3) of the Charter may set on the level of fines imposed on undertakings having breached EU competition rules.
334. Regrettably, however, this appeal does not lend itself to such a discussion. Apart from isolated remarks concerning the disproportionate nature of the fine, in particular in comparison to fines imposed previously in rebate cases under Article 102 TFEU, the appellant does not explain in what way the assessment of the General Court breaches the principle of proportionality. (200) Intel simply asks the Court to determine itself what penalty, if any, is proportionate in the circumstances of the present case.
335. In that connection, as is well known, it is not for the Court to substitute, on grounds of fairness, its own assessment for that of the General Court in relation to the amount of the fine. Exceptionally, where the Court considers that the level of the penalty is not merely inappropriate, but also excessive to the point of being disproportionate, the Court can find that the General Court has committed an error in law on account of the inappropriateness of the amount of a fine. (201) The fact that the fine imposed in the decision at issue (EUR 1.06 billion) was record-breaking at the time does not in itself make it inappropriate or indeed, disproportionate, as seems to be suggested by the appellant.
336. In fact, the arguments pertaining to proportionality call into question, in reality, certain findings of fact and, in particular, the assessment of evidence at first instance. (202) Unlike in the context of the other grounds of appeal put forward in the present case, the legal error allegedly committed by the General Court is not readily detectable from the pleadings of the appellant. As already pointed out, it is not for the Court to reassess, in the context of an appeal, the facts or evidence. No substantiated claim regarding a manifest error in the assessment of the facts has been put forward in the present case. Nor is a distortion of evidence of the kind alleged by Intel obvious from the documents in the case file. Indeed, to be reviewable on appeal by the Court, it must be possible to detect the alleged distortion without a fresh assessment of the facts. (203)
337. That is why I agree with the Commission that the arguments put forward by the appellant in relation to the proportionality of the fine must be declared inadmissible.
338. The second part of the sixth ground of appeal deals with the retroactive application of the Commission’s 2006 fining guidelines to conduct that partially predated them. It asks: to what extent is the Commission bound by its fining guidelines?
339. The case-law of the Court is clear on that issue and does not assist the appellant.
340. It is settled law that the Commission is not barred from adjusting the level of the fines (upwards), within the limits indicated in Regulation No 1/2003, should that be necessary to ensure the implementation of EU competition policy. This is because the proper application of the EU competition rules requires that the Commission may at any time adjust the level of fines to the needs of that policy. (204) In that context, the principle of non-retroactivity can only have an impact on the Commission’s discretion to set the fine if the change in question was not reasonably foreseeable at the time when the infringements concerned were committed. (205)
341. More fundamentally still, the Court has held that undertakings involved in an administrative procedure in which fines may be imposed cannot acquire a legitimate expectation that a particular method of calculating the fines will be employed. Rather, the undertakings in question must consider the possibility that the Commission can raise, at any time, the level of the fines by reference to that applied in the past. That is true not only where the Commission raises the level of the amount of fines imposing fines in individual decisions but also if that increase takes effect by the application, in particular cases, of rules of conduct of general application, such as the 2006 Guidelines. (206)
342. Those statements suggest to me that as long as the fine imposed remains within the confines set in Article 23(2) of Regulation No 1/2003, the appellant cannot usefully rely on the principle of non-retroactivity to contest the fine imposed on the basis of the 2006 Guidelines. That is not least because those Guidelines were already in force before the impugned conduct had ended. Indeed, it is Regulation No 1/2003 that defines, as the applicable law, the limits of the Commission’s discretion in imposing a fine for the infringement of EU competition rules, not the fining Guidelines that set out in more detail how the Commission intends to use that discretion.
343. In light of the above, the second part of the sixth ground of appeal should be dismissed as unfounded. Accordingly, the sixth ground of appeal ought to be dismissed.
VI – Consequences of the assessment
344. Under the first paragraph of Article 61 of the Statute of the Court of Justice, the Court is to set aside the judgment of the General Court if the appeal is well founded. Where the proceedings so permit, it may itself give final judgment in the matter. It may also refer the case back to the General Court.
345. I have concluded that the first, second, third, fourth and fifth grounds of appeal should be upheld. As a consequence, the judgment under appeal should be set aside.
346. In the light of the nature of the errors committed by the General Court with regard to the first, second, third and fifth grounds of appeal, the state of the present proceedings does not in my view permit final judgment to be given. That is because a decision on the merits (whether or not the rebates and payments offered by Intel constitute an abuse of dominance contrary to Article 102 TFEU, as well as whether the Lenovo agreements had any immediate, substantial and foreseeable anticompetitive effect within the EEA) hinges upon an examination of all the circumstances of the case and, as the case may be, of the actual or potential effect of Intel’s conduct on competition within the internal market. That, in turn, involves an assessment of the facts which the General Court is better placed to carry out.
347. On the other hand, as concerns the fourth ground of appeal regarding the infringement of the appellant’s rights of defence, the Court would seem, from the outset, sufficiently informed to rule upon the annulment of the decision at issue. Nonetheless, the facts available and the exchange of views before this Court lead me to propose a referral, also on this point, to the General Court. More specifically, the parties should be given adequate opportunity to express their views on the consequences to be drawn from the procedural irregularity in question and more particularly, on whether the decision at issue should be annulled in its entirety (as was the case in Solvay (207)) or only in so far as it concerns Intel’s conduct vis-à-vis Dell.
348. Consequently, I propose that the Court should refer the case back to the General Court for a fresh review.
VII – Conclusion
349. In the light of the foregoing considerations, I propose that the Court should:
(1) set aside the judgment of the General Court of the European Union of 12 June 2014 in Case T‑286/09 Intel v Commission;
(2) refer the case back to the General Court;
(3) order that the costs be reserved.