Initial thoughts
What is all the fuss about judgment of the General Court of 26 January 2022 in case T-286/09 RENV Intel Corporation Inc. v European Commission (Intel III). No doubt it is a milestone in the revision process facing antitrust rules which, albeit started by the European Commission more than ten years ago, has only gained momentum in EU case-law over recent years – especially in the domain of abuse of dominance under Article 102 of the Treaty on the Functioning of the European Union (TFEU). Yet, how much credit is it to be given for revolutionising the traditionally fragmentary legal test of exclusionary abuses into a common and more economic analytical framework? If this jurisprudential trend is to be viewed as the “unification” (for romantic that it might sound) of the patchwork of hitherto inconsistent legal tests (and standards), it is necessary to iron out per se rules making some categories of abuse (e.g. tying or loyalty/exclusivity rebates) illegal regardless of the anticompetitive effect – some sort of “by object” abuses, using Article 101 TFEU (agreements) terminology which is however not supposed to apply to Article 102 TFEU breaches.
The first unanimously credited attempt at unifying exclusionary abuses, to be traced back to the 2009 Guidance Paper,[1] already aspired to replace presumptions in the form of per se or “by object” rules with the so-called “more economic approach” to the analysis of anticompetitive foreclosure effect. Nearly a decade of more or less invariable relegation, including by the Commission itself (which soon enough sacrificed its good intentions on the altar of the greater enforcement efficiency afforded by per se rules), the Court of Justice stirred the embers of the more economic approach in the Intel II judgement[2], which ordered the referral back leading up to the present judgment in Intel III.
Since then, other rulings have further paved the way for the unification of exclusionary abuses into the two-pronged general abuse legal test – which is heavily reliant on factual (and economic) case-by-case analysis and thus traditionally left to miscellaneous exclusionary abuse not fitting in any per se pigeon hole: if the dominant company’s conduct (i) either does not qualify as “competition on the merits” (i.e. it admits of no rational economic explanation other than the anticompetitive foreclosure of competitors) or reveals an outright foreclosure strategy, and (ii) is capable of having foreclosure effects (potential is enough), it would be an abuse of dominance. If (i) is not fulfilled (i.e. the conduct amounts to competition on the merits), an abuse can only be found where the exceptional circumstances of the “essential facilities” doctrine (i.e. indispensability of the advantage that the dominant company keeps to itself and elimination of effective competition if not shared) are present[3]. Faced with such an easy and elegant solution the inconsistent set of old-fashioned per se rules seem to go spare – although they will die hard!
Background
Against this background, Intel III confronts one of the most entrenched per se rules or “by object” abuses, i.e. the presumption against exclusivity rebates, after it was dispraised by the minimalistic ruling in Intel II which annulled on cassation the judgment of the General Court in Intel I[4] and referred the case back to first instance. Intel II in turn upheld the EC Decision,[5] where the Commission imposed on Intel a EUR 1,060,000,000 fine for a single and continuous infringement of Article 102 TFEU consisting of abusing its dominant position in the worldwide market for x86 architecture microprocessors (CPUs) (featuring high barriers to entry that allowed Intel to maintain market shares of around or over 70 % all along the relevant period and caused to all rivals except the complainant, AMD, to exit the market) through a two-fold strategy allegedly aimed at foreclosing AMD:
(i) granting “exclusivity rebates” to four original equipment manufacturers (OEMs), Dell, Lenovo, HP and NEC, de jure or de facto subject to their purchasing all or almost all of their x86 CPU requirements from Intel (between 80 % and 100 %), and making payments to the distributor MSH on condition that the latter exclusively sell computers equipped with x86 CPUs produced by Intel; and
(ii) making payments to OEMs HP, Acer and Lenovo, labelled as “naked restrictions”, conditional upon either their postponing or cancelling the launch of AMD x86 CPU-based products or their obliging their distributors to allocate AMD’ products to less profitable customers or not to stock those at all.
Intel’s grounds of appeal against the EC Decision (dismissed by the General Court in Intel I) were the Commission’s (i) failure to observe the burden and standard of proof in presuming the exclusivity incentive and errors in the application of the as efficient competitor (AEC) test; (ii) breach of the territoriality principle under public international law (by establishing the Commission’s jurisdiction based only on the qualified effects of the conduct in the European Economic Area); (iii) violation of Intel’s rights of the defence by relying on unrecorded meetings with this company as evidence; (iv) absence of an overall foreclosure strategy; (v) absence of intentionality or negligence because Intel would be unaware of the anticompetitive nature of its conduct; and (vi) incorrect calculation and disproportionate nature of fines.
On cassation (Intel II), the Court of Justice quashed the Intel I judgment on the grounds that the General Court had wrongly failed to take into consideration Intel’s arguments and evidence seeking to expose errors committed by the Commission regarding the capability of the controversial rebates to restrict competition and, in particular, in applying the AEC test – and referred the case back to the General Court.
Summary of the Intel III judgement
Intel’s first plea concerns the Commission’s departure from the jurisprudential method of assessing a rebate scheme’s capability to restrict competition. On this point, the General Court begins by recalling the reasoning in Intel II: although the Commission inferred from the Hoffmann-La Roche case-law[6] that exclusivity rebates are anticompetitive “by their very nature” or “object” and hence there is no requirement to prove their capability to have foreclosure effects, the watchdog did examine in practice such capability by applying the AEC test and did attach a genuine relevance to the outcome for its assessment of the competition restriction. Therefore, since it is established that Intel submitted, during the administrative procedure, on the basis of supporting evidence, that its conduct was not capable of producing the alleged foreclosure effects, the Commission was obliged to examine at least the five criteria which paragraph 139 of the Intel II judgment picks up from the Hoffmann-La Roche case-law –i.e. extent of the dominant position, share of the market covered by practice, conditions and arrangements for granting the rebates, duration and amount of rebates, and possible existence of a strategy aiming to drive out competitors that are at least as efficient as the dominant company.[7]
It must be noted that, in Intel II, the Court of Justice only mandated the Commission to prove the capability of exclusivity rebates to have a foreclosure effect, particularly on the basis of the AEC test, in the specific scenario where both the Commission had decided to carry out such an analysis in practice and attached a genuine relevance to the outcome and the investigated company had challenged it – the rationale being safeguarding Intel’s rights of defence.[8] Contrariwise, the General Court seems to stretch the Intel II judgment to the point of deriving a full-fledged substantive rule whereby the presumption of the anticompetitive object or nature of exclusivity rebates is rebuttable.[9] In other words, the Commission’s obligation to examine the effects no longer stems from the procedural need to avoid depriving the investigated company of its defence rights, but from the substantive absence of a per se rule (or iuris et de iure presumption) that certain categories of rebates are abusive by nature.
Intel’s second and third pleas for annulment elaborate on the first plea as regards, respectively, the errors in the Commission’s AEC test and the failure to properly examine the criteria in paragraph 139 of Intel II.
As regards the AEC test, it is based upon the premise that Intel is an unavoidable trading partner (given, in particular, the nature of its product, its brand image and its profile) so that OEMs would always buy at least part of their x86 CPU requirements from it irrespective of the quality of its competitors’ offerings. Hence, customers would be willing and able to switch only a certain (“contestable”) share of their demand to an alternative supplier and Intel would be able to use the non-contestable share as leverage to reduce the price on the contestable share. On the contrary, competitors would only have the smaller contestable share to spread out the rebates that they would need to grant to OEMs to compensate for Intel’s rebates over the whole volume that they purchase from Intel which they would forgo if (as a result of their switching) they no longer meet the exclusivity or quasi-exclusivity condition – this economic effect is known as “suction effect”.[10]
The AEC test consists of determining whether the price at which a competitor should have offered its x86 CPUs to compensate in absolute terms for Intel’s payments and rebates (“effective price”) is lower than Intel’s average avoidable cost (average of the costs that could have been avoided if the company had not produced a marginal amount of output, and thus a proxy for the variable cost[11]) in order to determine whether such a competitor with the same costs (as efficient) as Intel would have incurred losses so that even Inter would not have been able to compete viably with itself had not it been for the non-contestable share. In particular, the methodology used in the EC Decision compares the contestable share with the “required share” that a competitor with a unit cost equivalent to Intel’s average avoidable cost must obtain in order to be able to spread out Intel’s absolute rebate volume without incurring losses.
Besides, the General Court recalls that its jurisdiction is limited to the review of the legality of the EC Decision and that it cannot take into account the additional arguments submitted by the Commission to support its AEC test methodology without substituting its own analysis for that institution’s, which runs counter the case-law on proceedings for annulment. On this basis, it examines the evidence provided by Intel by distinguishing, in relation to the standard of proof, cases of indirect proof, where the Commission relies on the supposition that the facts established cannot be explained other than by the existence of anticompetitive behaviour and it is sufficient for the undertaking concerned to provide another plausible explanation of the facts; and cases of evidence which is in principle sufficient to establish the existence of the infringement, where the standard of proof required of the undertaking concerned is higher in that it must demonstrate to the requisite legal standard the existence of the circumstance that it relies on and the fact that such circumstance calls into question the probative value of the evidence relied on by the Commission.[12]
Then, the General Court goes on to examine the Commission’s methodological errors put forward by Intel in relation to each of the OEMs concerned:
(i) The fact that the Commission relied, concerning Dell, on third party evidence (such as documents from Dell) of which Intel was unaware in order to estimate the contestable share is found not to infringe the principle of legal certainty.
(ii) Intel’s submissions do cast doubts, concerning Dell, on the calculation of the contestable share in the EC Decision and thus on the fact that for certain stretch of the relevant period the required was lower (so there would be no foreclosure effect during that lapse); concerning HP, on the total period considered in the AEC test; concerning NEC, on the fact that the evidence (essentially NEC’s internal documents) relied on by the Commission supported the finding of the rebates’ being subject to exclusivity, and on the extrapolation of the results obtained for a part of the relevant period to prove the foreclosure capability to the rest of the period; concerning Lenovo, on the quantification of the non-cash advantages given by Intel, which vitiated all the components of the AEC test; and, concerning MSH, on the double rebate method whereby the fact that this distributor benefited from the rebates received by NEC when purchasing computers from it during part of the relevant period leads to the conclusion that it benefited to the same extent from the rebates received by other manufacturers, without proving that MSH purchased computers from them, and for the whole of the relevant period.
(iii) The Commission’s alternative calculation in the case of Dell (relating to a different part of the relevant period) was not capable of correcting those errors and the reinforcing factors that it introduced to back its findings (the increase in rebates to OEMs other than Dell, the fact that Dell purchased products other than x86 CPUs, the use of figures in the case of HP that were allegedly more favourable to Intel, and the possibility for Intel to allocate HP’s rebates to another OEM in the event that HP purchased its x86 CPUs from AMD) were not sufficient in themselves to support the potential exclusionary effect.
Finally, the General Court upholds Intel’s third plea based on the fact the Commission applied only two of the five criteria in paragraph 139 of Intel II: the extent of Intel’s dominant position and the conditions and arrangements for granting the rebates. It concludes that the analysis of those criteria is flawed in that the Commission in fact improperly disregarded the criteria relating to share of the market covered by the practice (by merely establishing the strategic importance of the OEMs which benefited from the rebates because of their share, their strong presence in the most profitable segment and their ability to legitimise new processors – contrary not only to paragraph 139 of Intel II but also to the 2009 Guidance Paper). In view of this finding, the General Court did not consider it necessary to examine whether the Commission also disregarded the criteria regarding amount of rebates and possible exclusionary strategy.[13]
Conclusion
The General Court does not hesitate to endorse an ambitious reading of the Intel II judgment, where the Court of Justice was careful enough to keep its take on how to introduce the more economic analysis in the assessment of exclusionary abuses methodological and procedural by limiting its conclusions to cases where the investigated undertaking challenged the application of the per se rule. The General Court goes further and interprets the cassation judgment as actually turning the per se rule into a rebuttable presumption. This way, it endows the procedural reasoning of the Court of Justice with substantive essence by proclaiming the Commission’s strict obligation to analyse at least the five economic criteria in paragraph 139 of Intel II – in which one may discern the concretisation of the two limbs of the general (or unified) test for exclusionary abuse: in the first four criteria the capability to have foreclosure effects can be made out, while the conduct not constituting competition on the merits seems to come into view in the possible foreclosure strategy criterion.
To achieve this, the General Court resorts to a literal and even inflexible interpretation (perhaps too much to fit in with the casuistic spirit of the more economic approach or with the openness of the two-limb unified exclusionary abuse test) which elevates the five criteria to the status of a binding exhaustive list and which only appears limited by the case-law according to which the AEC test is only one of various tools available to the Commission to assess those criteria[14] – so that the watchdog is only obliged not to make methodological errors in its application and to take account of the results if it decides to use it. For those reasons, Intel III forges ahead with the unification of the analytical framework for exclusionary abuses into the double test of competition not based on the merits (or foreclosure strategy) and capability to have foreclosure effects, although perhaps less explicitly than other outstanding milestones in this trend that we have seen in the last two years (see Generics, Slovak Telekom or Google – cited above). The breakthroughs in these examples focused rather on the delimitation of the first part of the test by sizing the concept of competition not based on the merits, while Intel III embarks on an equally important contribution to the unification venture: dismantling per se rules to provide the second part of the test (potential foreclosure effect) with effectiveness.
[1] See the Commission Communication of 24 February 2009 – Guidance on the Commission’s enforcement priorities in its application of Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings.
[2] See judgment of the Court of Justice of 6 September 2017 in case C-413/14 P Intel Corporation Inc. v European Commission.
[3] See e.g. judgments of the Court of Justice of 30 January 2020 in Case C 307/18 Generics (UK) Ltd. and others and Competition and Markets Authority, of 25 March 2021 in Case C 165/19 P Slovak Telekom, a.s. v. European Commission, or of the General Court of 10 November 2021 in Case T-612/17 Google LLC and Alphabet, Inc. v. European Commission.
[4] See judgment of the General Court of 12 June 2014 in Case T-286/09 Intel Corporation Inc. v European Commission.
[5] See decision of the European Commission of 13 May 2009 in Case COMP/C-3/37.990 Intel.
[6] See judgment of the Court of Justice of 13 February 1979 in Case 85/76 Hoffmann-La Roche v Commission.
[7] Intel III, paragraphs 121, 122, 124, 125 and 126.
[8] Intel II, paragraphs 139 to 144.
[9] Intel III, paragraphs 144, 148 and 149.
[10] Intel III, paragraphs 152 to 158 of the Judgment
[11] 2009 Guidance Paper, paragraph 26.
[12] Intel III, paragraphs 150, 165 and 166.
[13] Intel III, paragraphs 493 to 500, 515 to 520 and 521.
[14] See the judgment of the Court of Justice of 6 October 2015 in case C- 23/14 Post Danmark [II], paragraph 61.
