Previous business/academic article Next business/academic article
Business Articles Awards > Dominance

The EU General Court upholds the European Commission’s Decision Regarding Exclusivity Rebates on the Microprocessor Market (Intel)

James S. Venit, e-Competitions Bulletin June 2014, Art. N° 67164

See James Venit's resume

On 12 June 2014 the General Court issued a judgment upholding in its entirety the European Commission’s decision of 13 May 2009 imposing a fine of €1.06 billion on Intel for abusing its dominant position in the market for x86 central processing units ("CPUs"). In its ruling on the substantive issues, the General Court upheld the Commission’s findings that Intel, which during the relevant period had held a market share of about 70% in the global market for x86 CPUs, had granted rebates and made payments to four OEMs, Dell, HP, NEC and Lenovo, and the retailer MSH that were conditioned on exclusivity or quasi-exclusivity, and had made cash payments to HP, Acer and Lenovo — characterized as "naked restrictions" by the Commission — conditioned on those OEMs’ cancelling or postponing the launch, or restricting the distribution, of PCs incorporating x86 CPUs supplied by Intel’s only competitor, Advanced Micro Devices ("AMD"). The General Court further upheld the Commission’s finding that both abuses constituted a long-term comprehensive strategy to foreclose AMD which was implemented by Intel from October 2002 to December 2007.

The Court’s judgment confirmed, inter alia, that there need be no analysis of actual effects or consumer harm to determine the anti-competitive effects of discounts conditioned on exclusivity or quasi-exclusivity, even in an ex post case, and that there is no de minimis threshold under Article 102. It also found that there is no need to establish the existence of a causal link between the discount and the customer’s purchasing decision and rejected the need for a price/cost test to determine whether the discounts had the potential to foreclose AMD. [3] Although the Court acknowledged that discounts conditioned on exclusivity may be legal if economically justified, subject to that theoretical exception, it applied a strict per se approach to Intel’s rebates even where the exclusivity was limited to a market segment or specific product line and took the position that even if foreclosure had amounted to only 14% of total worldwide demand, that would be "significant."

Introduction

On 12 June 2014 the General Court issued a judgment upholding in its entirety the European Commission’s decision of 13 May 2009 imposing a fine of €1.06 billion on Intel for abusing its dominant position in the market for x86 central processing units ("CPUs"). [1] In its ruling on the substantive issues, the General Court upheld the Commission’s findings that Intel, which during the relevant period had held a market share of about 70% in the global market for x86 CPUs, had granted rebates and made payments to four OEMs, Dell, HP, NEC and Lenovo, and the retailer MSH that were conditioned on exclusivity or quasi-exclusivity, and had made cash payments to HP, Acer and Lenovo — characterized as "naked restrictions" by the Commission — conditioned on those OEMs’ cancelling or postponing the launch, or restricting the distribution, of PCs incorporating x86 CPUs supplied by Intel’s only competitor, Advanced Micro Devices ("AMD"). The General Court further upheld the Commission’s finding that both abuses constituted a long-term comprehensive strategy to foreclose AMD which was implemented by Intel from October 2002 to December 2007. [2]

The Court’s judgment confirmed, inter alia, that there need be no analysis of actual effects or consumer harm to determine the anti-competitive effects of discounts conditioned on exclusivity or quasi-exclusivity, even in an ex post case, and that there is no de minimis threshold under Article 102. It also found that there is no need to establish the existence of a causal link between the discount and the customer’s purchasing decision and rejected the need for a price/cost test to determine whether the discounts had the potential to foreclose AMD. [3] Although the Court acknowledged that discounts conditioned on exclusivity may be legal if economically justified, subject to that theoretical exception, it applied a strict per se approach to Intel’s rebates even where the exclusivity was limited to a market segment or specific product line and took the position that even if foreclosure had amounted to only 14% of total worldwide demand, that would be "significant."

Overview

The Intel judgment represents a serious step backwards for the adoption of an effects-based approach to Article 102. In a case which was touted by the Commission as the first to apply a cost-based test to determine whether rebates were exclusionary [4], the General Court has held that:

• A cost-based test designed to determine whether a rival can meet the dominant firm’s discounts in the contestable part of the market is irrelevant to a finding of infringement under Article 102, since even if it is possible for the rival to compete for the contestable share of the market, the discounts may make it more difficult for it to do so and that in itself suffices for a finding of infringement.

• Even in an ex post case in which the conduct has already taken place, it is not necessary for the Commission to establish actual foreclosure or consumer harm. The potential to harm a rival by making it more difficult for it to compete for the contestable part of the market and the constraint on the freedom of customers to use other sources of supply and the increased difficulty of rivals to get access to the market resulting from exclusivity obligations are sufficient to establish the existence of an infringement. This is the case even where it can be shown that the rival greatly increased its sales and profitability during the period of the alleged infringement since, absent the exclusionary conduct, these increases might have even been greater.

• A corollary of the absence of any need to show actual effects is that it is also not necessary to prove that the conditional discount was the decisive or primary cause of the customer’s purchasing decisions even in a case where there is no written agreement and the exclusivity can only be inferred retroactively. In effect, this means that factors such as the quality and availability of the rival’s product are irrelevant to the analysis once there is evidence of conditionality.

• The exclusivity (or quasi-exclusive) condition does not necessarily have to account for all or most of the customer’s total demand in relation to the relevant product market. In one case, the exclusive condition infringed Article 102 even though it covered less than 30% of one OEM’s total demand with conditionality being limited to a market segment. As a result, it appears that it may no longer be possible to rely on the "all or most of its requirements" language in Hoffman-La Roche [5] or the hypothetical 80%-75% benchmark for exclusivity identified in, respectively, the block exemption regulation and guidelines applicable to vertical agreements [6] and Hoffmann-La Roche [7] for assessing whether there is near or total exclusivity in an agreement involving a dominant firm.

• There is no de minimis threshold for infringement of Article 102. In any event, foreclosure from 14% of total worldwide demand for the relevant product was ’significant.’ The finding of the Court of Justice in Van den Bergh Foods [8] that 40% of the market had been foreclosed is merely a finding of fact and does not establish a threshold for foreclosure under Article 102. Rather, foreclosure from any part of the market is sufficient to establish an infringement.

• It is no defense that customers may have offered limited periods of exclusivity as part of their strategy to play their only two suppliers off against each other in order to obtain more favorable pricing, that there was no binding written agreement or that the quasi-exclusive agreement was terminable without cause by either party on 30 days’ notice.

• A strategic plan to exclude can be inferred ex post from the conduct of the dominant firm, and aspirational emails, pre-dating the actual conduct by four years, may be relied on as evidence of the strategic plan.

While consistent in almost all respects with existing case law such as Hoffmann-La Roche [9] and Tomra, [10] the Intel Court’s approach and conclusions are very much at odds with the spirit of the judgment dated 27 March 2012 in Post Danmark [11] in which the full chamber of the Court of Justice indicated, inter alia, that:

• Absent a finding of intent to exclude, discounts that are above the seller’s variable cost are unlikely to infringe Article 102 because they can be matched by an equally efficient competitor.

• Actual foreclosure is relevant in an ex post case.

• The existing jurisprudence concerning the special obligations of dominant firms not to further distort competition should be applied only to former State-owned monopolies that have inherited their dominant positions.

It is not clear whether the Court of Justice’s approach in Post Danmark can be extended to apply to discounts conditioned on near or total exclusivity or to cases in which a court of first instance has concluded, rightly or wrongly, that the dominant firm’s discounts were part of a plan to exclude a rival. However, as discussed below, there are strong arguments that Post Danmark’s views concerning the relevance of actual effects in an ex post case and its approach to the ’special obligations’ of dominant firms are not invalidated where exclusivity and a strategic plan are involved.

The Commission’s Decision

The Commission’s decision of 13 May 2009 had established two infringements

• The conditioning of discounts on exclusivity or quasi exclusivity in respect of four OEMS (Dell, HP, NEC and Lenovo) and one retailer (MSH).

• Naked restrictions involving payments by Intel to three OEMs (HP, Lenovo and Acer) in exchange for their abandoning or delaying the launch of computers equipped with AMD products or restricting their distribution either geographically or by customer segment.

The Commission further found that this conduct, in the period from October 2002 to December 2007 constituted a single strategic plan to exclude Intel’s only competitor, AMD, and thus, in effect, amounted to a single continuous infringement during that period. [12]

The Commission fined Intel €1.06 billion, the largest fine ever imposed on a single firm.

The General Court’s Judgment

The Court’s judgment, which upholds the Commission’s decision on all points, is some 263 pages long. However, its key substantive legal conclusions, including the Court’s legal findings concerning naked restraints and jurisdiction, are contained in the first 48 pages of the judgment. Thereafter the Court addresses procedural issues, its detailed findings of fact with respect to each of the alleged infringements, the existence of the strategy to exclude and the level of the fine.

The Court’s key substantive legal findings concerning exclusive discounts and naked restrictions may be summarized as follows:

• Contractual or de facto exclusive or quasi-exclusive discounts (even if only applicable to certain market segments) infringe Article 102 without any need to show actual foreclosure or consumer harm even in an ex post case (paras 76, 77, 86, 103).

• The Commission is not required to employ a cost-based (AEC test) to establish the potential to foreclose (paras 149-153).

• Even in a de facto case in which there is no written agreement, the Commission does not have to establish a decisive causal link between the discount and the customer’s purchasing decision (para 104).

• Exclusivity need not be for all or nearly all of a customer’s demand; rather, exclusivity for a certain market segment, even if it represents only a relatively small part of the customer’s total demand for the relevant product market, may be treated as an exclusive or quasi-exclusive condition (paras 129 and 132).

• There is no de minimis threshold for the application of Article 102, but, in any event, foreclosure of about 14% of worldwide demand would be ’significant’ (paras 116 – 119, 194).

• The positive performance of the dominant firm’s rival is not relevant to the assessment of abuse (para 136).

• The fact that customers may offer exclusivity in order to get more favorable pricing is irrelevant (para 791).

• Pricing conduct that has no objective justification other than to exclude a rival constitutes a naked restriction and infringes Article 102 (paras 202-205).

Discussion

In reaching its conclusions the Court, reaffirming prior jurisprudence, distinguished between three types of rebate systems:

Type one – quantity rebates. This category covers non-retroactive quantity rebate systems linked solely to the volume of purchases which are generally considered not to have a foreclosure effect. It is permissible for the dominant supplier to pass on the costs savings from higher volume purchases to the customer (para 75).

Type two – exclusive/quasi exclusive rebates. Because they are designed to restrict the buyer’s freedom to choose his sources of supply and to deny other suppliers access to the market, rebates, such as those in Hoffmann-La Roche [13] (and Intel), which are conditioned on the customer’s purchasing all or most of its requirements from the dominant firm will always infringe Article 102 save in exceptional circumstances where they can be objectively justified by efficiencies (paras 76-77). According to the Court, the "capability of tying customers to the undertaking in a dominant position is inherent in exclusivity rebates" and "is designed to prevent customers from obtaining their supplies from competing producers." (Para 86) As a result, there is no need to examine all the circumstances to determine whether the rebate is designed to foreclose rivals (para 86). Given the "inherent" foreclosure effects of exclusivity discounts, this virtual per se approach is "justified by the special responsibility that an undertaking in a dominant position has not to allow its conduct to impair undistorted competition and the fact that exclusive supply conditions in respect of a substantial proportion of purchasers by a customer constitute an unacceptable obstacle to access to the market." (Para 90)

Type three – loyalty-inducing rebates. This category includes rebates (such as those in Michelin I [14] and Tomra [15]) which are not directly conditioned on near or total exclusivity but which may have a loyalty-inducing effect, including rebates based on individual sales objectives. The legality of these rebates requires consideration of "all the circumstances" and in particular the rules governing the grant of the rebate and whether the rebate system is economically justified (para 78). However, there is no need to establish actual foreclosure or apply an AEC test in respect of such rebates (para 103) and there is no de minimis threshold if the rebates are loyalty-inducing (paras 116 and 119).

The rejection of the AEC test and the use of the term (but not the concept of) ’naked restrictions’ are novel points, as is the application of exclusivity to a market segment rather than the customer’s total demand for the relevant products. [16] On the other hand, the Court’s rejection of the need to show actual effects, its assertion that there is no de minimis threshold under Article 102 and its conclusion concerning the irrelevance of who offers the exclusivity commitment are all based on prior jurisprudence, although, as discussed above, rejection of the need to show actual effects has recently been called into question by the Court of Justice in Post Danmark [17]. To this extent, the problem with the Intel judgment is less its novelty than its failure to be novel in the spirit of Post Danmark and, in particular, its maintenance of the view that the existence of dominance leaves so little room for competition that dominant firms have a special obligation to compete very carefully if at all.

Implications of the Intel Judgment

It is unclear to what extent Intel can be limited to its facts and circumstances and to the unique convergence of consistently high market shares that allegedly made Intel an unavoidable trading partner, the alleged strategic plan to exclude its only rival, the aggravating use of alleged naked restrictions incapable of pro-competitive justification and the allegations concerning Intel’s attempts to conceal its conduct. The Court’s conclusions concerning the absence of any need to prove actual effects, a causal link between discounts and customers’ supply choices or consumer harm when Type 2 exclusive rebates are involved and the absence of any de minimis threshold under Article 102 are not unique to Intel and would only seem to confirm the existing jurisprudence applicable to Type 2 rebates.

In this context, the rejection of the AEC test, while novel, is not particularly surprising. This does not make the Court’s reaffirmation of the existing Article 102 jurisprudence any less disappointing in light of the Commission’s Article 102 Guidance paper and the expectations of a possible change in the approach to Article 102 aroused by the Court of Justice’s judgment in Post Danmark [18] discussed immediately below.

As concerns the other two arguably ’novel’ substantive elements of the Intel judgment — the finding of exclusivity in the case of HP, and the acceptance of a single strategy to foreclose AMD over a five-year period where much of the conduct in question was, individually, of relatively brief duration — it is the former that is likely to have the most significant practical implications going forward. When attempting to determine whether near or total exclusivity has been imposed, it may no longer be possible to base that assessment on the totality of the customer’s demand for the relevant product if the customer itself has diverse product lines, one or more of which are subject to near or total exclusivity obligations even though these obligations cover a relatively small percentage of the customer’s overall demand for the relevant products.

Dissonance with Post Danmark

The General Court rejected the relevance of Post Danmark on the grounds that it involved Type 3 and not Type 2 rebates (para 100).

It is not clear whether the existence of an alleged plan to exclude AMD or the use of discounts conditioned on some measure of exclusivity are sufficient to rule out the type of analysis employed in Post Danmark. That question may be answered by a second Post Danmark referral in January [19] or by an Intel appeal of the General Court’s judgment. As things now stand, the Post Danmark and Intel judgments seem highly incompatible in concept and spirit, and it can be seriously questioned whether that incompatibility can be justified by the presence in Intel of exclusivity rebates or the existence of a plan to foreclose.

A strong argument can be made, at least in an ex-post case where the effects of the conduct can be assessed, that there should still be a requirement to take actual effects into account and to show a causal nexus between the conditional discount and the customer’s purchasing decision, even in the presence of exclusivity and an alleged plan to foreclose. First, one of the main arguments cited in Intel in support of a per se approach which does not require actual effects – that the Commission should not have to wait for actual exclusion to occur before it takes enforcement action (para 252) [20] — is not relevant where the conduct has already occurred. The other argument cited by the Commission and the Court — that AMD might have done even better absent the exclusionary conduct (para 136) — arguably remains valid in an ex post case. However, it is also a highly speculative claim for which the Commission should have the burden of proof with the latter requiring more than unsupported assertions that the rival might have performed even better absent the pricing conduct, particularly where there is evidence that the rival lacked capacity or produced products of either inferior quality or which were not well adapted to key market segments. Second, it is far from clear that exclusivity and the existence of intent or a plan to exclude a rival are decisive factors that preclude the approach taken in Post Danmark. Discounts conditioned on exclusivity and a loyalty-inducing discount differ only in form, not function. The key issue in both cases is whether an as efficient rival can profitably match the discount. As concerns plans and intent, the desire to win (whose unavoidable corollary is to defeat one’s rival) is the hallmark of competition. [21] To ignore or condemn this is the product of a competition culture in which too much emphasis has been placed on cooperation and too little on competition. [22] In the end, what counts is not the intention or the manner in which it is expressed, but the means chosen. And when the means involve discounting, there is a strong argument that the appropriate quantitative tests are required to determine whether the discounting involves pro or anti-competitive effects.

Conclusion

The Intel case was intended to help modernize the approach to Article 102 by developing an effects-based approach. Instead it has turned out to have produced the exact opposite result, given, inter alia, the General Court’s firm rejection of any need to show actual effects, consumer harm or apply a cost-based test and its reliance on the traditional Article 102 jurisprudence concerning the special obligation of dominant firms to compete very cautiously if at all. Subsequent market developments — AMD’s stable or declining position for x86 CPUs and the overall decline in PC demand given the growth of smart phones and tablets which are powered by different processors for which Intel’s share is not significant — suggest that the Commission’s decision has had little or no impact on this dynamic and fast-moving high tech marketplace. However, unless overturned or modified on appeal, the judgment to which it has given rise will serve to reinforce the strict, non-effects based approach to Article 102 which arguably chills, rather than protects, competition. Against the background of the proposed shift to a more effects-based approach, the General Court’s robust and unflinching affirmation of the status quo ante can only be regarded as a disappointment and major step backwards.

The views expressed herein are solely those of the author, have not been reviewed by or discussed with Intel and should not be viewed as in any way reflecting Intel’s views.

Footnotes

[1Case T-286/09 Intel Corp. v Commission, judgment of 12 June 2014, nyr.

[2The Court also upheld the Commission’s finding that Intel had sought to conceal its conduct. Intel at paras 1542-1551.

[3In its decision the Commission had applied the ’as efficient competitor’ test (the "AEC test") to determine whether AMD could supply the contestable part of customers’ demand above cost and stated that the results of the AEC test corroborated its findings that Intel’s discounts were exclusionary. In its judgment the General Court rejected the relevance of the AEC test in cases of both exclusive and loyalty-inducing discounts (paras 143-166) and declined to examine whether the test – which had taken up some 150 pages of the Commission’s decision – had been properly applied.

[4See speech by Commissioner Neelie Kroes, "Commission Enforcement Policy and the Need for a Competitive Solution to the Crisis", Address to the Irish Centre for European Law, Dublin, 17 July 2009: "This workability [of an effects-based approach] is proven by its applicaiton in such cases as the Intel rebate prohibition decision (from May 2009). We applied the equally efficient competitor test using the methodology set out in the Article 82 Guidance."

[5Case 85/76 Hoffmann-LaRoche v Commission [1979] ECR 461, para 89.

[6Commission Regulation (EU) No 330/2010 of 20 April 2010 on the application of Article 103(1) of the Treaty on the Functioning of the European Union to categories of vertical agreements and concerted practices, OJ 2010 L102/1 (the "VBER") and Guidelines on vertical restraints OJ 2010 C130/1. Although the VBER and the vertical guidelines do not address whether the 80% threshold would be applicable in the case of a dominant firm, and the safe harbor identified in para 89 of Hoffmann-La Roche stops at 50%, the 80% figure has sometimes been used as guidance concerning the parameters of exclusivity.

[7Note 5 at para 83(c).

[8Case T-65/98 Van den Bergh Foods v Commission [2003] ECR II-4653.

[9Note 5.

[10Case T-155/06 Tomra Systems v Commission [2010] ECR II-4361, and Case C-549/10 P Tomra and others v Commission, judgment of 19 April 2012, nyr.

[11Case C-209/10 Post Danmark v Konkurrenceradet, judgment of 27 March 2012, nyr.

[12Of the arrangements in question, only those relating to MSH and Dell covered all (MSH) or nearly all (Dell) of this five-year period. The duration of the other agreements and the naked restraints was considerably shorter.

[13Note 5.

[14Case C-322/81 Michelin v Commission [1983] ECR 3461.

[15Note 10.

[16The Court’s denial of the need to show causality can only be appreciated in the context of those cases in which the Commission based its inference of exclusivity on the conduct of the dominant firm and/or relied on the alleged incentivizing effect of Intel’s rebates.

[17Note 11.

[18Note 11. See E. Rousseva and M. Marquis, Hell Freezes Over, A Climate Change for Assessing Exclusionary Conduct under Article 102 TFEU, Journal of Competition Law and Practice, Vol. 4 No. 1 (2013).

[19Case C-23/14 Post Danmark AS v Konkurrenceradet, 16 January 2014 (appeal pending). In this case the Danish court has asked the Court of Justice to provide guidance on a number of critical questions relating to discounts including, inter alia, the relevance of an AEC test and whether any foreclosure effect must be appreciable.

[20The Court makes this argument in support of its conclusion that the Commission is not required to establish actual effects in order to justify its exercise of jurisdiction under public international law (para 251). However, this argument is equally applicable as a justification for not requiring the Commission to establish actual effects in order to prove the existence of an infringement.

[21See the Joint Opinion of Advocate General Vesterdorf in Case T-1/189 Rhone Poulenc v Commission [1991] ECR II-867: "...some relaxation of the requirements of proof must be regarded as unobjectionable. Otherwise, in many cases the Commission would in all likelihood have to abandon prosecution from the outset in cases where there is unquestionably an unlawful cartel but where it is not possible to adduce detailed proof of each party’s involvement in the cartel’s activities."

[22See D. Gerber, Law and Competition in Twentieth Century Europe (Oxford, 1998), 75 and G. Amato, Antitrust and the Bounds of Power: the Dilemma of Liberal Democracy in the History of the Market (Oxford, 1997).

© 2017 - Institute of Competition Law Download our brochure