The EU General Court's 2022 Intel Judgment: Back to Square One of the Intel Saga

Abstract: The latest judgment of the General Court in the Intel case annulled the EU Commission’s decision from 2009 imposing a €1.06 billion fine on Intel for abusing its dominant position by offering fidelity rebate schemes (case T‑286/09 Intel v Commission ECLI:EU:T:2022:19). The judgment reaffirms the application of an “effects-based” approach which requires careful economic analysis in order to establish the abusive nature of fidelity rebates. The judgment demonstrates that the presumption that fidelity rebates are restrictions of competition by object can be rebutted by the dominant company. It also clarifies that the as efficient competitor (AEC) test is not an indispensable part of the assessment in examining the foreclosure capability of all rebate systems but can be a relevant factor where the Commission has carried it out as part of its assessment of the anticompetitive effects of the rebate schemes. This Insight seeks to examine how this clarification can be translated into concrete lessons not only for future cases but also for other cases dealing with similar issues (i.e. the Qualcomm and Google Shopping cases) and, in particular, the significance of the AEC test as a specific tool to evaluate the anticompetitive effects of fidelity rebates. The Insight concludes that the recent judgment leaves more questions than answers regarding the application of the AEC test, and that it can be seen as signalling the demise of the application of this test for future cases.

Keywords: EU competition law – abuse of dominant position – fidelity rebates – effects-based approach – as efficient competitor test – Intel.

I. Introduction

This Insight examines the application of the as efficient competitor test (AEC test) in light of the 2022 judgment of the General Court (GC) in Intel.[1] In recent years, the application of the AEC test has been one of the most controversial topics in the field of EU competition law. The controversy arose from the fact that the AEC test had been endorsed by the EU Courts as regards abuse of dominance cases concerning pricing practices (TeliaSonera, Deutsche Telekom and Post Danmark I)[2] but rejected in relation to the anticompetitive effects in cases in which the competition issue identified was not one of pricing (the 2014 GC judgment in Intel, Post Danmark II, Google Shopping, Qualcomm).[3] One possible explanation could be that the AEC test should be interpreted as a conceptual principle; often conflated with a price-cost test, which is one type of evidence to verify a possible exclusionary abuse. What was rejected in the cases above was the application of the price-cost test; the principle that art. 102 TFEU is only concerned about the exclusion of equally efficient rivals stands. The 2022 GC judgment reaffirms this principle and confirms that the AEC test (in terms of price-cost test) is not an indispensable part of the assessment of anticompetitive foreclosure. However, it can nevertheless be a relevant factor where the Commission chooses to carry out this test as part of its assessment. The Insight concludes that there are still many open questions and uncertainties regarding the application of the test.

II. Background

II.1. The European Commission Decision

In its decision of 13 May 2009, the European Commission found that fidelity rebates provided by Intel to Dell, HP, NEC and Lenovo and one retailer (MSH) which were conditional upon those companies purchasing all or nearly all of their x86 Central Processing Units (CPUs) from Intel violated art. 102 TFEU, and fined it in an amount of over a billion euros.[4] The Commission found that Intel had committed a single and continuous infringement of art. 102 TFEU by implementing a strategy, the aim of which was to foreclose Intel’s only competitor, Advanced Micro Devices (AMD) from the market.[5] According to the Commission, Intel engaged in two separate types of exclusionary abuse of its dominant position, forming part of a single strategy to foreclose AMD,[6] namely: conditional rebates and payments and so-called “naked restrictions”, the effects of each conduct reinforcing the other.[7] In addition, it found that the rebates had the effect of restricting those companies’ freedom to choose their source of supply and prevent other competitors from supplying them with corporate desktop x86 CPUs over the period in question.[8]

The Commission conducted an assessment to evaluate the effect of fidelity rebates including a price-cost test but emphasised that this assessment was not necessary, and it was conducted merely for completeness. By referring to the settled case law, the Commission concluded that the assessment of an actual foreclosure effect of fidelity rebates granted by a dominant company was not required for an infringement of art. 102 TFEU to be found. It is sufficient to demonstrate that the conduct of the dominant undertaking is capable or likely to restrict competition,[9] which means that a violation of art. 102 TFEU may also result from conduct by a dominant company that is anticompetitive by object.[10]

II.2. The 2014 General Court judgment in Intel

On appeal, the GC upheld the Commission’s decision and reasserted the settled case law concerning fidelity rebates under art. 102 TFEU.[11] The GC concluded that the Commission was not required to prove a causal link between the practices under consideration and the actual effects on the market and that the price-cost test was not required for finding an infringement of art. 102 TFEU. In its reasoning, the GC distinguished between three categories of rebates – quantity rebates, exclusivity rebates and “other” rebates, and made clear that the first category is generally considered lawful. The GC considered the second category to be exclusive by nature if there is no objective justification for granting it, renaming them “exclusivity rebates” (because they are granted on the condition that customers obtain all or most of their requirements from the undertaking in a dominant position). The GC confirmed the Commission’s position that the rebates granted to Dell, HP, NEC, and Lenovo fell within this second category and therefore their anticompetitive effects can be presumed. For that reason, it was unnecessary to undertake an analysis of their actual effects.[12]

The third category of rebates comprised “other rebates” systems, granted on the basis of certain non-exclusivity related conditions. According to the GC, these could not be presumed anticompetitive, and as such it was necessary to consider all the circumstances in order to assess whether they may have foreclosure effects. The GC further clarified that: “(…) even in the case of rebates falling within the third category, for which an examination of the circumstances of the case is necessary, it is not essential to carry out an AEC test”.[13] Notably, the GC distinguished between pricing and non-pricing abuse practices and stated that the AEC test may be a necessary part of the assessment of price-based abuses (which clearly suggests that the Court equated the AEC test with a price-cost test) because “it is impossible to assess whether a price is abusive without comparing it to other prices and costs”,[14] whereas in the case of exclusivity rebates, it is the condition of exclusive or quasi-exclusive supply that makes access to the market more difficult. It is therefore the conditions that is abusive, rather than the amount of the rebate.[15] The GC considered the relevance of the coverage of the practices in question and concluded that in markets where the structure of competition is already weakened by the mere presence of a dominant company, even a small further weakening of the degree of competition may constitute an abuse of dominant position.[16] Finally, the GC considered that the duration of the supply contracts is not relevant but it is important to consider the possibility of termination/switching to an alternative source of supply without incurring costs. The court concluded that the incentive for customers to purchase exclusively from Intel was based on the existence of a financial incentive, which in practice would prevent them from terminating the contract, regardless of the possibility of termination.[17]

II.3. The Intel judgment of the Court of Justice

In a judgment issued on 6 September 2017, the Court of Justice (CJEU) overturned the GC’s decision and referred the case back for further assessment of the arguments put forward by Intel regarding whether the rebates at issue were capable of restricting competition.[18]

The CJEU began its reasoning by reiterating that the purpose of art. 102 TFEU is to ensure that effective competition is not distorted, clarifying that a dominant company is not prevented from competing on the merits, and that only if a competitor at least as efficient as the dominant company in terms of, among other things, price, choice, quality or innovation is excluded should the conduct be considered abusive (this principle had already been established in the case law in Post Danmark I).[19] Next, the CJEU reiterated settled case law on the concept of dominance and the special responsibility of dominant firms.[20] The “special responsibility” means that a dominant company can be prohibited from adopting pricing practices that can exclude a competitor as efficient as the dominant company which strengthens its dominant position. As a result, “not all competition by means of price may be regarded as legitimate”.[21] From this perspective, the AEC test should be interpreted as a conceptual principle according to which art. 102 TFEU is only concerned about the exclusion of equally efficient rivals. However, the CJEU did not distinguish the AEC test as a principle from the price-cost test, which can be considered as simply one type of evidence to verify a possible exclusion of equally efficient rivals.[22]

The CJEU also reiterated settled case law on fidelity rebates according to which if a dominant company ties its purchasers to obtain all or most of their requirements exclusively from that undertaking, it abuses its dominant position within the meaning of art. 102 TFEU.[23] However, the CJEU clarified that this presumption can be rebutted if the defendant provides supportive evidence that its conduct is not capable of restricting competition; a possibility already established under art. 101 TFEU.[24] In that case, the Commission is required to analyse all the circumstances including the extent of the undertaking’s dominant position on the relevant market, the coverage and the duration of the practice and the conditions and arrangements for granting the rebates. The Commission must also assess the possible existence of a strategy aiming to exclude competitors that are at least as efficient as the dominant undertaking.[25] In this respect, the CJEU stressed the importance of trading off efficiency benefits against foreclosure risks.[26]

The Court further clarified that, in cases when the Commission carries out an in-depth analysis of all the circumstances, including an AEC test (in terms of a price-cost test) and the defendant is able to present evidence to contest the validity of the Commission’s findings, the GC must examine all of these arguments.[27] Moreover, the Court noted that although the Commission considered that the evaluation of all the circumstances and, in particular the price-cost test, was unnecessary to find an abuse of dominant position, it nevertheless chose to carry out such an analysis. As a result, the test played an important role in the Commission’s assessment.[28] For that reason, the GC was required to examine all of Intel’s arguments concerning the price-cost test.[29] The CJEU clarified that the correctness of the implementation of the test should be assessed because it was in the decision and the defendant challenged the Commission’s analysis by providing evidence that challenged the correctness of this analysis. In other words, the case was sent back to the GC on procedural grounds because the judgment had failed to take into account Intel’s arguments in the name of the rights of defence. Finally, the GC’s findings with respect to the naked restrictions were not annulled by the CJEU.

III. The 2022 General Court decision in Intel

On remand, in its 2022 judgment in Intel, the GC annulled the Commission's decision, which according to the court was justified by a single error resulting from the failure to take into consideration, in the initial judgment, Intel’s arguments that challenged the Commission’s AEC analysis, which was applied to test the legality of the loyalty rebates. The GC was clear that the naked restrictions were not subject to the same standard, and they did not require further assessment.[30] The GC also accepted the characterisation of the rebates at issue as “exclusivity rebates”.[31]

The main reasoning of the judgment is drawn from the CJEU’s 2017 ruling, according to which the presumption of illegality of fidelity rebates granted by a dominant company stands but that presumption does not allow the Commission to disregard evidence submitted by the dominant company during the administrative procedure, in which case the Commission has the obligation to assess the anticompetitive effects of rebate schemes.

The ratio of the GC’s judgment was that the presumption of illegality of fidelity rebates can be rebutted if the dominant company provides evidence that its conduct is not capable of anticompetitive distortion of competition. In that case the Commission must analyse the foreclosure effect by reference to the criteria outlined by the CJEU’s 2017 decision, namely: the extent of the dominant position; the conditions and arrangements for granting the rebates; the share of the market covered by the practices and their duration and the possible existence of an exclusionary strategy.[32] In addition, the AEC test is not an indispensable part of this analysis but because the Commission decided to conduct the test, the GC had to revisit the Commission’s AEC test analysis in accordance with the CJEU’s judgment.[33]

The GC accepted that the evidence adduced by Intel cast doubt on the correctness of the Commission’s findings of the AEC analysis.[34] In particular, it concluded that the Commission made an error in its application of the AEC test, affecting in particular the calculation of the contestable shares (i.e. the proportion of a customer's demand that could be captured by an as-efficient-competitor) and the value of the conditional rebates. The court found that the evidence relied on by the Commission to conclude that the rebates granted to Dell and HP were capable of having a foreclosure effect throughout the entire infringement period was not established to the requisite legal standard.[35] In addition, the court found that the Commission had not established, to the requisite legal standard, the validity of the conclusion that Intel’s rebates granted to Lenovo were capable of having or likely to have anticompetitive foreclosure effects on account of errors made by the Commission in the assessment of the non-cash advantages offered by Intel to Lenovo. According to the court, this ultimately affected all the component parts of the examination of the rebates granted to that OEM.[36]

Similarly, as regards the rebates granted to NEC, the GC found that the Commission ”made two errors of assessment, first, by using an exaggerated value for the conditional rebates and, second, by extrapolating the results which it obtained for the fourth quarter of 2002 to the entire period of the alleged infringement.”[37] Next, the court accepted that the Commission’s AEC analysis for the rebates granted to MSH was validated by an error, given that the Commission extrapolated the results obtained for the fourth quarter of 2002 for the whole of the alleged infringement period.[38]

Finally, the GC concluded that the Commission had failed to properly evaluate two out of the five criteria that should be assessed in order to establish a foreclosure capability as set out in para. 139 of the CJEU judgment, namely the share of market covered and the duration of the rebate schemes.[39]

The Commission has decided to appeal the GC judgment. It will therefore be interesting to see what happens next.[40]

IV. Critical analysis and further discussion

The main reason for the annulment of the Commission’s decision was the failure of the GC, in its initial judgment, to take into consideration Intel’s arguments challenging the Commission’s AEC analysis (in terms of a price-cost test). A considerable part of the GC’s decision was devoted to assessing the evidence used by the Commission in the application of the price-cost test and the arguments submitted by the applicants.[41] However, despite the careful review of the price-cost test conducted by the Commission, the GC's decision in Intel left more questions than answers, as it remains unclear how the results of the test, if applied correctly, fit with the rest of the evaluative criteria/market conditions. In addition, the question of whether the price-cost test is a supportive or decisive factor in finding an infringement of art. 102 TFEU is also omitted. Next, does the positive result of the price-cost test mean that the conduct is not an abuse of dominance even if the other circumstances suffice to show the risk of anticompetitive foreclosure? The correct question that the court was expected to deal with is: “Should the Commission have been using the price-cost test at all in this particular case?”[42] In contrast, the GC in its initial judgment considered the price-cost test to be irrelevant for the assessment of exclusivity rebates for two reasons. First, the GC clarified that the price-cost test is limited to pricing practices and is thus irrelevant for the evaluation of exclusivity rebates. In the Intel case, exclusivity rebates were regarded as unrelated to pricing conduct. The same position can be found in GC’s judgment in the Google Shopping case, where the court provided the important clarification that the AEC test (in terms of a price-cost test) is applicable only in pricing practices and does not make sense in cases where the competition issue identified is not one of pricing.[43]

Second, the GC clarified that the test cannot capture the rebates’ anticompetitive nature, and that foreclosure effects could arise even if an as-efficient competitor could theoretically enter the market. This statement indicates that the GC considered that the test is prone to false negatives in holding that even if a competitor is able to cover its costs, this does not mean that there is no foreclosure effect.[44] This statement could be interpreted as suggesting that even if the test is passed by the dominant company, the existence of other evidence, such as unavoidable trading partner status, the significant part of demand secured for the dominant company, retroactivity of rebates in combination with additional anticompetitive conditions, i.e. the “naked restrictions”, would be sufficient to find Intel’s rebate system to be capable of harming competition. In support of this view, we might look at the relevant economic theories providing support for the conclusion that, under these circumstances, fidelity rebates can be anticompetitive even if prices are above an appropriate measure of cost.[45]

Next, the GC in its initial judgment held that the price-cost test is also not required for the evaluation of the “other” types of rebates because the assessment of “all the circumstances” was considered sufficient to demonstrate the existence of a loyalty mechanism, which was deemed to amount to an anticompetitive effect without need for a price-cost test. This position was adopted in AG Kokott’s opinion[46] and repeated in the Post Danmark II judgment, where the CJEU regarded the AEC test as neither legally required nor decisive for establishing an abuse, which might be seen as limiting its usefulness in general.[47] The CJEU considered the application of a price-cost test to be irrelevant for this particular case, for two main reasons. Firstly, the Court considered that the characteristics of the market in this particular case could not accommodate a competitor as efficient as Post Danmark, in which case the presence of a less efficient competitor still might impose a competitive constraint on the dominant company.[48] Secondly, it held that the application of the AEC test does not constitute a necessary condition for a finding of abuse, which means that there is no legal obligation to make use of that test.

Finally, the GC’s rejection of the price-cost test in its initial judgment is in line with the position of many academics who argue that the test is subject to significant implementation errors for a number of reasons.[49] Firstly, the accuracy of the test depends on the proper estimation of the contestable share of sales, which is a difficult, expensive and unpredictable task.[50] For example, in the context of bundled rebates, the discount for the bundle should be attributed to the competitive product in the bundle (which is a distinct product), whereas in the context of single product rebates, the discount should be attributed to the contestable share of demand for one product. From this perspective, although the economic logic of the test is similar for both single-product retroactive rebates and bundled rebates, its practical implementation differs significantly in terms of an estimation of the contestable part of demand.

This raises the question of how to identify the contestable share of demand. One may argue that contestable share is equal to the market share of the dominant company. On this view, if a dominant company has 70 per cent market share, the contestable part would be 30 per cent. Alternatively, the contestable share may relate to the dominant company’s discount level, i.e. if the threshold for obtaining rebates is 80 per cent, then the contestable part is 20 per cent.[51] Others may argue that contestable share is the difference between non-contestable units and the threshold for obtaining the discounts, i.e. if the non-contestable share is 60 per cent and the threshold for obtaining rebates is 80 per cent, it means that contestable share is 20 per cent.[52] Thus, the uncertainty of defining contestable share may lead to an unpredictable outcome, leading to disagreement between the two sides relying on different calculations.[53] For these reasons, the application of the test in practice is questionable. Secondly, the identification of the appropriate measure of cost might be an issue in industries with high fixed costs and relatively low marginal costs, and with corresponding difficulties in accounting sunk costs.[54]

Based on the above, it can be suggested that this modified price-cost test, although economically rational, is subject to implementation errors and cannot reliably identify whether the conduct would lead to anticompetitive exclusion. In particular, and taking account of a considerable set of academic sources, fidelity rebates granted by a dominant company can lead to anticompetitive exclusion even if the dominant company’s price is above an appropriate measure of costs.

Thus, what matters is an exclusion of a competitor that is at least as efficient as the dominant one, regardless of whether a price-cost test is met or not. The Commission’s Guidance Paper recognises this and proposes that the assessment of anticompetitive foreclosure in cases of non-price conduct, such as exclusive purchasing, refusal to supply and tying should be evaluated without a price-cost test.[55] Arguably, the case law has long recognised that some forms of conduct can be exclusionary without involving below-cost pricing.[56] Thus, if the concept of the AEC test is that only exclusion of an as-efficient competitor is capable of harming consumers, it still can be an effects-based approach even if the assessment is carried out without a formal price-cost test. In this sense, the concept of the AEC test might be more broadly interpreted than the view that the AEC test is only a price-cost test. Moreover, the concept of an AEC test, not the price-cost test in itself, can be correctly equated with the effects-based approach.

However, the Guidance Paper lacks clarity as to how to differentiate between conditional rebates that have effects similar to an exclusive purchasing obligation (non-price conduct) and those that might be regarded as pricing conduct. Some forms of conduct, such as tying and refusal to deal, can easily be categorised as non-price conduct, while predatory pricing and margin squeeze can readily be categorised as pricing conduct. However, for conditional rebates it might not be so obvious.[57] Indeed, it is unclear when conditional rebates fall within a pricing or non-pricing category, which is crucial for the choice of the appropriate framework for their assessment. If conditional rebates are regarded as pricing conduct, an application of a price-cost test would be relevant for the evaluation of their anticompetitive effect; if they are regarded as non-price conduct, an evaluation of qualitative evidence rather than comparing price and cost would be relevant.

It is therefore doubtful that it is appropriate for the Guidance Paper to classify some rebates as amounting to exclusive dealing (non-price conduct) while proposing to evaluate their effect by applying an innovative price-cost test, which is extremely complex, prone to implementation errors and which no court endorses so far.[58] It was expected that, on remand in the Intel case, the GC would raise this point as a central issue but it did not do so.

Finally, the GC position that the Commission had failed to properly evaluate the share of the market covered and the duration of the rebate schemes is in contradiction with its statement in its initial judgment, according to which even a small further weakening of the degree of competition may constitute an abuse of dominant position. Accordingly, the CJEU in Post Danmark II took the view that it was not appropriate to create a de minimis threshold above which a practice should be considered anticompetitive merely because competition was already weakened by the presence of the dominant company.[59] As readers will know, the EU Courts have so far refused to evaluate the coverage of fidelity rebates. A possible explanation might be that in some markets characterised by high fixed costs and constant demand, a rival needs to achieve minimum efficient scale (MES) in order to enter the market or to compete effectively with the dominant company if it is already in the market. In these markets, the dominant company might tie an insignificant part of demand, which might nonetheless be large enough to prevent the rivals from achieving MES.[60] The position of the GC regarding the duration of the rebates is also in contrast to its position in the initial judgment, where it refused to accept the duration of exclusivity rebates as short because of the cumulative effect of multiple agreements. Arguably, the relevance of the reference period had been acknowledged by the EU Courts, although the case law does not provide clear indications as to how long is enough for an anticompetitive effect to be presumed.[61] However, according to some empirical studies in the economic literature, the duration of the reference period (even that it is an essential part of any rebate system) cannot be endorsed as a part of the economic assessment of retroactive rebates and their potential foreclosure effect because the rebate percentage, the threshold and the amount already bought is sufficient for the conclusion in that respect.[62] Other empirical models reported that looking only at the length of the contract is misleading; instead, it is important to assess “to what extent a contract of a given length locks the parties into a relationship” due to the penalties that a customer would have to incur in order to terminate a contract.[63] These statements could be interpreted as suggesting that the reference period in itself is not a sufficient indicator to be taken into account; instead, the possibility of terminating the agreement with short notice without penalties, such as a termination of the contract or requirement to return the rebates, would suggest that the practice is capable of harming competition.[64] From this perspective, the GC’s reasoning not to accept the duration as short was based on the existence of a financial incentive, which in practice prevented customers from terminating the contract.[65] From this perspective, an evaluation of the reference period in itself seems redundant. Here again, the recent judgment of the GC in Intel failed to address these issues.

V. Conclusion

The recent judgment of the GC in Intel is not revolutionary. It reaffirms that art. 102 TFEU is only concerned about the exclusion of equally efficient rivals and that the presumption of illegality for fidelity rebates used by a dominant company is rebuttable - principles that have already been clarified by the CJEU. Arguably, the judgment failed to address many important questions regarding the application of the AEC test, simply because the GC was following the framework set out by the CJEU in its previous judgment in the same case. The GC assessed the evidence used by the Commission in the application of the price-cost test and the arguments submitted by the applicants, and concluded that the Commission had made an error in its application of the AEC test. However, the GC did not touch upon the question as to whether the Commission should have been using the price-cost test at all in this particular case. Finally, the GC has only concluded that the Commission had failed to properly evaluate the share of the market covered and the duration of the rebate schemes as evidence to determine the capability of Intel’s rebates to have a foreclosure effect. It did not engage further with its position in the initial judgment, according to which the argument that the duration of exclusivity rebates was short could not be accepted because of the cumulative effects of multiple agreements. Nor did it engage further with and its position regarding market coverage, according to which even a small further weakening of the degree of competition may constitute an abuse of dominant position. From the above, it is doubtful, however, that in future cases the Commission will assess rebates that amount to exclusive dealing (non-price conduct) by applying an innovative price-cost test. Such a test is extremely complex, prone to implementation errors, and so far has not been endorsed by the EU Courts as appropriate.