Skip to main content

Clifford Chance

Clifford Chance

Antitrust

Talking Tech

The General Court annuls the 2009 Intel Decision – When are conditional rebates abusive?

Consumer Tech Antitrust 3 February 2022

On 26 January 2022, the General Court (GC) quashed the 2009 decision of the European Commission (EC) which fined Intel EUR 1.06 billion for implementing an abusive rebate scheme alongside other restrictive practices. Ruling for the second time on the matter, after its first judgment upholding the decision was reversed on appeal, the GC found that the EC did not sufficiently evidence the alleged anticompetitive effects resulting from Intel's rebates and annulled the finding of an infringement and the fine in its entirety.

Key issues at a glance

  • The EU General Court (GC) annulled the European Commission's decision that had found Intel's conditional rebates to be abusive. 
  • The GC applied the guidance of the highest EU Court, according to which conditional rebates cannot be treated as abusive per se. Consequently, the GC was required to assess the merits of Intel's claims that the EC's decision failed to establish the rebates' capability of having foreclosure effects. 
  • To show that rebates are capable of having foreclosure effects, the EC is not required to prove that they prevent asefficient competitors from profitably offering the product at a competitive price. However, if the EC decides to carry out such an assessment, the GC can review it thoroughly if the decision is challenged. 
  • The GC found that the EC's application of the as-efficientcompetitor test was vitiated by errors and that the EC did not adequately prove that Intel's rebates were capable of foreclosing competition. 
  • The judgment highlights the importance of defendants adducing robust economic evidence to show that their rebates are not capable of foreclosing competition.

The Case History

In 2014, the GC ruled for the first time on Intel's challenge of the EC decision, which had found that Intel abused its dominant position in the market for x86 central processing units (CPUs). According to the EC, the abuse consisted of two practices. First, Intel granted rebates to retailers, PC manufacturers, and other customers on the condition that they only stocked computers with Intel's x86 CPUs or purchased virtually all of their x86 CPUs from Intel. Second, Intel made payments to PC manufacturers in exchange for them not launching, or postponing the launch of, computers incorporating competitors' x86 CPUs. The EC found that both practices were capable of producing anticompetitive foreclosure effects.

Intel challenged the decision and made several arguments including, among others, that the EC was required to demonstrate the allegedly abusive rebates were capable of having foreclosure effects on the market and had failed to do so. Intel in particular criticised the EC's application of the as-efficientcompetitor test (AEC test) which, according to the EC, had shown that a hypothetical competitor with the same costs of production as Intel would not have been able to sell its x86 CPUs at a competitive price while maintaining profitability.

The GC dismissed Intel's arguments on the basis that, under the standard set out in the 1979 Hoffman-La Roche judgment, rebates which are conditional upon the customer purchasing all or almost all of their requirements for a product from the dominant company are abusive by nature, so there was no need to ascertain whether they were capable of having foreclosure effects on the market. This, in turn, rendered Intel's criticism of the AEC test ineffective. 

Intel appealed the GC's judgment to the Court of Justice (CJ). As we reported at the time, the CJ's 2017 judgment redefined the legal test applicable to conditional rebates implemented by a dominant undertaking. The CJ held that the EC may presume that such rebates are capable of causing anticompetitive foreclosure, but that presumption does not allow it to disregard evidence which is adduced by the investigated company to show that its conduct does not restrict competition.

The EC is required to consider that evidence and conduct an in-depth analysis of the alleged effects, taking into account the following five elements:

  • the extent of the dominant position on the relevant market
  • the share of the market covered by the practice
  • the conditions and arrangements for granting the rebates
  • the rebates' duration and amount
  • the possible existence of a strategy aiming to exclude from the market competitors that are at least as efficient. 

The CJ also criticised the GC for ignoring Intel's arguments on errors in the decision's application of the AEC test. The CJ held that when the EC applies the AEC test, even if it is not required to do so and does so simply for the sake of completeness, the GC must review the application of the test to ascertain whether the results satisfy the burden of proof. The CJ sent the case back to the GC for it to reassess Intel's challenge in light of the CJ's judgment.

General Court Annulment

Following the CJ's guidance, the GC again considered the EC's assessment of foreclosure effects, this time in light of the evidence adduced by Intel.

First, the GC determined whether the EC's AEC test regarding rebates granted by Intel to Dell, HP, NEC, Lenovo, and MSH allowed it to conclude that these rebates were capable of having exclusionary effects. In a detailed client-by-client analysis, the GC found that the EC's application of the AEC test was in each case vitiated by errors. One error consisted in the EC assuming that the share of demand for x86 CPUs contestable by the hypothetical as-efficient-competitor was 7% on the basis of one internal document, while other evidence suggested that that share was significantly higher. For another customer, the EC claimed foreclosure effects for the entire November 2002-May 2005 period, despite having failed to include the November 2002-May 2003 data in its analysis. These and other errors led the GC to find that the AEC test in the decision did not allow the EC to conclude that, as a result of Intel's rebates, an as-efficient competitor could not have competed with Intel for these clients.

Having invalidated the EC's findings based on the AEC test, the GC assessed whether the capability of Intel's rebates to foreclose competition had been demonstrated in the decision on the basis of the five criteria set by the CJ in the 2017 judgement. However, according to the GC, neither the share of the market covered by Intel's conduct (the rebates only concerned business PCs, while the market included both home and business PCs), nor the duration of the rebates, had been properly analysed in the decision, precluding the EC from demonstrating, to the requisite legal standard, that Intel's rebates were capable of having anticompetitive foreclosure effects. These errors lead the GC to annul the EC's findings as regards the abusive nature of Intel's rebates.

The GC did not reanalyse the naked restrictions, so the decision's finding that these restrictions were abusive still stands.

However, because the GC was unable to identify the part of the fine related to the rebates from the part related to the naked restrictions, it annulled the fine in its entirety.

The GC's annulment of the fine for errors in the substantive finding of an abuse is significant. The EC cannot have a second bite at the apple and investigate Intel's rebates afresh. It remains to be seen whether the EC will issue a new decision imposing a new, presumably significantly lower, fine for the naked restrictions alone.

The EC may also choose to appeal the GC's judgment to the CJ, which would add another chapter to the Intel saga.

Implications for conditional rebates and other types of alleged abuses

The GC heavily criticised the EC for failing to meet the applicable standard of proof in its AEC test. In particular, it faulted the EC for cherry picking evidence and relying on evidential shortcuts, such as extrapolating data from one segment of the market or one time period to prove an infringement in the entire market and for the entire alleged duration. Even with those shortcuts, the EC's application of the AEC test ran to over 150 pages of its decision; carrying out such analyses to the GC's exacting standards in the future will be even more burdensome.

However, it should not be assumed that such burdens will deter the EC (or other EU competition authorities) from bringing cases in the area of conditional rebates in the future. In particular, the CJ's 2017 judgment allows the EC to rely on an evidential presumption that exclusivity rebates are anticompetitive, which must then be rebutted by the dominant company with evidence that it was not, such as an AEC analysis. Where an AEC analysis is submitted by the dominant company, it may be expected that the EC will hold it to the same high standards as its own analysis was held by the GC in the Intel case.