Over a five-year period, Dell allegedly received payments from Intel averaging $300 million per quarter.a The Attorney General of the State of New York, Andrew M. Cuomo, accuses the Santa Clara-based chip firm of making these payments to force its OEM customer not to use AMD's x86 CPUs in its computers, in violation of antitrust law. Intel is alleged to have infringed Section 2 of the Sherman Act, which opposes behavior by firms aimed at creating or preserving a monopoly other than by merit. In December 2009, the Federal Trade Commission also filed suit against Intel.b The FTC accuses the chip maker of numerous anticompetitive unfair practices, including various payments to its customers in the form of lump sums or discounts.
In Europe, the case is closed, or almost. The billions of dollars that Intel paid to Dell, HP, and several other firms were deemed anticompetitive behavior. The European Commission found that the payments amounted to a strategy to exclude AMD from the microprocessor market. They were considered akin to rebates and restrictions imposed on buyers, which are incompatible with European antitrust law. The Commission found against Intel in May 2009 and fined the firm almost $2 billion.c So, instead of going to its customers, Intel's money replenished the coffers of the European Union! Intel immediately appealed to the EU Court of First Instance in Luxembourg. It also signed a $1.25 billion settlement agreement with Dell to put an end to its antitrust and patent allegations.
Intel considers the payments it made to customers were, on the contrary, a reflection of vigorous competition and beneficial to consumers.
Who's right? Who's wrong? The parties offer diametrically opposed versions of the story.
The story told by the plaintiff State of New York and by the European Commission can be summed up as follows. Intel and AMD are practically the only manufacturers of x86 CPUs, the micro-processors inside most computers. Although four times the size of AMD, Intel was outpaced by the smaller firm in innovation. In particular, Intel is having more trouble negotiating the transition from 32-bit architecture to the 64-bit architecture that makes computers more powerful. According to New York Attorney General Cuomo, Intel has a "big competitive hole in its product development roadmap." In 2003, AMD was the first to introduce a newgeneration processor for the high-end, high-margin corporate server market. Intel feared its competitor would erode its profits on this segment, since business users would be eager to purchase AMD-based desktops and notebooks.
Who's right? Who's wrong? The parties offer diametrically opposed versions of the story.
To prevent that market shift, Intel paid Dell and HP to purchase Intel microprocessors almost exclusively, and paid Acer and Lenovo to delay the launch of their AMD-based notebooks. In other words, Intel paid its customers to protect a segment of its market. Dell was by far the biggest beneficiary of these practices. Between February 2002 and January 2007, Dell received more than $6 billion in return for maintaining an exclusive procurement agreement with Intel. Without these payments, Dell would have reported a loss in some quarters. According to the State of New York, the Federal Trade Commission, and the European Commission, the money that Intel paid its customers was conditional on their boycotting AMD's products. In technical terms, the retroactive rebates given to some OEM customers are loyalty rebates, and the restrictions imposed on OEMs' sales policies are naked restrictions. In Europe, both are generally prohibited because they are perceived to be anticompetitive because they tend to exclude competitors and reduce consumer choice.
Intel's version is nothing like the previous story.d Since 2000, the Santa Clara-based chip maker has faced aggressive price competition from AMD and it has responded by defending itself fairly. AMD's failure to succeed in some market segments is due to its own shortcomings, especially insufficient production capacity, not to any action by Intel. Between 2002 and 2007, the price of microprocessors fell by 36% on average per year and AMD's market share among the main computer manufacturers has risen from 8% to 22%. These figures contradict the claims that Intel has behaved like a monopolist and AMD has been squeezed out of the market. Computer manufacturers know how to play off their two suppliers to their advantage. Intel claims that the payments to customers were not tied to exclusive or near-exclusive purchasing commitments, but were volume-based discounts enabled by economies of scale in production. Thanks to Intel's discount policy, consumers benefit from lower prices. The prices were always higher than Intel's costs. Therefore Intel cannot be accused of selling below cost to drive out AMD.
Whose story should we believe? How can we tell who's right?
In order to decide between the two versions, the first thing to do is of course to look at the facts. This is not easy for an outside observer (including this writer) because the evidence is off limits. Only the public statements and official decisions are available. In the U.S. lawsuits, the State of New York's complaint and the FTC's statements total less than 100 pages. In the EU case, the material is more abundant. The European Commission's decision against Intel runs to more than 500 pages and cites numerous statements by the parties. For example, a Lenovo purchasing manager wrote in an email message dated December 11, 2006, "Last week Lenovo cut a lucrative deal with Intel. As a result of this, we will not be introducing AMD-based products in 2007 for our notebook products." Thousands of figures are also reported. Unfortunately, in order to respect business secrecy, almost all the figures have been deleted from the public version of the decision. Thus, there is no information about the amount of the discounts granted to Dell.
A factual approach is also hampered by the absence of formal contracts. What Intel requested in exchange for the payments to its customers is not mentioned in written documents. Most of the agreements were oral and sealed with a handshake. The few written agreements raise no antitrust concerns. The State of New York and the European Commission accuse Intel of having deliberately removed from the written documents the litigious clauses with respect to antitrust law. If the allegations were proved true, the antitrust agencies would be dealing with a situation akin to a cartel. Since the agreements were secret, evidence is scant and often only indirect.
The problem of access to the data and evidence makes it impossible to verify the validity of the different viewpoints.
For want of being able to decide on the basis of the facts, an outside observer can call theory to the rescue.
The first principle to recall is that antitrust law seeks to protect consumers, not competitors. It does not oppose the elimination of less-efficient competitors; it prohibits behavior of firms that results in higher prices or lower-quality products. While clearly detrimental to AMD, did Intel's actions harm consumers?
In the case of the naked restrictions, the damage to consumers is not in doubt. Let's take the example of Intel's lump sum payments to HP, Lenovo, and Acer in exchange for delaying the launch of their AMD-based computers. That practice (if proven) did hurt consumers: some had to wait before buying the product they preferred, while others, in more of a hurry, had to buy hardware that was not their first choice. Moreover, consumers who were not interested in buying AMD-based desktops and notebooks did not gain anything. The money paid by Intel did not affect the OEMs' marginal cost and, consequently, the price of their computers. Intel and the firms it paid off were the only beneficiaries of these transactions.
The case of the rebates is a more complicated situation. When rebates are linked to volumes purchased, they are good for consumers. They enable manufacturers to pass on some of the savings from economies of scale in production and distribution. In other words, they bring prices down for the consumer. But retroactive rebates tied to market share targets (for example, the buyer receives a rebate if it covers X% of its requirements with the same supplier) are a different story. If a competitor wants to obtain a significant share of the customer's purchases, it must compensate for the loss of the rebates. For example, if Intel offers $100,000 on the condition that HP fulfills 95% of its requirements with Intel, AMD will be forced to offer the same amount if it wants HP to buy more than 5% of its chips from AMD. That threshold effect can have surprising effects. It would explain, for example, why HP refused AMD's offer of a million Athlon XP processors free of charge. If the gift is worth less than the rebate forfeited by not purchasing 95% of its requirements from Intel, it is rational for HP to refuse it.
Economic literaturee shows this threshold effect can lead to the exclusion of competitors that are at least as efficient as the firm offering the rebates. And consumers lose out on two counts. First, there are no more competitors to push down the price set by the dominant firm. So consumers pay higher prices. Second, there is no competitive pressure driving the firm to innovate. So products are manufactured at higher cost and their quality stagnates.
The European Commission sought to show that Intel's loyalty rebates indeed had a foreclosure effect. According to the Commission, a rival with the same costs as Intel would have been excluded. Intel contests this conclusion by finding fault with the Commission's calculations. But once again, the problem of access to the data and evidence makes it impossible to verify the validity of the different viewpoints. Theory without the facts is unfortunately of little use for vindicating either the defendant Intel or the plaintiffs.
d. See "Why the European Commission's Intel decision is Wrong," and "Intel's Response to the EC's Provisional Non-Confidential Version of the Commission Decision of 13 May 2009," September 21, 2009; http://www.intel.com/pressroom/legal/news.htm
e. See, for example, Nicolas Economides, "Loyalty/Requirement Rebates and the Antitrust Modernization Commission: What is the Appropriate Liability Standard?, Antitrust Bulletin 54, 2 (Summer 2009), 259279.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2010 ACM, Inc.