The Rambus case is one of the first times that the Commission has dealt with Standard-Setting Organisations (SSOs) under Article 102 TFEU. Like many recent single-firm conduct cases, the same facts were under review by competition authorities on both sides of the Atlantic. However, contrary to some other well known cases, the same conclusion was eventually reached, as both the EU and US proceedings were closed without any finding of infringement on the part of Rambus.
The Rambus case raised very interesting issues regarding the application of Article 102 TFEU to SSOs and about the scope of Article 102 TFEU, generally. This article aims at contributing to the debate by analysing the basis of the Commission’s preliminary assessment, and by setting out some of the arguments in the case which are not set out in the settlement decision because they came after the preliminary assessment (i.e. arguments made in response to the statement of objections) and on which the Commission took no position.
The facts and the history of the case
Rambus is a small1 innovative company founded in 1990 that develops memory interface technology used in DRAMs (Dynamic Random Access Memory a type of electronic memory used to transfer and store data) and other memory devices such as memory controllers. Rambus’ founders, Professors Michael Farmwald and Mark Horowitz, solved the “bottleneck problem” – the problem that memory technology would limit computer performance because CPU speed had been increasing at a faster rate than DRAM speed. Rambus’ revolutionary DRAM design offered a tenfold improvement over traditional DRAMs. As a pure engineering and research company, Rambus has no manufacturing facility. The licensing of its technology constitutes its sole source of income.
From 1992 to 1996, Rambus was a member of JEDEC,2 a US organisation that sets standards in semiconductors. While a member of JEDEC, Rambus held no patent or patent application that read on any JEDEC standard.3 Rambus never promoted its technology for inclusion in JEDEC and was even inappropriately prevented from doing so.4 However, most of the JEDEC members were well aware of Rambus’ technology and its business model.5
In June 2002, the FTC issued a monopolisation complaint against Rambus for alleged exclusionary conduct (in particular, what was claimed to be a “patent ambush”).6 A few months later, two DRAM manufacturers lodged a complaint before the European Commission on the exact same grounds. Rambus was again accused of having engaged in an alleged “patent ambush”.7 It was accused of having failed to disclose patent applications and its intention to patent technology that was discussed at JEDEC,8 albeit that the discussions on most of the standards that were the subject of the complaint did not start until long after Rambus left JEDEC. As a consequence, JEDEC members were allegedly deprived of the opportunity to develop standards that did not read on Rambus’ technology. After having left JEDEC, Rambus started in 2000 to assert its patents (all of which were granted after it left JEDEC based on applications filed after it left JEDEC) and to request payment of royalties for the manufacture of JEDEC-compliant DRAMs. This, it was argued, would not have been possible absent its alleged deceptive conduct.
Rambus responded that this was not a patent ambush, but rather a technology hijack. In other words, certain JEDEC members had inserted Rambus’ technology into JEDEC standards in knowledge that Rambus was seeking royalties on such technology. Rambus also responded that certain companies, acting collectively, had used a number of means, including artificially inflating the price of Rambus-designed DRAMs, with a view to driving Rambus out of the market. A trial on these allegations is scheduled for Spring 2010 in the California Superior Court.9 Rambus invoked those companies’ conduct as a defence in the EU case.
In the United States, the FTC’s Chief Administrative Law Judge dismissed the FTC’s monopolisation complaint against Rambus in February 2004 after a 54-day long trial.10 In August 2006, the full FTC reversed the ALJ decision and found that Rambus had engaged in illegal monopolisation.11 This decision was later quashed by the DC Court of Appeals in April 2008,12 which found that the FTC’s decision was wrong as a matter of law – and also cast doubt on the FTC’s approach to the facts. The Court held that the FTC had “taken an aggressive interpretation of rather weak evidence.”13 The FTC formally closed the case in May 2009.14
In parallel proceedings, Rambus had sought damages for patent infringement against DRAM manufacturers, who defended on grounds of fraud, equitable estoppels and antitrust. In March 2008, the jury of the District Court for the Northern District of California unanimously found against Hynix and in favour of Rambus.15 The Court later confirmed the findings of facts and conclusions of law and entered in March 2009 into final judgment ordering Hynix to pay royalties.16
In the European Union, two years after having issued a Statement of Objections, the Commission closed its investigations without imposing any fine and without finding any infringement.17 In June 2009, Rambus proposed commitments with the view to enabling all parties to put aside their past differences and move towards an environment where the industry resolves patent questions via licensing discussions rather than costly litigation. Rambus committed to limit its future royalties rates (on a worldwide basis) to a maximum of 1.5% for the bundled licence for certain DRAM memory chips (with a royalty holiday for certain older DRAM memory types)18 and a maximum of 2.65% for memory controllers reducing to 2% in April 2010, both on a going-forward basis.19 Rambus’ commitments were market tested and accepted by the Commission on 9 December 2009, thus also recognising Rambus’ right to charge royalties in the future.
Very recently, Samsung and Rambus agreed a settlement of all pending claims under which Samsung was to invest $200 million in Rambus stock, make an initial payment of $200 million and pay $500 million over the next five years.20
Alleged dominant position
In its preliminary assessment, the Commission left open the exact market definition.21 The Commission had provisionally considered that it was “commercially indispensable” for DRAM manufacturers to comply with JEDEC standards22 and that the industry was “locked in to JEDEC standards”.23 On that basis, it had provisionally considered that Rambus was in a dominant position when it asserted its patents.24 Rambus denied having ever held any dominant position.25 It was also noted by Rambus that the method applied by the Commission to reach its preliminary finding of dominance was questionable in light of the settled case-law holding that the definition of the relevant market is “a necessary precondition” of any assessment under Article 102 TFEU.26
The complaining DRAM manufacturers’ position was inherently contradictory with their claim that Rambus was dominant. On the one hand, in antitrust proceedings, they alleged that Rambus was dominant on the basis of patents which read on JEDEC standards. On the other, in patent infringement proceedings, they always denied that Rambus’ patents were either valid or infringed (i.e. the patents did not read on the standards) and refused to pay any royalties. However, the finding of dominance in this case necessarily implies recognition of the fact that Rambus’ patents were valid and infringed.
In addition, the Commission’s preliminary assessment did not take into account the DRAM manufacturers’ significant countervailing buying power.27 Rambus’ patents, even if valid and infringed, could not guarantee that it was able to exercise dominance in downstream markets as DRAM manufacturers were able to utilise their own market power to dictate the terms on which licences were accepted. Far from being in a position to act “independently of its customers”,28 Rambus was dependent on the semiconductor industry to manufacture DRAMs incorporating its technology. For years, Rambus had been unable to collect royalties from most DRAM manufacturers and reported losses.29 In 2008, only a small fraction (well under 40%) of DRAMs produced worldwide were made pursuant to a license from Rambus.
Moreover, one might question whether the market shares of the products incorporating the patented technology could accurately reflect market power in technology markets, such as the DRAM market, which are typically volatile.30
As regards industry lock-in, we submit that its existence must be assessed at the point of time when the industry considers adopting a new generation of chips. Market data showed that, in the DRAM industry, standards are quickly replaced and different standards coexist at all times. Rambus presented several expert reports demonstrating that switching costs were not prohibitive for the industry and testimonies given during the US proceedings also supported the absence of lock-in. In 2004 the ALJ found that “DRAM manufacturers were not locked in to using Rambus’ technologies at any point of time”.31
Alleged duty to disclose patents and patent applications to JEDEC
In its preliminary assessment, the Commission had found that Rambus was under a disclosure duty on the basis of (i) the “obligation to disclose issued and pending patents relating to the standard-setting work of JEDEC incumbent upon every member of the organisation”;32 (ii) the “expectations of other participants”;33 and (iii) the “underlying duty of good faith [under Article 102 TFEU] that is binding on a participant in a standard-setting process”.34 These three grounds will be analysed in turn. We believe there are sound elements to question the existence of a disclosure duty on any of these grounds.
JEDEC’s rules and policies
The Commission’s preliminary assessment considered that a disclosure duty may exist on the basis of certain phrases in certain JEDEC manuals. This followed the approach of the FTC in 2006. However, the FTC’s interpretation of JEDEC’s policies was severely criticised by the DC Court of Appeals.35 The US courts that have considered the matter have all described JEDEC’s patent disclosure policies as suffering from “a staggering lack of defining details”,36 and have rejected the idea that Rambus was subject to a clear duty to disclose. Judge Whyte concluded “not only did Rambus not have an obligation to disclose pending or anticipated patent applications, it had sound reasons for not doing so.”37 The existence or non-existence of a duty of disclosure in a US SSO whose rules are subject to US law is to our mind a question of US law. In our view, this point became res judicata after the US courts’ ruling on that issue.
JEDEC members’ expectations
The analysis of the actual practice within JEDEC, as confirmed by several testimonies from the US proceedings, shows that JEDEC members did not systematically disclose all their patents. The Federal Circuit held that “if these members perceived the duty to encompass any patent or application with a vague relationship to the JEDEC standard, the record would likely contain a substantially greater number of disclosed patents and applications.”38 Moreover, to the question “did JEDEC members share a clearly defined expectation that members would disclose relevant knowledge they had about patent applications or the intent to file patent applications on technology being considered for adoption as a JEDEC standard?” the jury of the Northern District of California replied “NO”.39 As a matter of comity, we submit the Commission would have had difficulty in ignoring factual determinations made by judicial bodies in the jurisdiction where the disputed facts occurred.
There is also a broader point when authorities consider basing an antitrust duty on the “expectations” of competitors. At their most basic level, SSO activities are a type of concerted action by a group of competitors – something normally viewed with suspicion by competition law. While the benefits from SSOs will often outweigh the risks associated with that coordination, antitrust risks do remain. It is submitted that the best way to manage such risks is to ensure that the SSO’s rules are clear and unequivocal and known by all in advance – this is surely preferable than for conduct in the SSO to be judged according to the “expectations” of rivals that are only expressed subsequently.
(In Rambus’ case, those subsequently expressed “expectations” were clearly influenced by a commercial interest in trying to avoid paying royalties.) More broadly, if antitrust duties are imposed based on “expectations”, then those expectations must be reasonable – otherwise in the extreme case a determined group of competitors could use “expectations” to rein in a successful competitor under the guise of an SSO.
Duty of good faith
We submit that there are reasons to doubt that Article 102 TFEU permits the imposition of an implicit disclosure duty on the basis of a general duty of good faith, not least when such a duty is not set forth in the rules of the non-European SSO. It would in our submission constitute an inappropriate invasion by competition law into the field of “unfair trading” law. By contrast to Section 5 of the FTC Act which prohibits unfair or deceptive acts or practices,40 there are no general EU law principles on such matters as good faith and legitimate expectation. Until now such principles have been regarded as being private law issues solely regulated by national laws. In other words, “unfair trade” falls outside the scope of Article 102 TFEU, which focuses on the maintenance of competition and the protection of consumers and not on the enforcement of “fair” business standards.
Relevance of SSO rules
Commission’s preliminary assessment was that “an actual breach of the precise rules of a standard-setting body would not be a necessary requirement for a finding of abuse.”41 This statement (albeit not a final finding by the Commission) has potentially broad implications beyond the specific Rambus case.
In effect, this statement suggests that competition law would rewrite and supersede the rules adopted by SSOs. It overlooks the fact that some SSOs may legitimately decide not to enact any disclosure duty in light of the inherent significant administrative costs that could hamper their standard-setting work. Furthermore, when an SSO decides whether to impose a mandatory disclosure duty, this requires a careful determination of its exact personal, material and temporal scope: for example, disclose by whom (everybody vs technology promoters); what to disclose (patent vs patent application, essential vs related patents); and when to disclose (at ballot vs earlier). The application of a one-size-fits-all approach appears inappropriate and could even be contradictory to Commission guidelines stressing that the assessment of each standard-setting agreement “must take into account the nature of the standard”.42 Finally, in the absence of any EU precedent, such rulemaking would also appear problematic given its retroactive character: no-one could have foreseen such an application of what was then Article 86 of the EC Treaty in the early 1990s.
The alleged abuse – and the extent to which monopolisation should be caught by Article 102 TFEU
In its preliminary assessment, the Commission considered that Rambus might have engaged in “intentional deceptive conduct in the context of standard-setting process by not disclosing the existence of patents and patent applications which it later claimed were relevant to the adopted standard”.43 The Commission provisionally considered that Rambus might have abused its dominant position “by claiming royalties for the use of its patents from JEDEC-compliant DRAM manufacturers at a level which, absent its allegedly intentional deceptive conduct, it would not have been able to charge”.44
In essence, the Commission’s preliminary view thus alleged that Rambus breached its disclosure duty and charged excessive royalties. In response, Rambus denied that “it has engaged in abusive conduct that has adversely affected competition within the EEA or that it has engaged in a patent ambush or any form of intentionally deceptive conduct”.45
Alleged intentional deceptive conduct
We submit that the existence of a deceptive conduct is a factual question that is also largely a question of US law since it is closely linked to the interpretation of the rules in force within the SSO. US courts have held that there has been no deceptive conduct.46 The Northern District of California jury found that Rambus did not act in bad faith or engage in deceptive conduct.47 In response to the question “For each market you have found to exist, do you find it more likely than not that Rambus engaged in anticompetitive conduct?” the jury answered “NO”. In response to the question “Did Rambus utter half-truths about its intellectual property coverage or potential coverage of products compliant with synchronous DRAM standards then being considered by JEDEC by disclosing some facts but failing to disclose other important facts, making the disclosure deceptive?” the jury again answered “NO”. We believe it would have been difficult for the Commission to justify reaching the opposite conclusion on the basis of the same set of documentary evidence, especially as it did not have the benefit of hearing live testimony from the complainant DRAM manufacturers’ representatives at JEDEC. It is a matter of record that a number of JEDEC members were
aware of Rambus’ business model and were put on notice that Rambus might have relevant IPRs. In particular, potential scope of Rambus’ patents was a matter of public record following the publication in 1991 of its PCT application.48 Moreover, Rambus’ refusal to comment about the precise scope of its potential IP rights at an early stage “could not have been reasonably construed as a statement that Rambus lacked any form of intellectual property” covering features under consideration at JEDEC.49 Furthermore, Rambus never promoted its technology for inclusion in JEDEC standards. On the contrary, when Rambus did inquire about presenting its own technology to JEDEC it was inappropriately barred from doing so by the Chairman of the meeting, IBM’s Gordon Kelley. Such a move was unprecedented and according to the ALJ demonstrated a “clear conflict of interests”.50
Is monopolisation an offence under Article 102 TFEU?
In its preliminary assessment, the Commission alleged that Rambus became dominant after the inclusion of its technology into JEDEC standards – thus as a result of its alleged deceptive conduct.
Even assuming for the sake of argument that Rambus employed some “underhand” strategy (something that has been rejected by the relevant US Courts and which Rambus strongly denies) which led to its dominant position, there would be a fundamental problem with this approach. Unlike US antitrust law, where Section 2 of the Sherman Act prohibits the acquisition of a dominant position by unfair means,51 under EU law the mere creation of a dominant position is not an abuse. Article 102 TFEU only prohibits the abuse of dominance and not the creation of dominance itself. This was confirmed by the Court, which has consistently held that “only the strengthening of dominant positions and not their creation can be controlled under Article  of the Treaty”.52
Can “exceptional circumstances” bring monopolisation within the scope of Article 102 TFEU by the back door?
In its preliminary assessment, the Commission sought to avoid the fact that monopolisation is not an offence under EU law by alleging that the abuse occurred when Rambus asserted its patents and not directly when Rambus allegedly failed while at JEDEC to disclose supposedly relevant patent applications or intentions to file such applications. However, the Commission provisionally characterised the royalties charged by Rambus as “excessive” because of its earlier conduct: they were allegedly “incompatible with Article 102 TFEU, in light of the specific circumstances of this case”.53
We submit that, had the case gone to a final decision, the Commission would not have been justified to rely on a vague concept of exceptional/specific circumstances to impose duties under Article 102 TFEU on an undertaking at a time when it was not dominant. Article 102 TFEU places dominant firms under a “special responsibility” not to conduct themselves in such a way as to impair undistorted competition in the common market.54 It follows that, by contrast, non-dominant undertakings are under no such duty as far as competition law is concerned. In effect, if a final Commission decision were to have relied on Rambus’ past conduct to find the royalties excessive, such a precedent would impose a general obligation on all undertakings not to act abusively, or take the risk of having proceedings brought against them if and when they eventually become dominant. It would effectively eliminate the core distinction between dominant and non-dominant firms at the heart of Article 102 TFEU and, consequently, dilute the “special responsibility” of dominant firms not to act anti-competitively. If the abusive conduct and dominance need not coincide in time, the only remaining difference would be whether proceedings could be started immediately or whether competition authorities first have to wait for the company to become dominant. In past cases, the undertaking’s dominance has always been contemporaneous with its abusive conduct.
When is enforcing an IP right to seek royalties abusive?
The case law suggests that the enforcement of an intellectual property right does not constitute an abuse of a dominant position.55 Almost all the previous cases on IP rights under Article 102 TFEU have concerned refusals to license. By contrast, Rambus was happy to license its IP rights – it just wanted to be paid a royalty for so doing. It is submitted that the case law provides that a lawsuit by a dominant company in such circumstances could only violate Article 102 TFEU under very unusual conditions, namely that “the action (i) cannot reasonably be considered as an attempt to establish the rights of the undertaking concerned and can therefore only serve to harass the opposite party and (ii) it is conceived in the framework of a plan whose goal is to eliminate competition.”56 Rambus’ patent litigation fulfilled neither of these two cumulative conditions. It was largely successful and intended to obtain royalties for the use of its technology while enabling DRAM manufacturers to pursue their activities on a downstream market where Rambus is absent.
Alleged excessive royalties
The Commission’s preliminary assessment that Rambus may have been charging “excessive” royalties was open to criticism on two grounds. First, it was not supported by mathematical or economic analysis based on any specific methodology. Second, it was based on the provisional conclusion that “save for Rambus’ alleged deceit, JEDEC Members were likely to have designed a ‘patent-free’ standard around Rambus’ patents”.57 Similar accusations by the FTC were quashed by the DC Court of Appeals.
First, according to the two-stage test espoused by the Court in United Brands,58 whether a price is abusive depends firstly on whether it bears any “reasonable relation to the economic value of the product supplied”: if it does not, it then becomes necessary to consider whether the price is either “unfair in itself or when compared to other competing products.”59 Rambus provided both a cost-performance analysis60 and an analysis comparing the license rate sought under standard valuation techniques, including looking at other comparable technologies. Both analyses confirmed that its rate was reasonable.
Moreover, the royalties Rambus sought were lower than the rates of the compulsory licences imposed by the District Court for the Northern District of California.
Second, it is not obvious that the JEDEC Members would have chosen an alternative technology in the “but-for” world. Contrary to the Commission’s preliminary views,61 JEDEC members did take into account issues of quality and cost/performance and did not simply look at alternatives in absolute cost terms.
The Commission noted in its preliminary assessment the existence of alternative technologies.62 However, it is not sufficient to show that alternatives were available, technically viable and commercially viable to prove that JEDEC members would not have chosen Rambus’ technology. The relevant test is whether the alternatives were superior from a cost-performance perspective, taking into account the royalties on the alternative options. And the experts relied upon by Rambus concluded that the Rambus technologies were significantly better on a cost-performance basis.
Particularly because there was no allegation in this case that Rambus misrepresented to JEDEC the qualities of its technology, the very fact that JEDEC did select Rambus’ technology placed the weight of history on Rambus’ side. JEDEC members adopted Rambus’ technology because of its superiority over other technologies.
Moreover, the Commission’s preliminary view that JEDEC members “were likely to have designed a ‘patent-free’ standard”63 failed to take account of the practice of cross-licensing within JEDEC and the fact that JEDEC did incorporate patented technology into its standards.64
Manufacturers may use cross-licensing arrangements as a way of not charging each other royalties for the technology they use, instead choosing to make profits at the manufacturing level – in essence a form of barter. Alternatively, cross-licensing arrangements between manufacturers may involve a payment – this would normally occur when one manufacturer holds a superior patent position to another manufacturer, and the superior patent owner receives royalties in addition to the cross-license rights. For the present discussion, the key point is that a cross-licensing situation is in fact a bargained-for exchange of value and often does include payments of money by one to the other. In both instances, the cross-licensing manufacturers are in the technology market based on the patents they hold.
Even in cross-licensing situations where the patent position is balanced and there are no cash payments back and forth, there is still bargained-for value, and thus inherent payment for the rights to use the patents of the other participants in the so-called “royalty-free” SSO. And while members of a cross-licensing group with a strong or balanced patent position may not have had to pay royalties, any new entrant would have to pay significant royalties. It is important in such cases, that regulators remain neutral as between different business models. If a standard is only royaltyfree for a certain category of participant, but not others, then it is not royalty-free.
In that context, there are in particular strong policy reasons to encourage innovators, like Rambus, rather than suggesting they should receive no royalty to favour another class of stakeholder. Parties with a licensing business model (like Rambus) engage in the same bargained-for exchange as manufacturers with crosslicensing arrangements. There is one difference, namely that innovators with no manufacturing arm, such as Rambus, would only typically receive a royalty payment from the manufacturers. Innovative companies such as Rambus will not participate in SSOs – and the economy as a whole will suffer – if their legitimate interests are not respected.
Shortcomings of commitments decisions
Article 9 commitment decisions do not contain any adjudication of the issues. They simply present the Commission’s preliminary views as set in the statement of objections – before any evidence is submitted in response. Additional submissions would have been filed, additional evidence would likely have been produced, new legal developments might have occurred. The Commission’s position might have changed but any such change would not be reflected in the Decision, except via the fact that the Commission has determined that there was no longer grounds for action on its part and that the proceedings should be brought to an end.
In our view, it is regrettable that commitments decisions do not contain any summary of the defendant’s arguments. There is a serious risk that third party readers may fail to grasp fully the nature of the controversy or even be misled as they are only provided with one side of the arguments, and a preliminary one at that. It is therefore evident that new legal principles may not be created by means of such commitment decisions.
The importance of Rambus resides not so much in the final decision which, as a commitment decision, has limited precedential value, as in the issues raised by the case. The case demonstrates the importance for SSOs to have clear rules in order to avoid disputes on the scope of SSO members’ obligations. Every SSO should enact its own rules, tailored to its needs. As regards disclosure duty, for instance, it is for each SSO to balance any disclosure duty with its inherent administrative costs. The experience of the Rambus case indicates that such rules must be clear, in writing and available to all participants. Ex post reliance on the alleged expectations of a group of competitors, is no substitute for the need of ex ante rules.
How should competition law regulate standard-setting organisations will remain without doubt debated for years to come. The Commission is expected to revise its Horizontal Guidelines in the course of 2010 and will no doubt draw on its experience in the Rambus case and other cases. We will await their publication with interest. We would submit that the key point that the guidelines should make to SSOs is that the onus is on them to put in place clear and effective rules from the outset. If SSOs do so, then antitrust authorities should only need to intervene in exceptional cases.