Tuesday, September 3, 2019

The big elephant in the FTC v. Qualcomm antitrust room: compulsory licensing of standard-essential patents

This is the second follow-up to the publication of Qualcomm's Ninth Circuit opening appellate brief in the FTC antitrust case, and a direct follow-up to the post on Qualcomm's argument against an antitrust duty to deal to extend exhaustive standard-essential patent (SEP) licenses to rival chipset makers.

For a recap, Qualcomm's opening brief

  • says Judge Lucy H. Koh of the United States District Court for the Northern District of California mistakenly based her holding of an antitrust duty to deal (with competitors) on the Aspen Skiing Supreme Court ruling, with Qualcomm attacking the abandonment of profitable voluntary conduct more fiercely than the other factors, pointing to the Ninth Circuit's post-Aspen no-duty-to-deal findings in Metronet and Aerotec; and

  • tackling the FTC's holistic (in other words, "grand evil scheme") alternative reasoning (the next best thing to a right-for-the-wrong-reasons argument) by pointing to the Supreme Court's 2008 LinkLine decision that says you can't prove an antitrust violation based on a wholesale-retail margin, but you need to prove either a duty to deal on the wholesale side or predatory pricing at the retail level--and the price-squeeze theory disallowed by LinkLine can't be avoided by mere labels (on that part, I agree, though Qualcomm's unique business model does raise issues beyond LinkLine's scope in my view).

Unlike at the stay stage, the FTC now has to defend Judge Koh's actual reasoning (though right-for-the-wrong-reasons fallbacks don't hurt). Contrary to what Qualcomm naturally asserts, one can shoehorn the case at hand into the Aspen pattern, but it's not trivial as there are gaps that one needs to bridge.

Even I as an outspoken critic of Qualcomm's business model don't doubt for a second that Qualcomm's pre-Quanta agreements with competing cellular baseband chipset makers weren't intended to be exhaustive. But does that make them involuntary? Qualcomm can't claim that it signed those agreements at gunpoint or to comply with regulations. The Supreme Court didn't say in Quanta that it overruled, or expanded, an older doctrine. Instead, the highest court in the land said that this had been the law for 150 years, and even the (relatively speaking) more recent case that governed Quanta is from the 1940s, long before there was such a company as Qualcomm. I readily believe Qualcomm that they had misinterpreted the law, but does that make the agreements they signed until then involuntary?

Additionally or alternatively, one can hold against Qualcomm that it entered into FRAND licensing declarations such as the two with respect to which Judge Koh entered summary judgment that they undoubtedly required Qualcomm to extend FRAND licenses to rival chipset makers. Again, Qualcomm's self-delusionary opinion might have been that it could circumvent that obligation, but we're talking about a voluntary course of dealing. (To be clear, I'm not saying that breach of contract is an antitrust violation.)

Then there's the "good for the goose" alternative to a conventional voluntary-conduct finding. Qualcomm itself required everyone, even companies like Ericsson that never did so, to grant exhaustive licenses to them that benefited their chipset customers. I was personally in that Paris courtroom where Apple avoided a French preliminary injunction against an iPhone model because Samsung's license agreement with Qualcomm covered Apple as Qualcomm's customer. What Qualcomm did in its dealings with other patent holders may very well serve as a substitute for a more conventional "abandonment of profitable voluntary course of dealing" theory.

But I wouldn't have used the verb "to shoehorn" if the Aspen-to-Qualcomm pattern matching was extremely simple and comfortable. I still believe it's doable and defensible, and better than the LinkLine pitfall, but it does take some effort.

Besides Aspen shoehorning and navigating around LinkLine, there's actually a third approach. The FTC won't be able to go down that road because it's politically difficult enough for the FTC to even just defend its trial win, so there's no realistic way the FTC could agree on a more aggressive theory. However, the FTC's future amici curiae will be free to argue what they want, and amici often go beyond what the party they support says. And even if nobody raised that issue here because everyone believes it's more conservative to just stick to Aspen (a shoe that can fit, but doesn't fit like a glove), there will be other cases down the road, involving either Qualcomm or other SEP holders, such as that Continental v. Avanci case presently before Judge Koh (Qualcomm is one of Avanci's contributors, although the focus there is mostly on Nokia).

That tertium is the big elephant in the room: straightforward compulsory licensing.

Yeah, I know that the mere thought of it stirs up a hornets' nest. I'm not one to shy from controversy and have advocated compulsory SEP licensing for a long time. This is a litigation and policy blog, not a popularity contest.

In 2014, the Congressional Research Service (a department of the Library of Congress) published a paper by an unnamed visiting scholar, Compulsory Licensing of Patented Inventions (PDF). While significant parts of the document focus on international trade with a particular emphasis on pharmaceutical patents, it provides a pretty good overview. It addresses two kinds of compulsory licensing:

  • compulsory licensing in the narrow sense of license agreements having to be concluded due to legal obligations, and

  • compulsory licensing as the net effect of an inability to obtain and enforce injunctive relief (with courts awarding an ongoing royalty).

The paper notes that "for some, the distinction between a compulsory license and an 'ongoing royalty' is one without a difference." In some cases that may be right, but when it's about component-level SEP licenses, the question is whether or not a license is exhaustive. Given the Supreme Court's generally comprehensive and expansive take on patent exhaustion, there's a strong argument that a component maker paying a court-ordered ongoing royalty (and, thanks to the denial of an injunction, being allowed to continue to implemented the patented technique) must be able to sell the effectively-licensed product, and the downstream must benefit from exhaustion. I have even raised the question on this blog whether Qualcomm's promise not to assert patents against chipset makers triggers promissory estoppel and whether such promissory estoppel should also have an exhaustive effect. Arguably, those theories would be reconcilable with the Lexmark approach and the fact that courts believe a covenant not to sue is a license by any other name for the purposes of an exhaustion analysis.

Courts around the globe have struggled with SEP injunctions for a long time. If any jurisdiction had allowed any SEP holder to preclude anyone else from implementing a standard by insisting on the exclusionary rights that normally go with a patent, there would have been chaos:

  • Some patent holders, especially trolls (since no one can countersue them for infringement), would have extracted way supra-FRAND royalties, and as a result of royalty stacking, it would have become economically impossible to implement any standards affected by such behavior.

  • Some other patent holders would have leveraged their SEPs to shut out any new entrants.

The standardization system would have broken down. The economy at large would have suffered massive damage.

But there's a conundrum: patents are exclusionary rights. They are, by definition, monopolies. However, the specific problem with SEPs is that if you had, say, 500 truly essential patents reading on a standard, and if those 500 were held by, say, 50 different patent holders, then any single one (!) of those 50 patent holders could preclude anyone else from practicing the standard, be it for the purpose of exortion or that of monopolization.

U.S. and EU (especially, but not only, German) law went in completely different directions as courts recognized the problem and were looking for a solution:

  • U.S. courts typically threw out or didn't even reach antitrust claims, but they held SEP holders to their FRAND licensing pledges on the basis of contract law.

  • EU, particularly German, courts didn't have that option, however, for lack of third-party beneficiary rights. The contributors to a standard-setting process make their FRAND promises to standard-setting organizations (SSOs). But implementers of a given standard are third parties, and EU (particularly German) law doesn't recognize the right of a third-party beneficiary to enforce an agreement between others. In order to poke a hole into the exclusionary rights of patent holders (in that regard, Germany is even out of compliance with EU law), they resorted to an old Roman-law doctrine ("dolo agit qui petit quod statim redditurus est") that means you're not allowed to claim what you'd have to return immediately for some other legal reason. And that other legal reason (for which one is not allowed to seek injunctions over SEPs) would be an antitrust violation.

    The original German Orange-Book-Standard doctrine was based on a case where there wasn't even a FRAND pledge. It was a de facto standard. Later, the Court of Justice of the EU brought a bit more balance into that analysis with its Huawei v. ZTE decision, which is also based on antitrust law (based on "settled case-law that the exercise of an exclusive right linked to an intellectual-property right by the proprietor may, in exceptional circumstances, involve abusive conduct for the purposes of Article 102 TFEU," which is the unilateral-conduct paragraph of EU law).

I was never completely satisfied with either jurisdiction's problem-solving approach--and I don't just mean the details of how to implement it, but structural limitations:

  • Why not have the best of both worlds?

    • U.S. courts can solve the SEP injunction problem on the basis of contract law whenever there is a FRAND commitment, but what if there isn't because one or more companies created a standard without making a FRAND pledge? Then you need to resort to antitrust law. Even when you don't need it, why not let it serve as another deterrent to prevent SEP abuse?

    • EU/German judges can't legislate from the bench and enrich their laws with enforcement rights for third-party beneficiaries. But in cases where a FRAND pledge has been made, a breach of that commitment could be held against a SEP abuser in addition to all other considerations, such as on the basis of considering it particularly malicious and deceptive conduct to make a public statement that encourages others to implement a standard only to renege on it later.

  • The actual starting point of the problem is neither unilateral abusive conduct after the adoption of a standard nor a FRAND pledge. It starts with the fact that a group of companies sits at a table to agree on a standard that is, by definition, exclusionary: exclusionary with respect to alternative technologies. Of course, standardization can be beneficial. But without a FRAND pledge, it wouldn't be. That's why some regulators such as the EU Commission make it pretty clear in their horizontal-cooperation guidelines that standards development may be a great thing, but only if all contributors make FRAND licensing promises. In other words, the upside of standard-setting may outweight the downside of inevitably excluding alternative technologies--but while you may be allowed to exclude competing technologies, regulators are not going to let you exclude competing companies (they must at least have access to a license on FRAND terms).

    A FRAND pledge is not just a contract like any other. It's a contractual commitment that is needed to legalize what would otherwise be an illegal cartel.

    Exclusionary use of SEPs, be it through the enforcement of injunctive relief or a refusal to extend licenses, is not just unilateral conduct. It's an abuse of rights that are rooted in a process that involves the creation of a cartel.

In Aspen Skiing, you had one company operating three resorts in an area, and the smaller player having just one. But the big bully on the block simply had the financial strength to build in part and acquire in part those three resorts--as opposed to having gotten into that high-leverage position by means of initially participating in a cartel, legalizing it through a FRAND pledge, and later reneging on the pledge.

What I miss in the FTC v. Qualcomm component-level licensing context is the cartel-based background of FRAND-pledged SEPs. Again, the whole reason why regulators require the FRAND pledge (because otherwise they might go after standard-setters even before they ever assert an essential patent) is to ensure that other companies are not excluded from the market. Apart from the often rather clear language of FRAND pledges, it follows from that "exclude technologies, but don't exclude companies" logic that SEPs must be licensed to all comers at all levels of the supply chain.

The CJEU in Huawei v. ZTE and the DOJ in the Motorola Mobility contexts made the availability of injunctive relief over SEPs subject to an analysis of both sides' conduct: hold-up vs. hold-out. That's how they sought to strike a balance between the exclusive nature of patent rights and the anticompetitive effects of SEP abuse. It's one of the most fundamental conflicts one can imagine: antitrust law is anti-monopoly law, while patents are monopolies. That's why compulsory licensing is sometimes inevitable.

In the EU, Microsoft was subjected to compulsory IP licensing with respect to a Windows network protocol long before the more recent wave of SEP cases. And even in the U.S., there's an interesting precedent that I only learned about from the Congressional research paper mentioned above: the May 2002 final judgment in United States v. 3D Systems Corp. and DTM Corp., a merger case (okay, the third major pillar of competition law--after unilateral conduct and cartels--that comes up here). The DOJ's Antitrust Division approved an acquisition subject to a divestiture, and in order to make sure that patent enforcement wouldn't render the divestiture useless from a competition point of view, the primary defendant was obligated to extend IP licenses to a future acquirer of the unit to be divested.

It's not a Supreme Court duty-to-deal precedent like Aspen Skiing, but it is very interesting. And it happened under a Republican administration (Bush 43).

This is already a fairly long blog post, and all I actually wanted was to add a broader perspective to the question of compulsory SEP licensing to competing component makers. Considering the reason for which the participants in standard-setting processes feel forced to make a FRAND licensing promise in the first place, there really is a pressing need under competition law to ensure that someone like Qualcomm must grant SEP licenses on FRAND terms to rival chipset makers, regardless of whether or not there had been some voluntary and profitable course of dealing that was discontinued at some point.

Share with other professionals via LinkedIn: