Sunday, August 25, 2019

Qualcomm argues Quanta changed patent exhaustion law, requiring Qualcomm to adjust terms of dealings with other chipmakers

This is my first follow-up to yesterday's post, which just served to make Qualcomm's Ninth Circuit opening brief in its appeal of the FTC v. Qualcomm ruling public and to discuss the likely effect of the sheer length of that filing on the further schedule.

While it's imperative to see the forest among all the trees, it simply is a huge case, so this first commentary on Qualcomm's opening brief on appeal will focus on the question of chipset-level licensing (Section I of the brief, which ends on page 70 of the PDF).

In order to analyze what Qualcomm is saying in that incredibly important part, I've re-read a couple dozen other documents, most of which are court decisions cited by Qualcomm or others in this context. On this basis I can provide you with a bird's-eye view on the questions this part of the appeal turns on.

First, Qualcomm says Judge Lucy H. Koh of the United States District Court for the Northern District of California determined that the market-dominating cellular baseband chipset maker had an "antitrust duty to deal" with its competitors by granting them an exhaustive SEP license on FRAND terms.

In connection with Qualcomm's motion to stay, the FTC had (in that context, unsuccessfully) argued that the obligation to grant licenses to other chipset makers was not simply and strictly an antitrust duty to deal in a narrow sense, but there was more to it. It was about the overall conduct and its anticompetitive implications. Qualcomm already claimed in connection with its motion to stay enforcement that the FTC wasn't even trying to defend Judge Koh's rationale.

After further research, I think Qualcomm is closer to the truth here than the FTC as far as the nature of Judge Koh's chipset-licensing decision is concerned. In her conclusions of law, Judge Koh "conclude[d] that Qualcomm has an antitrust duty to license its SEPs to rival modem chip suppliers." It couldn't be any clearer.

What the FTC did in the stay context was to broaden the perspective and present a "right for the wrong reasons" kind of theory (except they wouldn't concede the "wrong" part, and would claim that Judge Koh actually had the right reasons because she got everything right at the holistic level).

Further below I'll also address Qualcomm's linkLine attack vector against the alternative holistic theory. But at the start of the analysis it's just about whether or not Judge Koh concluded there was an antitrust duty to deal. Yes, she did, as Qualcomm accurately notes, and unlike Qualcomm I think she was right to do so, though some people's mileage may vary. I hope the FTC will draw the appropriate conclusions from its inability to sell its grand-picture type of theory from a parallel universe to the motions panel, and will now focus on defending what Judge Koh actually wrote as opposed to a strawman.

Second, Qualcomm discusses the chipset licensing part of Judge Koh's duty-to-deal decision as if, under subsequent Supreme Court (Trinko) and Ninth Circuit () precedent, the sole exceptional basis for an antitrust duty to deal--were Aspen Skiing, a case in which the operator of three ski resorts in an area abusively refused to continue to cooperate with the operator of the fourth resort in order to sell a ticket covering all resorts in that area, which for various reasons was a very attractive offering.

While it is true that the Supreme Court and the Ninth Circuit have on different occasions made it clear that an antitrust duty to deal must be a rare exception, and have therefore opposed exceedingly expansive interpretations of Aspen Skiing, I disagree with Qualcomm's "Aspen Skiing or bust" (my words, not theirs, as you might have guessed) position.

Judge Koh's ruling calls the Aspen Skiing precedent "[o]ne circumstance where a monopolist has a duty to deal with rivals" (emphasis added). One. Not "the only one."

This is consistent with the Supreme Court's Verizon v. Trinko (often just named after the latter) decision, which describes Aspen Skiing as "[t]he leading case imposing §2 liability for refusal to deal with competitors" (emphasis added). The leading one. Again, not the only one. After concluding that the telecommunications case before the Supreme Court in Trinko didn't match the Aspen Skiing pattern (for a reason I'll address further below), it held that "[t]raditional antitrust principles do not justify adding the present case to the few existing exceptions from the proposition that there is no duty to aid competitors" as Verizon was already subject to oversight by regulatory agencies and the downside of "antitrust intervention" would in this case have outweighed the "slight benefits."

I wrote last month (in a post that also summarized the Aspen Skiing case that "nowhere does Judge Koh say or suggest that Qualcomm's obligation to extend FRAND licenses to rival chipset makers is like a photographic image of the fact pattern in Aspen." Those cases involve different industries. It's true that Judge Koh heavily relies upon Aspen, whch is not surprising given that it's the closest precedent out there, but I reject any narrowing of the issue to the question of whether or not this is Aspen 2.0 just like I disagree with the FTC's overbroad characterization of what Judge Koh actually held.

Third, the part of Judge Koh's Aspen analysis that Qualcomm focuses on is the question of whether the monopolist abandoned a profitable course of dealing in a way that would sacrifice short-term profits because of long-term benefits from weakening, or preferably getting rid of, a competitor.

In Aspen, the courts apparently assumed that the larger one of the two players would have made more money in the short term, meaning in a given skiing season, by continuing to cooperate with the competitor it wanted to weaken or drive out of business. In her FTC v. Qualcomm decision, Judge Koh points to what Qualcomm once told the IRS about changes to its licensing terms and the underlying strategic rationale.

Qualcomm argues that this still doesn't make it an Aspen case since Qualcomm switched from license deals with other cellular baseband chipset makers to covenants not to sue (more recently, covenants to sue last) because the law on patent exhaustion changed. Qualcomm's lawyers say their client never wanted to grant exhaustive licenses to chipset makers, but--in other words--the goal posts moved and they had to adjust, but since they never wanted to grant such exhaustive licenses, they never abandoned a voluntary course of dealing.

On this one, I agree on what Qualcomm did or did not want to achieve with its earlier chipmaker agreements (there can be no reasonable doubt in my mind about Qualcomm consistently having tried to structure such deals so as to avoid exhaustion), but for a variety of reasons, any single one of which would be self-sufficient, I ultimately disagree with Qualcomm on what that means for the case at hand.

The first one of those reasons is that we're not talking about a change of the legal framework resulting from legislation (such as if the America Invents Act or an international trade agreement or whatever other measure had enshrined patent exhaustion in statutory law, on a basis going beyond the case law that existed at the time). We're not even talking about the Supreme Court deciding to overrule an earlier decision. There was no legislative or judicial about-face here. The June 2008 Quanta v. LG Electronics decision (PDF) refers to roughly 150 years of tradition of the patent exhaustion doctrine and takes the position that what the Supreme Court held in 2008 had been the (case) law all along. Justice Thomas's per curiam says: "Univis governs this case." United States v. Univis Lens Co. is a decision from 1942--approximately the time Qualcomm's founders were born.

Even if Qualcomm didn't know that its attempts to avoid patent exhaustion weren't effective, Qualcomm still signed those license deals voluntarily, and the unanimous (!) Supreme Court decision in Quanta says that patent exhaustion was always meant to cover that kind of situation--for 150 years, or at least since 1942.

Another key fact to consider is that Qualcomm itself has for a long time demanded and obtained exhaustive licenses covering its own chipsets. Why couldn't a court just find that the abandonment-of-voluntary-dealing criterion of Aspen is alternatively also satisfied by a company's own conduct when the shoe is on the other foot (what's good for the goose...)? Judge Koh's findings of fact refer to those licenses, and even in the legal conclusions on chipset level licensing she mentions that "Ericsson also granted Qualcomm a chip-level license."

Again, I can understand why Judge Koh mostly discussed Aspen. But one can apply the philosophy underlying the concept of venire contra factum proprium to an Aspen-like (even if not identical to Aspen) fact pattern. If a monopolist imposes (and the record contains ample evidence that nobody liked to grant Qualcomm an exhaustive chipset-level back-license, but they all had to because of Qualcomm's enormous market power) a form of cooperation on others, then that is, in my view, potentially an even stronger point than the literal Aspen case of abandoning a course of profitable dealing.

The fact that Qualcomm's refusal to extend an exhaustive SEP license to chipmakers while always securing such licenses for its own chips is one of the most important reasons for which I thought from the beginning that the FTC would prevail on the chipset-licensing part. It would be a modification, or even a replacement, of one of the Aspen criteria, but far too logical and reasonable to even be described as "trailblazing" (which is the relatively most positive potential conclusion of the appeal that the motions panel imagines).

MediaTek's excellent amicus curiae brief in support of the FTC's opposition to Qualcomm's motion to stay mentions the above as the fourth one of "at least four prior voluntary courses of dealing" that could satisfy the related Aspen factor:

"(1) Qualcomm's voluntary FRAND commitments to license, which were designed to expand Qualcomm's profits by securing inclusion of Qualcomm technology in industry standards, (2) its previous record of licensing its patents to rivals as well as customers, (3) its exhaustive sales of chips (with patent rights included in the chip sale) in markets where it lacks monopoly power, and (4) its insistence that other SEP owners license their patents back to Qualcomm’s chip business (precisely the requirement to which Qualcomm now objects when applied to Qualcomm)."

With respect to those FRAND commitments, Qualcomm unsurprisingly denies that they count in this context--but the fact of the matter is that Qualcomm made some FRAND pledges that were very clear about that (as the summary judgment in the FTC case held).

All in all, there are some ways to defend the holding that Qualcomm had an antitrust duty to deal, and I hope the FTC and its amici will follow MediaTek's example rather than try to tell the FTC's holistic story that didn't get traction at the stay stage.

Fourth, even though the conduct at issue in Aspen Skiing allegedly sacrificed short-term profits for the sake of long-term competitive harm, I believe the concepts of "short-term" and "long-term" must be viewed differently in the fast-paced cellular baseband chipset market. Here, the short-term opportunity would be a license deal with chipmakers (whom one could charge once they ship their components to a device maker), and the long-term one would be Qualcomm collecting royalties from device makers later on. Aspen and Trinko don't specify a particular period of time between one business opportunity and the other.

Fifth, Qualcomm points to two Ninth Circuit no-duty-to-deal decisions, Metronet v. Qwest (2004) and Aerotec v. Honeywell (2016); the latter cites to the former.

Metronet has a lot in common with Trinko: it's about telecommunications networks, and there is governmental regulation in that field, so there's no reason why antitrust law should be invoked instead. Aerotec was about replacement parts sold to a third-party provider of repair services for a type of aircraft component (auxiliary power units). That's also quite far from SEP licensing. A replacement part can be sold only once in each case; patents are virtual and can be licensed any number of times (provided that the contracts contain safeguards against double-dipping).

Sixth, Qualcomm argues it's more efficient to license at the device level than to engage in multi-level licensing, and in this context complains that while there may be a market in terms of demand for chipset-level licenses, courts might have to set royalty rates. (Arguably, those are two separate, though interrelated, issues.) But it's one thing what rights someone (be it a chipmaker or a device maker) can enforce if all else fails and another thing what may be the most common outcome at the end of the day. A device maker who knows that a chipmaker is entitled to a license may still take a license to Qualcomm's entire portfolio--but the negotiation will be different if the important alternative of component-level licensing exists. And as for judicial determinations of royalty rates, that's simply something that can happen--and has sometimes happened--in connection with SEPs. In fact, potential judicial determinations of SEP royalty rates play a key role in connection with restrictions on SEP holders' access to injunctive relief.

Seventh, while I believe the antitrust duty to deal is defensible here, it's always good to have fallbacks and that's where the FTC and its amici may still argue that the grand evil scheme of Qualcomm's anticompetitive conduct justifies an obligation to extend chipset-level licenses. However, Qualcomm does have a point that the Supreme Court's 2008 LinkLine decision basically says you can't base your antitrust theory on the difference between wholesale and retail prices but you must either show predatory retail prices (here, that would mean predatory pricing by Qualcomm for its chipsets when patent royalties are not considered, which has indeed been identified by the European Commission) or a duty to deal at the wholesale level. Qualcomm is right that (in my words) you need at least one of those pillars, but can't just build a bridge without at least one bridgehead. This doesn't mean that LinkLine necessarily defeats a holistic theory here, but one has to navigate around it--otherwise the FTC's alternative theory might fall into the LinkLine trap.

Supporters of Judge Koh's decision need not worry. Qualcomm's opening brief isn't shockingly strong. It's very well-crafted for sure (with a couple of typos on page 37 presumably being attributable to last-minute edits because they wanted to reference the order granting the motion to stay), but it's pretty much what one could hear from them in the January trial: because their business model works for them, they believe it's legit.

However, the FTC and its amici must and undoubtedly will take this challenge very seriously. The appellate game is different from the bench trial, and Qualcomm's opening brief (I'll comment on the other parts on some other occasions) presents the law and the facts in a way that will definitely be "appealing" to any judges worried about an expansive interpretation of the antitrust laws.

Share with other professionals via LinkedIn: