Sunday, March 8, 2020

FTC's "No License-No Chips" theory in Qualcomm case may have been too aggressive to be affirmed

For the technology industry, the question of whether rival chipset makers are entitled to a FRAND license to Qualcomm's cellular standard-essential patents (SEPs) is the commercially most relevant part of the case. I just wrote about it in the previous post (the first one of a series of three posts on FTC v. Qualcomm as a follow-up to last month's appellate hearing). But chipset-level licensing has become a question of contract law, with or without a potential antitrust add-on. From a pure antitrust point of view, "No License-No Chips" is the centerpiece of FTC v. Qualcomm. Exclusive dealing, which allegedly delayed Intel's entry into the premium LTE chipset market, is more of a historic thing than something that would make much of a difference for the future (which is why I'm not even going to discuss it here).

"No License-No Chips" is Qualcomm's practice of separately collecting patent royalties on its chip. If Qualcomm just sold its chips without previously signing a patent license agreement, its patent rights would be exhausted. Normally, tech products are sold and the price covers the intellectual property involved. But Qualcomm has always done it differently in the cellular baseband market (though not in other business areas), and very successfully so.

Trial testimony was consistent and clear: many companies bowed to Qualcomm's royalty demands only because they needed Qualcomm's chips. Then they were stuck with license agreements that also applied to any devices they'd ship with baseband chips made by other companies than Qualcomm.

The FTC argued--and Judge Lucy H. Koh agreed--that this resulted in a surcharge: patent royalties were inflated by shifting a part of the monopoly chipset price to Qualcomm's licensing division.

Reasonable people can disagree on whether this practice constitutes an illegal practice of the "raising rivals' costs" (RRC) kind as the FTC alleged and Judge Koh held.

The Ninth Circuit panel that heard the case last month consisted of one judge who views the FTC case as a "house of cards" (Judge Consuelo Callahan), one judge who was unconvinced of the FTC's argument to say the least (Judge Johnnie Rawlinson), and Judge Stephen Murphy III, who is normally a district judge himself (Eastern District of Michigan) and a bit more deferential to Judge Koh's ruling than the two circuit judges. All in all, a reversal of the "No License-No Chips" part is significantly more likely than affirmance.

Other trial reports than mine made it sound like the FTC had already lost. Apart from the fact that the court took no position on chipset-level licensing (see my previous post), I believe the jury--in this case, the panel--is still out. Yes, Judge Callahan may very well get support from Judge Rawlinson, in which case a unanimous decision (with Judge Murphy II likely joining them) is possible. But even when I watched the recording of the hearing again this week, I felt that Judge Rawlinson wasn't near-certain to reverse. The hearing made the FTC's position look weaker than it is. Qualcomm was represented by a world-class lawyer, Thomas Goldstein, who made it extremely easy for the court to agree with him, while an academic representing the FTC, Brian Fletcher, made it unnecessarily hard for the judges to follow. And Judge Callahan was totally in the tank for Qualcomm and hostile to the FTC ("the house of cards starts to fall"), even arguing that Qualcomm dealing with Apple would be like David and Goliath, but there was no one on the bench who would have supported the FTC. Still, it's not a given that Judge Callahan will find another judge to form a majority with her on everything. On some aspects, probably. But not necessarily on all.

Raising rivals' costs (RRC) is a recognized theory in the monopoly-maintenance context. RRC has major advantages for monopolists over predatory pricing--above all, it's profitable in the short term, not only in the long term when others may have exited the market or may have been marginalized. Back in 1983, an FTC working paper already mentioned patentable inventions as a potential vehicle for RRC schemes.

Qualcomm's appellate strategy emphasized that the FTC had allegedly failed to prove a surcharge. But at the hearing, things appeared to go even better for Qualcomm: the judges seemed to doubt that even if there had been supra-FRAND royalties, those would have constituted a tax on rivals as opposed to simply effective patent monetization.

The FTC's "No License-No Chips" theory was optimized for judges who favor a rather expansive view of antitrust law: liberals, typically. But not for conservatives. In my next post I'll get back to the political implications, given that there could be an en banc review by 11 judges (not three), with a Democratic majority being very likely.

When asked for case law above the level of a district court that would support the "No License-No Chips" part of Judge Koh's ruling, the FTC's special counsel, Professor Fletcher, gave the 1922 United Shoe Machinery v. United States decision by the Supreme Court. The fact that it's almost 100 years old (old law can still be good law) is less of a concern than the following aspects:

  1. The patent royalty collected "on shoes operated upon by competing machines" was just part of a set of seven "restrictive [contractual] provisions," which were "tied together."

  2. As Qualcomm noted in its reply brief on appeal, United Shoe just like the Caldera case from the District of Utah regarding Microsoft's per-device operating system license fee are different from Qualcomm's business model because United Shoe and Microsoft were fully paid when competing products were used, while Qualcomm receives only its full patent royalty, but the chipset purchase price goes to a rival.

  3. The district court decision in United Shoe (which the Supreme Court affirmed) explicitly allowed the offender to license its patents on "reasonable nondiscriminatory" (RAND has a long tradition in antitrust law) terms.

The circuit judges didn't seem to be concerned about profit shifting from one Qualcomm division (chipsets) to another (licensing). And their thinking apparently is that any concerns over excessive royalties need to be addressed in patent infringement cases. The FTC's special counsel argued that companies can't challenge Qualcomm's terms because they need uninterrupted chipset supplies. But there was no indication of the court buying that argument to the extent that it would deem antitrust intervention necessary.

Upon further reflection, I have concluded that the FTC probably should have taken a more conservative (in every sense of the word) approach to "No License-No Chips." Qualcomm does have some arguments that aren't easily countered. For instance, Qualcomm has also concluded license agreements on its terms with companies that weren't dependent on their chipsets at the relevant time. Those companies might simply have accepted those terms because any royalty determination by a court would be very much based on an analysis of the terms of comparable agreements. There might have been other reasons. And an excessive royalty isn't less excessive just because some people accept it. But those are circumstances that really make it hard to prove any wrongdoing on Qualcomm's part. Being greedy is what companies are supposed to be.

Instead of arguing that Qualcomm's practice of charging patent royalties separately from, and as a precondition to, chipset sales is generally illegal, the FTC could--and with a view to conservative judges should--have focused on only the combination of that practice with rebates. Those rebates were granted in exchange for exclusivity and/or co-marketing commitments. With rebates involved, there is a case that competitors couldn't sell their products on a commercially viable basis. The FTC's case does raise the issue of those rebates (incentive payments), but just as a factor that exacerbates the situation. What I feel is that the FTC's basic "No License-No Chips" scenario should have included rebates.

Last fall, the American Antitrust Institute (AAI) and Public Knowledge (PK) submitted an amicus curiae brief in support of the FTC that illustrated the impact of "No License-No Chips" in conjunction with rebates (this post continues below the document):

The AAI/PK brief explained the numbers as follows:

"The illustration shows that if Qualcomm had charged a monopoly chipset price and a FRAND royalty, as were its right and voluntary contractual commitment, respectively, then Qualcomm would have faced pressure to drop its $30 all-in price to match its competitors’ $21 all-in price. However, because Qualcomm imposed a surcharge on its SEP royalty along with an exclusivity rebate, Qualcomm was able to maintain its all-in monopoly price of $30 while forcing up the all-in costs of its rivals' products to the higher level of $31, at which they would not be viable. As a result, entry is deterred."

Professor Fletcher also mentioned those rebates, but almost as an afterthought, or an add-on. The case (going back to the original complaint, long before Professor Fletcher represented the FTC) would have been narrower--but more solid--if the FTC had focused on the scenario the AAI/PK amicus brief outlined so well.

Otherwise, the FTC needs a court that is prepared to adopt novel theories that are somewhat aggressive and ambitious.

Share with other professionals via LinkedIn: