Thursday, January 10, 2019

Qualcomm offered Samsung and MediaTek a "covenant to sue last" instead of a chipset license

No one ever doubted that Qualcomm is an innovator. Any disagreements anyone has with them relate exclusively to how much leverage they can rightfully get from their patents and how they can leverage their market position in premium baseband chipsets. But Qualcomm's lawyers also seem quite innovative in their field. Only Qualcomm could come up with the idea of offering a "covenant to sue last" to companies who requested a license.

"Covenant to sue last." Not a straightforward license. Not a covenant not to sue (which at least means you'll be left alone). But a promise to sue someone only if all other alleged infringers have previously been sued. It's like saying: "We very much do reserve the right to come after you, but we'll deal with your competitors first. You may have to defend yourselves at some point, but if that happens, you'll at least be our most favored adversary timingwise. Now, in exchange for this tremendous favor, we want you to commit to all sorts of things that render you less competitive."

As I mentioned in my report on Huawei's testimony on Day 2 of the FTC v. Qualcomm antitrust trial in the Northern District of California, Qualcomm appears to be extremely afraid of patent exhaustion. Patent exhaustion--which protects the downstream from assertions of patents already licensed somewhere upstream--can result from an authorized sale (such as selling a chipset that substantially embodies certain patented inventions) as well as from a license (such as licensing a chipset maker who then sells the licensed product to device makers). Qualcomm is concerned about both exhaustion avenues, but in this post it's all about the second scenario (licensing).

As my Day 2 report also explained, the Federal Circuit held in its 2009 TransCore v. Electronic Transaction Consultants (ETC) opinion that exhaustion can be rooted not only in what is narrowly defined as a license but also by a covenant not to sue, which is a license by any other name given that a patent is, simply put, a license to sue infringers, so if you give up the right to sue, the beneficiary of such a covenant is effectively licensed.

It makes a lot of sense to focus on the actual, practical effects of a commercial agreement rather than hairsplitting and purely formalistic aspects. A lot depends on the judges. Some are more receptive to commercial arguments than others.

So Qualcomm and others realized at some point that even a covenant not to sue was not nearly as reliable an exhaustion-avoidance strategy as they once thought. But they still wanted to be able to do non-exhaustive deals in order to hamper competition from rival chipset makers through terms that came with restrictions (on whom they were allowed to sell products to) and reporting requirements (so Qualcomm would get sensitive information about a competitor's business). We hear from Lenovo on Day 1 (see the MediaTek paragraph of this post) that they had to fear that Qualcomm would enjoin MediaTek from selling chipsets to Lenovo the moment Lenovo wasn't going to have a license to Qualcomm's patent portfolio.

The name of the game for Qualcomm was and is, therefore, that it seeks to thread the needle and do convenient (and apparently anticompetitive) deals with rival chipset makers while navigating around patent exhaustion.

On Tuesday (Day 3), Andrew Hong of Samsung's chipset business and Intel's general manager Aicha Evans testified that Qualcomm simply refused to grant exhaustive patent licenses to them. In Samsung's case, it was partly about Samsung itself but the largest part of the testimony related to a contemplated joint venture, named Dragonfly, between Samsung and some Japanese companies. According to the testimony, the number one reason for which Dragonfly failed to materialize was that the various Dragonfly partners assumed NTT DoCoMo, which already had a license deal with Qualcomm, could provide the Dragonfly JV with a chipset-related license to Qualcomm's standard-essential patents. When Qualcomm made it clear that it wouldn't do this (Mr. Hong says Qualcomm told him they weren't going to support something that would have made Samsung a baseband chipset competitor within a year, given that it otherwise takes several years to enter that market), Project Dragonfly was over before it began.

Thanks to a particularly important one of Judge Koh's unsealing decisions, the FTC's pretrial brief tells us what Qualcomm actually did offer Samsung's chipset business at some point. An email by a former Qualcomm president said the following:

"[W]e were also asked for licenses by Intel and TI at a minimum, probably others (e.g., Samsung, Mediatek) as well, and we refused to enter into anything other than a non-exhaustive covenant (or covenant to sue last in the case of SS and MT)." (emphasis added)

No company that wants a real license is going to be satisfied with a "covenant to sue last." And while Qualcomm presumably told the companies it offered such a deal that in practical terms they were going to be fine since there's a huge number of companies in the world and there would always be some whom Qualcomm wouldn't sue, the problem is still that if you're general counsel of a chipset maker and your CEO asks you whether the company is reliably safe from patent assertions, you must answer: "It's not safe because Qualcomm could at some point decide to simply sue the whole world and then we'll be hit like anybody else." And, at any rate, such a covenant does not solve the problem of indemnification. Samsung's Mr. Hong said in his testimony that "in [his] experience the IP indemnification clause tends be one of the biggest items debated" in chipset supply negotiations.

Intel's Aicha Evans explained the same issue just from a different angle. She said that device makers obviously have to look at their total cost: what they pay for the chipset as well as any patent licensing costs related to it. If a chipset was licensed, they'd pay a price and that would be it. However, when there's a situation where Intel sells a chip and knows that a competitor (who'd actually like to drive Intel out of this business) is then going to collect patent royalties from the device maker that Intel doesn't even know (certainly not beforehand and, due to confidentiality clauses, usually not even afterwards), it makes it extremely hard for both Intel and the device maker to actually calculate their costs. It discourages purchases and investment.

During this trial, some Qualcomm-internal communications have been shown according to which Qualcomm's management always feared that licenses to rival chipset makers would have a devastating effect on Qualcomm's patent licensing business. Here's an interesting passage (just recently unsealed) from the FTC's trial brief:

"Qualcomm's internal documents recognize the impact that offering competitors FRAND licenses would have on Qualcomm's ability to secure elevated royalties from OEMs. In 2005, Qualcomm's Marvin Blecker explained that making a license available to a chip competitor would impair Qualcomm's ability to collect high royalties from OEM customers: 'we absolutely cannot give a chip supplier a full license to our IP with pass through rights to his customers as that would have the potential of severely impacting our subscriber licensing program.' [...] Qualcomm's views were unchanged in 2015, when it concluded that granting a FRAND license to Intel 'would destroy the whole current QTL [licensing] business.'"

The decision on Qualcomm's obligation under antitrust law (with respect to two FRAND pledges, this has already been resolved favorably under contract law) won't destroy Qualcomm's licensing business in its entirety, but it would make it considerably harder for Qualcomm to collect supra-FRAND royalties, and it would make it harder for Qualcomm to avoid competition on the merits with other chipset makers.

Four companies making chipsets have testified by now. Samsung's chipset division, Huawei (now specifically referring to the part of Mrs. Yu's testimony that related to chipset licensing), and Intel definitely supported the FTC. MediaTek tried to, but wasn't nearly as effective during the public part of the testimony. The sealed part related to some contract between Qualcomm and MediaTek, and maybe that document spoke for itself. But even if one is skeptical about MediaTek's testimony (again, not because of the intention, but because the witness wasn't as tactically shrewd as the other witnesses, most of whom are lawyers), the testimony the court heard from chipset makers appeared overwhelmingly favorable to the FTC's cause.

Share with other professionals via LinkedIn: