Thread by @ProfWrightGMU

I am briefly interrupting my Shark Week to share a few thoughts on FTC v. Qualcomm.

First, I’m flattered to have 3 articles relied upon by the court. This highlights my special skill of picking fantastic co-authors, including @geoffmanne @lmedwards_ & Judge Ginsburg. (1/x)

Second, please go read @geoffmanne and @randypicker as well as more critical threads from @ErikHovenkamp and @marklemley. There are a few others that view the decision as *checks notes* the end of the world. You can find those on your own. (2/x)

Disclosure: Qualcomm supports general educational programming @GAI_GMU. Qualcomm does not and has not funded any of the articles cited. I am personally recused from participating in this litigation -- and so am only sharing my personal views. (3/x)

Punchline: I agree with Randy & Geoff the case is correctly decided. The confusing language about what can be inferred from harm to OEMs is not well drafted. But it is dicta. And in context, I read it as less egregious than other commentators for reasons I'll explain below (4/x)

Recall, the FTC’s theory of harm were that QC’s alleged refusal to deal, “exclusive deal” with Apple, and royalty rates, resulted in harm to rivals, which in turn raised their costs (and chip prices), which in turn, harmed competition and consumers. (5/x)

Let's start with refusal to deal. This is the most straightforward part of the opinion. Aspen Skiing carves out an exception to Trinko’s general rule of “no antitrust duty to deal” which can apply under (at least) two conditions. (6/x)

The first is that the D: (1) terminates a voluntary and profitable course of dealing; the second is that (2) it sacrifices short-run profits for longer run profits associated with exclusion of competition. (7/x)

Neither condition held. The FTC offered a 20 year old email on (1), an email that suggested QC never granted exhaustive licenses to rival chip makers. CA9 also found that the switch to OEM licensing was profitable in the short and long run. So much for profit sacrifice. (8/x)

In my view, and as my article points out, this is a pretty straightforward application of Trinko / Aspen. And the court was correct to reject advances to expand exceptions. I do not think this holding is vulnerable on appeal. (9/x)

Much has been made about what the CA9 opinion says about harm to OEMs and harm to competition. Some of it unfair. But opinion is a bit of a mess on this score. But here is how I read it. Let's start with some background. (10/x)

The opinion is valuable in rejecting the view that breach of contract, without more, violates Section 2. The FTC argues it does in the SSO context. I’ve argued elsewhere it does not: papers.ssrn.com/sol3/papers.cf…

But there is an important debate here.

(11/x)

Trinko, NYNEX, and Rambus teach that charging the monopoly price, when the monopoly power is lawfully acquired, is lawful. If one breaches a FRAND commitment (or other K) to charge the monopoly price, that does not change things ABSENT a showing of harm to competition. (12/x)

Deception can result in harm to competition, as alleged in Rambus / Broadcom. But Rambus requires a P distinguish between a price increase resulting from lawfully acquired monopoly power one caused by harm to the competitive process. P bears that burden. (13/x)

In Rambus, the FTC failed that burden (and even dismissed that it was theirs). The same burden applies when P alleges breach of K constitutes a Section 2 violation. P bears the same burden to show a price increase was a function of harm to competition & not something else. (14/x)

Recall the FTC’s theory of harm was harm to chipset rivals resulted in harm to OEMs and final consumers. Much has been made about the CA9’s language (e.g., p. 30) indicating that harm to OEMs doesn't count. Poorly drafted. But let’s read that in context. (15/x)

FTC bears the burden of showing harm to competition. CA9 repeatedly finds no harm to rivals or the competitive process. And they repeatedly make the point that harm to the rival, under the FTC’s OWN theory, is necessary (but not sufficient.) (16/x)

E.g., at p. 36, “The FTC does not explain how QC’s alleged breach … itself impairs the opportunities of rivals.” CA9 emphasizes that competitors practice QC’s SEPs royalty-free and that “FTC identifies no such harm to competition.” (17/x)

The OEM discussion must be read in context. If your reading requires CA9 to fail to understand that OEMs are customers, it is worth double-checking. The Court cites both Rambus and NYNEX in explaining CORRECTLY that without more, this is not "harm in the antitrust sense." (18/x)

Finally, it is worth noting that discussion is dicta.
CA9 found persuasive QC's procompetitive justification for licensing at the OEM level only.

A final word about exclusivity. The FTC alleged QC’s 2013 deal w Apple foreclosed a substantial share of the CDMA market.(19/x)

The supposed victim, Intel, won Apple’s business the next year. Reasonable minds can differ here, but the CA9 decision lines up much more closely with exclusive dealing jurisprudence than did Judge Koh's decision. (20/x)

I read this largely as a court rejecting a theory broadly acknowledged as testing the limits of Sec 2. Nothing wrong that. But winning in that context usually requires more than a good theory. It requires quality proof of harm. There was none here. (21/x).

It will also usually require the D to be unable to, or fail to, provide a sensible business justification for its conduct. In this case, QC succeeded in explaining why the change to OEM-level licensing made economic sense. (22/x)

In short, the prospects on appeal for the FTC are dim. And while I'd love a SCOTUS exclusive dealing or RTD opinion (perhaps Aspen Skiing is now beyond the outer boundary of Section 2...), I'm skeptical SCOTUS would see the CA9 opinion as cert-worthy. (23/x).

There are also some fascinating dynamics at play in whether the FTC petitions for cert, both in terms of internal dynamics as well as how the DOJ might respond. But none of it makes cert likely. But I will save that for later.

Back to Shark Week. END.

(24/24)