Friday, May 12, 2017

FTC says Qualcomm is "using both carrots and sticks together" in dealings with device makers

The Federal Trade Commission (FTC) has just responded to Qualcomm's motion to dismiss its antitrust complaint in the Northern District of California (this post continues below the document):

17-05-12 FTC Opposition to QCOM Motion to Dismiss by Florian Mueller on Scribd

I published Qualcomm's motion in early April (toward the end of this blog post) and wrote at the time that Qualcomm would hardly get rid of the FTC case altogether. It's much less unlikely that the court might want the FTC to be a bit more specific in one respect or another, but in today's opposition filing the FTC has just explained very well how Qualcomm's exploitation of its dual monopoly (which does appear unique in some ways, at least to me) falls within established patterns of anticompetitive conduct.

The FTC filing helps Judge Lucy Koh, who will (after Qualcomm's forthcoming reply brief and a hearing) have to rule on the motion, in various ways. A particularly valuable aspect of the FTC brief is that it identifies and describes certain patterns of conduct and, basically, asks the court not to get confused by Qualcomm's labeling tactics. Qualcomm's argument that its conduct is legal depends heavily on terminology. That's why the very first parapgraph of today's filing by the FTC argues that Qualcomm's supra-FRAND patent royalties are a "tax" no matter how much Qualcomm insists that it's just "royalties" on its patents.

The FTC notes that "[a]ntitrust law eschews such formal labels and instead examines market realities." That's true. But no field of law is more closely related to politics and policy than antitrust law, and just like the fight over terminology is often the decisive battle in political matters, labels do matter in a competition context.

What makes the intersection of antitrust and intellectual property law so very interesting is an inherent dichotomy: intellectual property rights are monopolies (limited to 20 years in the specific case of patents), but antitrust law is an anti-monopoly law. If every legitimate patent royalty was considered a "tax" imposed by a "monopolist," antitrust law would apply very broadly, but patent rights would be devalued. However, if every (actual) monopolist could just make an end run around competition law by labeling a monopoly tax as a "patent royalty," patent rights would serve as a (powerful) pretext and anyone wielding a patent would be immune to antitrust scrutiny.

The FTC argues that Qualcomm's customers "accept elevated patent royalties they otherwise would refuse" only because they are forced "to negotiate in the shadow of Qualcomm's threat to withhold chips." As a result, "Qualcomm collects far more in royalties than other licensors in the industry with comparable patent portfolios."

I've been trying to find out as much as possible from the outside about Qualcomm's royalties (and how they compare to the license fees other wireless standard-essential patent holders receive). If you consider that Wall Street believes Nokia gets approximately $2 from Apple per device, and if you compare that number to the $5 per-device discount from Qualcomm that an arbitration panel awarded BlackBerry (strongly suggesting that it's merely a fraction of what BlackBerry paid), or Qualcomm's reduction of a quarterly forecast by $500 million after Apple stopped its payments, then the FTC's allegation is very plausible. And unlike me, the FTC does have access to a lot of confidential information. That's the beauty of investigative authority.

I just saw that Samsung has submitted an amicus brief in support of the FTC brief, so for now I'll keep my commentary on the FTC filing short, though I may very well write a follow-up next week or so. Just a few more observations I wanted to share immediately:

  • The FTC has the royalty base problem figured out:

    "Moreover, Qualcomm assumes that if its rate was ever FRAND, it must remain FRAND today because it has not changed. But the complaint alleges that 'handsets today offer a number of features' not offered by older handsets, and 'many of Qualcomm’s patents related to CDMA technology have expired.' [..] Thus, a 5% royalty on a 2006 phone is not economically equivalent to a 5% royalty on a 2017 smartphone." (emphasis added)

  • Just like the FTC refuses to play Qualcomm's labeling game, it also seeks to (re)focus the court's analysis on economic realities with respect to Qualcomm's agreements with companies like Apple under which Qualcomm effectively pays back some of the royalties it receives, provided that the recipients of such paybacks behave in certain ways or refrain from behaving in other ways:

    "Qualcomm argues that its payments show that it lacked the power to coerce OEMs into accepting anticompetitive license terms. [...] But the notion that using both carrots and sticks together undermines a finding of monopoly power or anticompetitive conduct runs counter to both economic logic and legal precedent." (emphasis added)

  • With respect to Qualcomm's refusal to grant patent licenses to competitors (i.e., rival chip makers), the FTC explains that a refusal to extend patent licenses to competitors may be acceptable under other circumstances but here, in light of a FRAND licensing commitment and when all aspects of Qualcomm's behavior are seen together, it's anticompetitive. In my initial reaction (of early April) I had also disagreed with Qualcomm's emphasis on litigation against competitors and noted that sometimes they do sue. So does the FTC, which points to Qualcomm's patent assertions against Broadcom. Admittedly, it's been a while since that happened, but it is not and should not be forgotten.

There appears to be some amicus brief activity. Yesterday, a lawyer for an industry body (ACT | The App Association) filed a notice of appearance, and Samsung has (as I mentioned further above) just filed a brief, which I'll study now and, if it contains some interesting new information, blog about immediately.

Share with other professionals via LinkedIn: