Today the European Commission has concluded a 4 year long antitrust investigation into anti-competitive business behaviour of Qualcomm against other players in the market, and has fined the company €242 million for abusing its market dominance in 3G baseband chipsets during the period of 2009 to 2011.

The investigation was formally opened on July 16th 2015 and particularly looked at Qualcomm’s behaviour in the late 3G/UMTS and early 4G era where it held a commanding lead over other vendors in supplying modem chipsets.

The Commission concluded that Qualcomm had engaged in predatory pricing of three chipsets with evidence that the company had aimed to strategically push out and eliminate new contenders in the market, with a specific mention of Icera.

The Commission goes into more detail in regards to Qualcomm’s pricing behaviour in the mid-2009 to mid-2011 period where it concluded that it sold UMTS chipsets below cost to Huawei and ZTE, two important customers, with the goal of eliminating Icera.

Icera was an up-and-coming UMTS and LTE vendor which had started to see success in the market, and ended up being acquired by Nvidia with plans of integrating the technology into the Tegra line-up of SoCs, one product of this venture ending up being the Tegra 4i. The Tegra 4i unfortunately saw very little success among vendors in the market even though the chipset was technically equivalent to the Snapdragon 800/801 SoCs at the time. Nvidia ended up shuttering the division in 2015 due to a lack of success.

Qualcomm has communicated that the company is planning to appeal the finding.

The fine comes shortly after a recent scathing ruling in the US where the FTC had accused the company of similar anti-competitive behaviour breaching antitrust laws, and several years of scrutiny and fines by several regulatory agencies of various countries around the world.

Related Reading:

Source: European Commission Press Release

POST A COMMENT

29 Comments

View All Comments

  • Mr Perfect - Thursday, July 18, 2019 - link

    I'd be interested on your take of this article at Ars Technica. https://arstechnica.com/tech-policy/2019/05/how-qu... Reply
  • Raqia - Thursday, July 18, 2019 - link

    You asked for it so here goes:

    Tim Lee means well but he doesn't understand the issue and copied the one sided and controversial conclusion of the judge almost verbatim without reading Qualcomm's arguments or considering the rebuttals by a former and sitting FTC commissioners, the US's chief antitrust lawyer, and a number of other experts who came out against the FTC. Since the judgement and that article, the DoJ, DoD and DoE have come out against this decision for a combination of threats to competition, innovation and national security.

    The judge first issued a partial summary judgement declaring that licensing of Qualcomm's SEPs should be tied to modems. This ignores that Qualcomm's essential patents extend to various other components of both the device and cellular towers. Even though modems are important accelerators for Qualcomm's cellular standards (it owns most of the most important and seminal 3G/4G LTE patents), much of the complexity lies in the difficult real-time software (which could be run on CPUs or FPGAs as well) used to interface with towers and is not inherently implemented by the hardware. There are also entirely different standards not owned by Qualcomm that run on most modems. It's analogous to how Quake could use a CPU software renderer or make calls to a 3D accelerator for part of its execution, but licensing for Quake is generally not tied to the 3D card and accelerators are a separate cost from IP. These standards aren't just f-you gate keeping mechanisms but have substantial value in conserving spectrum, improving reliability and improving handset power.

    Koh ruling also runs against another ruling by a Texas district court which declared that it is perfectly valid to assess licensing at the device level. Before someone cries foul at "taking a percentage" of the phone's value in licensing, Qualcomm's IP licensing fees are capped so it could equally well be interpreted as a discount off of a flat fee for lower cost OEMs who often make less intensive use of standards but still benefit from interfacing with common infrastructure. Also recall that software licensing is often based on the number of cpu cores or the amount of ram in a system and this kind of IP price discrimination occurs all the time when students or seniors get a discount for movies or professional software.

    Koh further concludes that Qualcomm's "no license no chips" policy violates their FRAND commitments. The fundamental right granted to patent holders is the right to exclude others from the practice of an invention, and Qualcomm is well within their rights to partially exercise this right when a counterparty has no license by denying them sales of their implementations. Counterparties were still mostly free to reverse engineer their own implementation or buy them from another vendor and Qualcomm did nothing to prevent this although it's within its rights to do so. The issue of FRAND is entirely a contractually defined issue, with early adopters assuming more risk but determining the ultimate rate for IP that later adopters who don't assume as much risk ultimately accept as precedent for what constitutes FRAND. What would have constituted a FRAND violation in this case would have been using its modems to leverage higher SEPs but this was not proved to have happened during the trial. Her attempt to use antitrust statutes is largely based on qualitative criteria and in no way prove consumer or competitive harm. In fact the exact opposite occurred under Qualcomm's licensing scheme.

    Finally, her remedy to renegotiate contracts was imposed on all of Qualcomm's contracts when the evidence brought against them applied only to the narrow "premium standalone modem" business where only Apple was a major customer. There was ample evidence that Apple's decision to use Qualcomm was based on its technical merits and the terms in the contract with Qualcomm were demanded by Apple and not Qualcomm. It undoubtedly had more power than the suppliers in this market and ultimately illegally withheld licensing and tortiously interfered with its contract manufacturers. Outside of this market, there was vibrant competition from multiple implementers including Huawei, Samsung, Mediatek, and up until recently Intel. There was no evidence to justy the remedies she ordered for business outside of this narrow market.

    In the end, Koh's judgement reads like an angry freshman's term paper where she made up her mind before listening to any evidence, or understanding either the patents or the market outcomes. I am not saying Qualcomm is faultless or never engaged in business malpractice, but I think the spirit of recent regulatory action is deeply misguided and dangerous. I am paying a lot of attention to this issue because the long term consequences of such misruling can be very severe to incentives for R&D that paves the way for future tech.
    Reply
  • Raqia - Thursday, July 18, 2019 - link

    Also note that Koh made her partial summary judgement prior to hearing the trial and without understanding the scope of Qualcomm's patents. Furthermore, IP licensing came out to an effective $7.50 an iPhone for what's probably the most complex commonly used consumer technologies (the modem accelerator for this software sets you back another $15 or so). Reply
  • Khato - Friday, July 19, 2019 - link

    Japan - The scope of the antitrust investigation in Japan was extremely limited. It was specifically with respect to IP agreements between Japanese smartphone manufacturers and Qualcomm. Basically, whether Qualcomm abused its market position to force these manufacturers to allow it royalty free access to manufacturer IP by using SoC/modem pricing as leverage. The end conclusion after years of investigation being that there wasn't adequate evidence to prove these claims.

    Taiwan - Sure, Qualcomm reached a 'settlement' to remove the black mark of being found guilty of antitrust violations. If they were actually not guilty though then that settlement wouldn't have included "commitments that ensure good-faith negotiations for the benefit of licensees and SEP owners" and TFTC keeping the $93 million fine.

    Apple-Intel - You know why Intel had to sell their modem below cost? It's the same way Qualcomm has choked all other competition to death. They just ensure that the licensing costs + competitor modem cost is equal to or greater than what they're charging for their own modem.

    Basically, Qualcomm is at least on par with Microsoft in terms of abusing their market position to stifle out competition. Their stranglehold on CDMA in particular is the only reason their SoCs have been so successful. The most obvious example of this abuse being the numerous instances of manufacturers offering two versions of a smartphone, one with the SoC they want to use that's sold everywhere else, and then the one using a Qualcomm SoC for the CDMA market.
    Reply
  • Raqia - Saturday, July 20, 2019 - link

    You haven't shown that their dealings in Japan are any different from their dealings elsewhere, and 9 years of hearings tells of an extensive and exhaustive examination of the issue. Further, the Taiwan decision was sharply split with only a narrow vote in favor of a fine initially, and the commissioners in agreement voted for a measure that completely misunderstood the nature of Qualcomm's inventions.

    In the US, Apple and Intel didn't pay licensing for over 2 years from early 2017 to 2019 and Intel sold modems below cost and Qualcomm did not so what you say doesn't apply at all (in fact Intel was dumping modems.) Two years is also an extremely long time in the semiconductor industry during which both much larger companies brought to bear their extensive resources to try to bring a viable implementation to market while enjoying the illegal privilege of not paying licensing for standards that they didn't invent. They still failed. Also, despite what you say, Samsung, Mediatek, and Huawei have viable competing solutions to Qualcomm's exactly because Qualcomm chose to declare their inventions into standards rather than holding them closed as Intel does with x86. All of these things were not admitted into evidence by judge Koh in the FTC trial who hampered proper fact finding by strictly imposing an unfair timeframe that didn't include Apple's switch to Intel modems while allowing the FTC to use evidence outside her initially defined timeframe.

    Both Koh, the initial majority of Taiwanese FTC and a lot of the public misunderstand the nature of Qualcomm's primary business. The primary thing that Qualcomm sells now and historically are cellular systems that define how handsets electromagnetically interface with towers. This is primarily a software and algorithm related notion rather than a hardware one, and it is apt to describe it as a complex electromagnetic grammar that is standard in the industry. Qualcomm only got into hardware when no one else could figure out how to efficiently implement their systems, but modems only serve as hardware accelerators for their patented grammar on power constrained client devices and are an entirely separate notion and cost from the patented IP that they play a part in implementing. Few people confound software costs with the cost of the hardware it runs on, but confusion is rampant in most discussions about Qualcomm, for example many peoples' surprise that licensing is demanded of handsets using non-Qualcomm modems.
    Reply
  • Raqia - Saturday, July 20, 2019 - link

    Your supposition that Qualcomm chipsets are only used for CDMA markets is incorrect. Samsung used exclusively Exynos for all markets in the Galaxy S6:

    https://www.google.com/amp/s/www.androidauthority....

    So it is capable of not using Qualcomm for any market it chooses. It uses Qualcomm because they make good chips and they want to be apprised of outside developments, to ensure supply, and to make good use of foundry capacity. Samsung and many others choose Qualcomm chipsets exclusively for tablet solutions with minimal sales related to a cellular licensing component when the Exynos or Kirin could be substituted because Snapdragons are compelling solutions.
    Reply
  • AshlayW - Tuesday, July 23, 2019 - link

    All major companies are scumbahs that chase profits over anything... It's human nature to be greedy.

    Yes, even AMD would abuse its position if it could.

    Good that governments still have the control (and the will) to keep these companies in check. What in 50 years? When evil corporations rule the world and have more control than government, all in the name of profit...
    Reply
  • AshlayW - Tuesday, July 23, 2019 - link

    Scumbags* Reply
  • godhascome - Thursday, July 25, 2019 - link

    The Orphic school, a puzzle religion that began in Thrace and spread to Greece in the fifth century BCE, https://www.godhascome.com/">GOLDEN AGE held comparable convictions about the beginning of man, similarly naming the ages with metals Reply

Log in

Don't have an account? Sign up now