Tuesday, November 18, 2008

Latest Broadcom/Qualcomm Skirmish

This morning, the IP SIG of the telecom council had one of its occasional meetings to review major developments in IP law. IP is obviously very important to the local industry, whether it be Qualcomm’s patent licensing business model or Ron Katznelson’s current efforts to promote patent reform.

I couldn’t make the IP SIG, but on the agenda were three cases. One case was Broadcom v. Qualcomm. Which one? you ask — after all, there are so many. In this case, the lawyers talked about the case that brought findings of legal misconduct and forced Lou Lupin to fall on his sword. Another case scheduled for the IP SIG meeting was the ruling in Quanta v. LG, where the SCOTUS set new standards for doctrine of “patent exhaustion.”

As it turns out, the Quanta ruling plays a prominent role in Broadcom’s latest lawsuit against Qualcomm, filed October 8 in federal court. This filing is just the latest in itsongoing series of litigation intended to force Qualcomm to license its patents on terms that Broadcom finds acceptable: a royalty-free cross license.

When I saw the latest Broadcom v.Qualcomm case, I asked one of San Diego’s leading IP lawyers, USD professor David McGowan, if he would be willing to comment for this blog. I thought of David because he had analyzed the Quanta case for the leading patent blog. I already knew him because we’d met at an open source research conference, and we share a common interest (particularly in IP law) at the intersection of law and economics.

As a special guest commentary, here is Prof. McGowan’s analysis of the original Broadcom complaint. Paragraph numbers refer to the original complaint as filed in the Pacer database system.

Inexhaustible: The Latest Broadcom / Qualcomm Skirmish
By David McGowan
Lyle L. Jones Professor of Competition and Innovation Law
University of San Diego School of Law

Broadcom recently sued Qualcomm for a declaration that certain of Qualcomm’s alleged licensing practices misuse Qualcomm’s patents. The suit is both predictable and plausible in light of the Supreme Court’s recent Quanta opinion.

Based on Qualcomm’s description of its licensing practices, the district court’s decision will likely turn on how it resolves a question left open in Quanta. From an economic point of view I hope the court reads Quanta narrowly. Such a reading would tend to favor Qualcomm. There is a plausible formal case for Broadcom’s claims, however, so the case must be taken seriously.

The complaint alleges Qualcomm double-dips by charging handset manufacturers a royalty on both chipsets and handsets. Qualcomm may earn chipset royalties through sales to handset manufacturers or from licenses to competing chipset manufacturers such as Broadcom. It also charges a royalty directly on handsets.

Broadcom claims the Quanta case holds that a patentee may not charge downstream royalties beyond the first sale of its technology. This claim implies Qualcomm may either charge handset manufacturers a royalty for the handset or for a chipset (directly from Qualcomm or indirectly from Broadcom) but not both. Broadcom is asking the court to declare that it need not pay royalties to Qualcomm when Broadcom sells chipsets to handset manufacturers who pay Qualcomm royalties on handsets. (¶5)

McgowandBroadcom alleges Qualcomm’s double dipping drains from the industry money that could be used to foster innovation and competition. I find this claim implausible. Qualcomm innovates and there is no general reason to favor innovation by Broadcom or handset manufacturers over innovation by Qualcomm. This complaint raises distributional issues, not questions of dynamic efficiency.

Broadcom also alleges Qualcomm’s licenses reduce competition. It offers two theories in support of this claim. The first is that Broadcom must either take Qualcomm licenses or risk infringement action. That theory is doubly flawed. It confuses an effect on a competitor, Broadcom, with an effect on competition generally and it has nothing to do with the double dipping claim, which is the only one to which the Quanta holding might apply. If Broadcom operates under a cloud of infringement that is because Qualcomm has exclusive rights in its inventions not because of its licensing model. Only the serendipitous fact that the licenses were negotiated before Quanta was decided offers Broadcom any maneuvering room here, and that is a legal not an economic point.

The same points hold for the allegation that Qualcomm’s practices deter entry and increase costs. (¶29) That charge would be no different if Qualcomm suppressed all manufacturing other than its own and took all its royalties in chips. It seems to me a charge against the patent system rather than tiered licensing.

The second theory is that Qualcomm licenses favor its products over competitors’ products. Little is said about this claim so it is hard to know what to make of it. At one level it is not clear why Qualcomm would want to do this. Its patents mean it can obtain revenue by royalty from Broadcom or sale to handset manufacturers. Economically it should be indifferent between the two revenues streams. Indeed, if Broadcom is a lower-cost producer than Qualcomm, the latter would maximize profits by charging a high royalty rather than producing itself.

The complaint also refers to Qualcomm using its patents to leverage royalties, and leverage may be what the second theory has in mind. (¶24) It is not clear what Broadcom means by this reference. The facts alleged do not support the typical leverage claim, which is that a firm with market power in one market will use it to extend the duration of power in that market or to extend its power to another market. Rather, this appears to be a case in which single monopoly rent theory has a lot of force.

If handset manufacturers must practice Qualcomm inventions they must deal with Qualcomm, which implies some royalty. It does not imply an infinite royalty, however (Qualcomm wants handset sales so it is in its interest to set a royalty that allows handset manufacturers competitive returns on their investments). Still less does it imply that Qualcomm would earn less charging a single handset royalty than charging a royalty on both handsets and chipsets.

Chipsets and phones are close complements used in a fixed ratio. Basic rational actor assumptions imply Qualcomm will get as much money as it can; they also imply Qualcomm can get that amount at any level of production (either chips or handsets) and cannot get more by charging x at one level and y at another.

Here is a simplified illustration supporting the latter point. Suppose it is the dominant strategy for handset manufacturers to set prices at the average reservation price consumers place on the handset (given the cost of a contract with a carrier). Let’s say the average reservation price is $10, handset manufacturers need to earn $6 per handset, and the only input cost is chipset royalties. In theory QC could charge $4 as a handset royalty or charge $3 for a chipset and $1 as a handset royalty (or $2 and $2, of course). What it could not do is charge $3 for a handset royalty and $2 for a chipset royalty. (There are various qualifications to this logic in the antitrust literature but BC’s complaint does not seem to implicate any of them.)

No doubt this analysis is too crude to capture all the facts of the case. One would need to know the deal structures better than I do to offer more precise analysis. The analysis does raise significant questions, however, which the complaint does not answer.

It is hard to say what is likely to happen. Quanta dealt with an unconditional license and Qualcomm has said its licenses are expressly conditional. The Court did not rule on such licenses. The district court therefore is not bound to rule for Broadcom. If one read the Supreme Court as sending the signal that patentees should only be able to engage in one transaction per product, though, one could extend Quanta to reach the result Broadcom seeks.

As I mentioned I favor the narrow reading. I approach the problem this way. First, as a default matter, do we expect parties or courts reach more socially efficient bargains for the exploitation of inventions? In my view, both economic theory and experience favor parties as the default bargainers. Living markets as they do, they have much better information and much keener incentives to get things right than a generalist judge faced with unfamiliar information spun vigorously by competing advocates.

Second, should courts deviate from that default? The answer may be yes but only if there is a good reason. Absent such a reason the arguments favoring the default rule carry through. In some cases (price fixing for example) there will be an economic reason to abandon the default but there seem to be no such reason here. In other cases precedent might compel a court to abandon a default. Here it does not because the licenses at issue seem to be materially different from those at issue in Quanta. (Nothing in Quanta calls the default itself into question.)

Third, if there is no substantive reason to abandon a sensible default position and precedent does not require abandonment should a court abandon it? To me the question answers itself. Once you get the default right, in my view, you need a reason to move away from it, and there is no such reason here.

Saturday, November 15, 2008

Admitting the obvious: Qualcomm knifes UMB

Admitting the obvious, CEO Paul Jacobs told analysts Thursday that Qualcomm is giving up on UMB. The Reuters story (referenced by Unstrung) is the only account I could find:
NEW YORK (Reuters) - Qualcomm Inc, seeking to cut costs in the face of slowing demand for cell phones, has stopped developing a next-generation wireless technology called Ultra Mobile Broadband.

The wireless chip maker will put its resources into a rival high-speed technology called Long Term Evolution, which some of its major customers, such as Verizon Wireless, have backed, said Qualcomm Chief Executive Paul Jacobs.
As I noted in April, Qualcomm’s hope of getting its UMB adopted died when one of its two major cdma2000 customers in the US, Verizon Wireless, went with LTE.

It’s funny how Qualcomm chose to announce this. The only 2008 press release mentioning UMB is a February announcement of their planned UMB/LTE chipsets (presumably now LTE only). In a (Qualcomm-enabled) search for UMB on the Qualcomm website, 6 of the first 15 links went to missing documents (presumably pro-UMB marketing materials that have now been removed). I grabbed the various UMB white papers (still there) for posterity.

The 2006 annual report (from December 2006) predicted UMB would be commercialized in 2009. In March 2007, Qualcomm announced a “complete solution” for UMB in time for CTIA.

With UMB dead, tweaks to EVDO will not be enough to justify keeping 3GPP2 alive forever; given their overlapping membership, perhaps they can negotiate a friendly takeover by 3GPP. At some point, the raison d’ĂȘtre for CDMA Development Group will also disappear.

As I also noted in April, early cdmaOne (then IS-95) fan (then Vodafone CEO) Arun Sarun urged the industry to favor LTE over WiMax. Qualcomm is in the odd position — for the first time in more than 15 years — of not having its own horse in the mobile standards race, but instead watching its previous rivals duke it out.

Still, Qualcomm’s new alliance is clear. With HSDPA/USDPA, Qualcomm has been contributing to improvements in W-CDMA and getting closer to the 3GPP telecom crowd, and now with the Nokia patent dispute settled, Qualcomm is a respected (and perhaps even respectable) figure in Sophia-Antipolis.

Meanwhile, Qualcomm is like Nokia in that it would just like Intel and WiMax to go away or find a tiny niche (like small-town last-mile distribution). Most of the world’s carriers are in the same camp. The one obvious exception is Qualcomm’s other major CDMA carrier, Sprint, with a huge bet on WiMax (bankrolled by Intel). I wonder if we will see other evidence of increasing distance between Qualcomm and Sprint.

Conversely, Qualcomm is proudly providing the chip for T-Mobile’s new gPhone. Perhaps Qualcomm will gain enough WCDMA/LTE share to make up for the shrinking CDMA market.

The Reuters story also said Qualcomm hopes to use its (ARM core) Snapdragon CPU to expand revenues beyond cellphones to pocket computers or laptops. This seems like a tough slog, as PC sales are also expected to decline and many firms (beyond Intel) will be trying to gain market share in anything that needs a CPU.

Monday, November 10, 2008

Valuing the Nokia settlement

On TheStreet, Marerk Fuchs dissects Qualcomm’s quarterly results released Thursday — specifically the higher than expected Q4 and the lower than expected predicted earnings for Q1.

His conclusion? Analysts are overlooking the boost from a onetime $560m payment by Nokia (as part of their patent settlement) and that without it, Qualcomm’s revenues have already started to decline.

According to its Q3 filings, Nokia is paying Qualcomm $2.3b (or $2.5b) in back and prepaid royalties.

In its earnings call transcript (as reported by Seeking Alpha), CEO Paul Jacobs said that Qualcomm received a $2.5b payment in October. He explained the profits as follows:
The agreement includes among other things, the non-refundable upfront payment of $2.50 billion received in early October. Ongoing royalties and the assignment of patents valued at $1. 8 billion and recorded in intangible assets.

At the end of fiscal 2008, unearned revenue attributable to the upfront consideration of the Nokia agreement was $3.9 billion, which will be recognized over the approximately remaining 14 year term of the license agreement. The value of the patents assigned to us last month will be amortized on a straight line basis to licensing cost of sales over their estimated useful life of approximately 15 years. The fourth quarter of fiscal 2008 includes six quarters of revenue amortization and the resumption of royalties for the second half of fiscal 2008.'
I have not seen anyone do a model of this to see how it compares to what Qualcomm would have received at list price. Nokia clearly bought down its royalties by assigning patents to Qualcomm, and also by its promise of non-assertion of its patents against Qualcomm. (In the rest of the world, we call that a cross-license).

I haven’t seen anyone analyze the value of the patents Nokia assigned (or won’t assert) to Qualcomm. Nokia and Qualcomm were spending lots of money on economic and IP experts, so perhaps with their current truce, there aren’t many expert left interested in studying this — or maybe just no one left being paid to study this.

Saturday, November 8, 2008

MetroPCS, Leap finally form alliance

A year ago, San Diego-based Leap Wireless rebuffed the attempt by its younger rival MetroPCS to buy it. The two firms together are still smaller than the #5 carrier, Alltel, although they would technically leap to #5 with 8+ million subs if they merged because Alltel is becoming part of Verizon Wireless.

It seemed like the decision not to merge was driven by Leap’s pride. It is possible that they were worried about the disruption and risks of a merger, but both firms’ need for economies of scale was compelling.

However, with the Big 4 offering unlimited service plans earlier this year — directly competing with the Cricket/MetroPCS raison d’etre — their need for scale is even greater. Both companies need buying and operating efficiencies, plus the ability to offer roaming to their customers.

Although we’ve been an all-CDMA household for nearly 10 years, roaming is the main reason our household has not switched to Metro or Cricket. We split our time between San JosĂ© (where Metro has been prominent for years) and San Diego (where Leap finally began offering service in December 2006).

The roaming issue is now solved. In September, the two firms agreed to settle their litigation, swapped spectrum and also agreed to a 10 year roaming deal. The two firms lack the scale to get decent terms from Alltel, Sprint or Verizon.

The other shoe dropped this week. On Thursday, the two firms announced that customers with all but their lowest priced plans can roam for free to the other’s network. The deal covers 300 markets.

However, the two carriers lack a national footprint, without New York, Boston, Chicago or Washington, DC. (No subscribers from the new Obama administration). MetroPCS does plan to enter Boston and NYC next year.

The roaming deal is a good mid-term expedient for both carriers. They are particularly well positioned to sell to sell to college students (or low income types) who lack a landline and thus need an unlimited use wireless plan for the lowest possible price.