Given the amount of money at stake, I'm surprised that Carnegie Mellon University v. Marvell Technology Group, Ltd., Civil Action No. 09-290 (W.D. Pa.) hasn't received more attention than it has so far--although the Patent Damages Blog has a good discussion of the most recent installment, the court's March 31, 2014 decision on enhanced damages, interest, and ongoing royalties. (Of course, given that the stakes are so high, I imagine that an appeal is inevitable.) I'll try to be brief in highlighting the major points relating to remedies, as culled from the court's opinions of August 24, 2012 (denying a motion to exclude the plaintiff's damages expert's testimony); September 23, 2013 (denying the defendant's motion for judgment as a matter of law); January 24, 2014 (denying the defendant's motion for judgment on the equitable defense of laches); and March 31, 2014 (on enhanced damages, interest, and ongoing royalties). All of these are available on Westlaw, and a link to the most recent opinion is also available from the aforementioned post on the Patent Damages Blog.
The plaintiff Carnegie Mellon University (CMU) is the owner of two patents "generally directed to the method of sequence detection in high density magnetic recording sequence detectors." The defendants are Marvell Technology Group, Ltd. and Marvell Semiconductor, Inc. ("Marvell"). The case was tried in late 2012 and a jury returned a verdict "in favor of CMU on infringement, validity, and wilfulness, awarding damages in the amount $1,169,140.271.00." CMU's damages theory was premised in large part on Marvell's use of the patents in suit during its sales cycle:
Marvell sells its chips through a lengthy, expensive sales cycle during which Marvell “invest[s] significant resources with each potential customer without any assurance of sales to that customer.” At the end of a given sales cycle, Marvell achieves a “design win” if “the customer decides to go into production” with Marvell and actually does so. Such a design win is generally a winner-take-all affair in the HDD industry, which typically results in the winner becoming the exclusive supplier for the customer's specific hard drive or generation of hard drives. Simulation programs are used throughout the sales cycle to formulate the concepts and basic designs, research and develop new products, refine and evaluate chip designs before incurring the cost of setting the chips in silicon, among other ways. Once the simulation programs have satisfactory results, an engineering sample chip is created in Asia and sent back to domestic offices for continued refinement, testing, and evaluation with Marvell engineers and its customers. If the customer is satisfied and places an order, the chips are put into volume production and manufactured in Asia. Following Marvell's “design win,” it would become the exclusive supplier for a customer's specific hard drive or generation of hard drives. A portion of these chips comes back to the United States through the chain of commerce in hard drives and/or laptops. Dr. Bajorek testified, the “sales cycle sequence” takes “three or four years to complete” and “with the exception of the chip making, which is made by a foundry in Taiwan, all the activities related to designing, simulating, designing [ sic ], testing, evaluating, qualifying the chips by Marvell as well as by its customers occurs in the United States.”
Much of this sales cycle activity uses the methods claimed in the CMU Patents. Through evidence regarding this sales cycle, CMU proved Marvell's direct infringement by: (1) use of the method in the KavcicViterbi simulator, which all chips are tested against to evaluate their performance and used to develop greater SNR gain; (2) use of the method by the Chip Simulators (KavcicPP, MNP, EMNP, and NLD Simulators) during the research, development, design, qualification and testing phases for the corresponding chips; and (3) use of the method in Accused Chips as engineering samples, sometimes called “golden chips,” that Marvell uses during the sales cycle. CMU demonstrated that these three scenarios of direct infringement all arise in the United States during Marvell's sales cycle. CMU also successfully argued contributory infringement and induced infringement through Marvell's customers' use of the patented method in the Accused Chips in the United States. CMU has never asserted infringement against Marvell for any use of its patented methods which did not occur in the United States, nor does it seek damages for instances of foreign infringement.
The question then was how to "quantify[ ] the volume or the value of the 'use' of the patented methods during the sales cycle. "[Q]uantifying a per use fee in this case is nearly impossible, as the patented method is literally run hundreds of millions of times per second." Instead, the court permitted the plaintiff's expert to quantify:
the use of the patented methods during this sales cycle based on a reasonable royalty for the sales that arose from the sales cycle. CMU proffered this theory and the Court ruled on several occasions that Marvell's sales could be an appropriate metric for assessing the value of the use of the patented methods in the U.S. by Marvell and its customers. . . .
Marvell has repeatedly challenged the Court's decision to allow CMU to value Marvell's use of the CMU Patents by considering all the chips that were sold under the aforementioned sales cycle. . . . [T]he Court reiterated, in response to [Marvell's] “emergency” motion:
CMU intends to prove that the alleged infringing method is used during Marvell's sales cycle, which is performed here in the United States, where both its engineers and customers are located. CMU seeks damages for this sales cycle infringement by claiming a reasonably royalty rate on all of the chips that are produced during this sales cycle and purchased based on the result of said cycle.
To be clear, CMU does not seek damages from alleged infringement of the Accused Chips that are never used in the United States, because the Court has held the extra-territorial sales are not infringing, it seeks damages on the infringement from the U.S. based sales cycle, and has chosen to quantify these damages by applying a per chip royalty rate on all Accused Chips produced under the sales cycle. Marvell will have a full opportunity at trial to argue that this quantification is unreasonable. . . .
As previously discussed, CMU seeks damages based on the sales that are created because of Marvell's infringement during the sales cycle. Marvell argues for the first time in its reply brief on its Rule 50(b) Motion that CMU has not entered enough evidence to show that sales took place in the United States in light of [Power Integrations, Inc. v. Fairchild Semiconductor Int'l, Inc., 711 F.3d 1348 (Fed. Cir. 2013)].
First and foremost, Marvell seems to conflate several issues in maintaining that CMU must prove that the sales were made in the United States. Marvell's liability is not predicated on “sales,” but rather on “use” under 35 U.S.C. § 271(a). Under CMU's theory, as proffered, it would seem that sales would not necessarily need to be made in the United States because as long as the sales were a direct result of Marvell's infringement in the United States during the United States sales cycle, the foreign sales would remain an appropriate component to value domestic infringement. Second, Marvell introduced no evidence at trial that any aspect of its sales took place outside the United States. Third, Power Integrations' damages theory was predicated on lost profits, while this matter focuses on finding a reasonable royalty or negotiated fee for the use of CMU's methods. . . .
I'm still trying to decide for myself whether this theory makes sense, though I'm inclined to think it does. On the one hand, but for the infringement, Marvell would have sought a license for its use in the U.S. of the patented methods, and arguably the amount of that license would have reflected Marvell's expected post-sales cycle benefits (even if those benefits themselves were attributable in part to foreign sales). On the other hand, this might seem like a way to avoid the general rule that you can't recover damages for the infringement of a U.S. patent based on conduct that occurs abroad. Perhaps the case demonstrates the artificiality of the rule against extraterritorial damages, but I'm not sure what we can do about that.
In any event, the court also deflected Marvell's various challenges (improper apportionment, etc.) to plaintiff's expert Catherine Lawton's methodology for arriving at the $0.50 per chip figure. Multiplied by a huge volume of chips, you get a huge royalty: $1,169,140,271. That's "billion" with a "b".
The March 31, 2014 opinion further awarded supplemental damages for Marvell's continued sales of accused chips from July 29, 2012 to November 2, 2013 in the amount of $79,550,288, and considered additional motions for interest, enhanced damages, and postjudgment royalties. As for prejudgment interest, the court stated that even though U.S. courts ordinarily award prejudgment interest following General Motors Corp. v. Devex Corp., 461 U.S. 648 (1983), "district courts retain discretion to limit or deny prejudgment interest in certain circumstances, such as where the patentee has been responsible for undue delays in prosecuting the lawsuit." More specifically:
The Third Circuit advises that a district court should consider and balance the following factors before exercising its discretion to award prejudgment interest:
Having fully considered the totality of the circumstances in this case, in light of the aforementioned precedent from the Supreme Court and Federal Circuit, demanding a showing of delays by the patentee and prejudice to the infringer, see Gen. Motors Corp., 461 U.S. at 655–56 and Crystal Semiconductor Corp., 246 F.3d at 1361–62, each of which were found satisfied in the context of the Court's laches decision, ( see Docket No. 920), and after weighing the factors set forth by the Court of Appeals for the Third Circuit, see Feather, 711 F.2d at 540, the Court finds that CMU's exceptional delays in prosecuting this case and its corresponding failure to timely investigate the infringement allegations fully justify denying CMU's request for prejudgment interest.
Thus, even though the court had denied Marvell's motion for judgment on the equitable defense of laches, CMU's unreasonable delays in asserting its claims against Marvell did have a major impact in the courts' decision to deny prejudgment interest.
Second, the court awarded enhanced damages in the amount of $279,181,106.00, equal to 1.23 times the actual damages award including the supplemental damages. (In its September 23, 2013, the court had affirmed the jury's advisory verdict that the infringement was willful.) Applying the Read v. Portec factors, the court concluded that a relatively modest (percentage-wise, that is) award was appropriate, again relying in part on CMU's delay in bringing suit as well as on the closeness of the damages issues, among other things.
Third, the court awarded postjudgment interest at a rate of 0.14% compounded annually.
Fourth, applying the eBay factors, the court denied CMU a permanent injunction in lieu of ongoing postjudgment royalties. In particular, the court found "unproven" CMU's theory that it faced irreparable harm because it might not be able to collect its damages judgment against Marvell. The decision not to award an injunction isn't all that surprising, in my view, but the part that did surprise me a bit was the court's decision to award ongoing postjudgment royalties at the same rate ($0.50) as the prejudgment royalties. (I have previously argued that this is, economically, the correct rule, but the Federal Circuit and other district courts often have awarded ongoing royalties at a higher rate. See, e.g., my discussion here.) The court's discussion of this issue was interesting for what it said about CMU's licensing strategy (or lack thereof):
At this point, CMU is a prevailing patentee which has been awarded one-hundred (100%) percent of the compensation it sought through a substantial award of compensatory damages, accounting for every $0.50 in Marvell's “excess profits” which its expert attributed to the inclusion of the patented methods in each chip. It has also received an additional twenty-three percent (23%) compensation on the pre-judgment royalties, which equates to an additional $0.115 per Accused Chip. All told, CMU has received one hundred twenty-three percent (123%) compensation for the pre-judgment infringement, or $0.615 per Accused Chip, for a judgment award totaling in excess of $1.5 billion.While this award can be appropriately characterized as very significant, the return on investment by CMU is at best described as staggering. The record is uncontested that CMU developed these patents through its DSSC program with public grants and the support of corporate donors. Its own financial investment in the research and development of the patents has never been quantified during these proceedings but under any measure would be considered extraordinarily insignificant when compared to the present award.Given the valuation of these “must have” and “industry standard” patents, CMU's marketing efforts, or lack thereof, is stunning to the Court. CMU then ignored repeated allegations of infringement from the inventors and took nearly six full years to decide whether to pursue litigation, never following up with Marvell during this time period. It received similar allegations about other entities but to this Court's knowledge has not pursued them. And, despite the verdict, CMU has not negotiated licenses with any other third parties.
From the Court's view, CMU appears largely satisfied with the exceptional returns on its minimal financial investments in these patents and only questions whether it will be able to collect the judgment against Marvell. Accordingly, the Court believes that CMU would accept an ongoing royalty of $0.50 per Accused Chip for the continuing infringement rather than the requested $1.50 per Accused Chip or some rate between those two figures, including the effective rate of $0.615 per Accused Chip which it has been awarded for the pre-judgment infringement and willfulness. CMU effectively seeks to have the Court impose a higher rate than the jury's assessment to punish Marvell for its willfulness rather than arriving at that figure after undertaking a comprehensive Georgia–Pacific analysis.
Perhaps the decision not to award a higher ongoing royalty is consistent with Ted Sichelman's thesis that in calculating damages courts (ideally) should consider what amount of damages would be necessary to preserve the incentive to invent . . . .
Thanks for posting this summary. Without having read the case, my initial thought is that CMU’s theory makes sense to far as extra-territoriality goes, but I am more skeptical about the per chip royalty base. If the patented technology is used in the sales cycles, then the royalty should be based on its value as compared to the best NIA in the sales cycle. In principle that might be very little. Suppose the patented method allowed faster simulation runs, which shortened the sales cycle by a month. I can’t see that one month on a 4 year cycle would be worth $1b to Marvell’s customers. (I'm looking forward to reading the case myself - but I have a pile of exams to grade first.)ReplyDelete