How Will the $2.18 Billion Verdict in VLSI Technologies v. Intel Impact Future Patent Valuations?

VLSI Technology, originally an integrated circuit manufacturer in the 1990s, is today a non-practicing, patent-owning entity, an affiliate of Fortress Investment Group. VLSI no longer makes or sells any products, but in 2019 it became the owner of U.S. Patent Nos. 7,523,373 and 7,725,759. Each of these patents relates to increasing the power and speed of computer processors. VLSI sued Intel, claiming that almost 1 billion microchips sold by Intel infringed VLSI’s patents. The jury ultimately agreed, entering a verdict of $2.18 billion dollars for VLSI. In simple terms, this amounted to an award of approximately $2 per chip (on a chip that likely is sold for $200 or less per chip based on third-party estimates).

VLSI’s Damages Theory

The exact damages calculations that VLSI presented are not part of the public record. However, the judge did allow the public to listen in on portions of the trial, and several attendees and participating attorneys have publicly commented on VLSI’s damages theory. Specifically, VLSI’s damages theory has been described as a “hedonistic regression” or “incremental benefit theory.” This type of analysis is a technique that is more theoretical rather than practical, however. It depends on a number of statistical assumptions and input that can vary significantly depending on the availability and quality of the underlying data. In other words, a small change in one assumption (which may have little data to support it) can have a significant impact on the conclusion.

Specifically, hedonistic regression attempts to place a value on a certain feature or attribute of a product. To use a simple example, if one is buying a children’s wagon, how important is it to a purchaser that the wagon is red?  Or put another way, what percentage of the total price can be attributed to the fact that the wagon is red instead of green? Hedonistic regression attempts to answer that question by measuring the amount by which prices will change with a change in the feature at issue. Sometimes, this is a straightforward exercise; if the wagon comes in both red and green and the red is priced higher, we can put a finger on the value of that attribute. But the question is rarely that easy.

In the VLSI case, the damages expert purportedly attempted to estimate the impact of different features of Intel’s chips (chiefly processor speed) on the price of those chips, by estimating the fluctuations in price based on microprocessor speed. He assumed that a given increase in speed would correspond to a given increase in price, holding all other attributes constant. Several problems with that theory, however, become immediately apparent. Namely, he had to make several assumptions about the price increase, assume that increases in speed are solely based on the patented contributions and that there is no other (or cheaper) way to do it, assume that the speed increases would result in no additional costs, and assume that Intel (and customers) would have been willing to pay the higher prices in a hypothetical sale.

At trial, Intel’s attack on VLSI’s damages theory is described as having been focused mainly on the lack of widespread acceptance for a regression analysis such as VLSI’s. Intel pointed out that a regression analysis had never been used in real-world patent valuations and is not an accepted methodology in that context (in contrast, regression analysis has been used to some degree in antitrust cases, where exact dollar figures are not important so much as market power or market share). Intel also tried to tie the damages to the fact that VLSI had never used the patent itself; that lack of use of the patent was indicative of its low value, which would have driven down the royalty at a hypothetical sale negotiation. Intel also relied on a more common type of damages calculation; one based on purportedly comparable licenses. Intel’s expert concluded that a reasonable royalty for use of the two patents should have been more on the order of $2.2 million (or about ¼ of a penny per chip). The jury, however, accepted VLSI’s theory.

What Impact Will This Decision Have on Patent Valuations?

The obvious short-term result of this verdict is that it will likely embolden other patentees/plaintiffs to seek similarly large damages award in patent lawsuits. Plaintiffs, especially in the popular patent forum of the Western District of Texas, may also attempt to point to the VLSI verdict to argue that a hedonistic regression theory is now a “tested” and “accepted” methodology for purposes of surviving an admissibility challenge. Indeed, VLSI itself sought an additional $3.1 billion in damages from Intel in a second jury trial that was just completed April 21, although in that case the jury ultimately sided with Intel.  Commentators, moreover, generally agree that the large $2.18 billion VLSI verdict is unlikely to survive challenge on appeal. Typically, when damages experts employ novel and untested methodologies, they also utilize more traditional and accepted methods to corroborate and support their conclusions. That does not appear to be the case here, as the novel method utilized by the VLSI expert resulted in a damages figure that is magnitudes different than the damages amount calculated by the Intel expert using a traditional method.

With respect to whether this decision will change how patent valuations are conducted in real-world, non-litigation contexts, that is very unlikely. Most real-world patent valuations that are used as part of real licensing negotiations rely on time-tested methodologies such as the relief-from-royalty method, which is a blend of market royalty rate analysis and an income-based approach. This method values a patent by calculating the royalties the patent owner would be required to pay to license the patent if the owner did not already own the patent. Unlike the damages methodology used by VLSI, the relief-from-royalty method utilizes actual observable market transactions to support a royalty rate, which can be applied to historical and/or projected income streams for the patent to determine its value. In real-life negotiations, if the parties’ respective valuations are light-years apart, the negotiations will either break down or the parties will move closer to a valuation methodology that both parties can agree to in principle. A regression analysis is very unlikely to become widely used in actual licensing or patent sale negotiations because of its inherent uncertainty, loose estimates, and assumptions. In comparison, valuation methods such as an income, market, or cost-based approaches, provide tangible, time-tested starting points for determining the value of a patent.

Image rights acquired by 123RF.com