Intel Cuts $2.2 Billion VLSI Verdict With Appellate Reversal (1)

A $2.2 billion verdict against Intel Corp.—one of the biggest patent damages awards in US history—was wiped out Monday by the Federal Circuit.

The appeals court said one of the two patents asserted by VLSI Technology LLC against the semiconductor giant wasn’t infringed. The three-judge panel affirmed a Waco, Texas, jury’s finding that a second patent was infringed, but vacated the damages award of $1.5 billion tied to that patent—US Patent No. 7,523,373.

The jury’s infringement findings and associated damages amounts—$1.5 billion tied to the ‘373 patent and $675 million tied to US Patent No. 7,725,759—would’ve been blockbuster verdicts by themselves. Taken together the March 2021 verdict was a landmark win for VLSI.

VLSI alleged that Intel used inventions claimed in the ‘373 patent to create power saving for two of its popular microprocessor products. In an opinion for a three-judge panel, Judge Richard G. Taranto of the US Court of Appeals for the Federal Circuit wrote that VLSI’s infringement theory was backed by “ample expert testimony” and “adequate support in Intel’s internal documents,” turned over as part of the case at the US District Court for the Western District of Texas.

But, Taranto continued, the jurors were led astray by VLSI when it came to damages, and specifically the way the firm quantified the power-saving features in Intel’s microprocessors attributable to the VLSI patent.

A VLSI expert, Murali Annavaram, fed the wrong inputs into a regression analysis the company employed to reach its estimate that Intel owed it $1.6 billion for infringement of the patent, Taranto wrote. The opinion says Annavaram erred by relying on a set of data labeled “Core C7" rather than, or in addition to, “Package C7" data. “The difference between the Core C7 residency data and the Package C7 residency data is on its face significant,” it states.

The appeals court didn’t rule on several other arguments raised by Intel taking issue with the district court proceedings, including whether VLSI should’ve been allowed to tell the jury about “concededly noncomparable licenses” that Intel entered into for use of different technologies or patents.

It also said Intel had not “persuasively shown that the regression analysis used” by VLSI is inherently “an improper or unreasonable one.”

Reversals

The panel reversed the Waco jury’s separate finding that Intel infringed ‘759 patent, which relates to “managing clock speed,” under the “doctrine of equivalents.” That doctrine allows for patent owners to win verdicts when an infringing product differs from a patented invention only in an “insubstantial” way, according to Taranto’s order.

Taranto wrote that “VLSI offered no meaningful testimony” showing how Intel’s products “were substantially the same as the elements” laid out in the patent.

Taranto also wrote that Waco-based federal judge Alan D. Albright erred by preventing Intel from arguing it’s immune from VLSI’s suit because of its 2012 patent licensing agreement with a third company, Finjan Inc.

Intel’s Finjan license wouldn’t have had relevance to VLSI’s lawsuit when it was filed in 2019. VLSI parent Fortress Investment Group LLC, however, acquired Finjan in July 2020. That acquisition, Intel argued in a September 2020 motion, made Finjan and VLSI “affiliates” as that term was defined in its license agreement with Finjan, which in turn meant Intel was licensed to use VLSI’s patents, in addition to Finjan’s.

Albright ruled after the trial concluded that Intel’s license defense had been asserted too late in the case, would cause prejudice to VLSI, and would’ve been futile even if it’d been allowed.

Taranto said that Intel had been diligent in pursuing the defense after it learned of Fortress’s Finjan acquisition. He wrote that Albright was wrong to toss out the defense without a “fuller analysis of the governing law than has yet occurred.” He didn’t rule on whether Intel or VLSI would ultimately prevail with regard to the defense, remanding it back to the Western District court.

Reaction

Intel was “pleased the appeals court vacated the judgment below,” according to a statement released Monday. It noted that both patents were held invalid in a parallel administrative proceeding at the US Patent and Trademark Office’s Patent Trial and Appeal Board.

“Intel looks forward to making its case to a jury that the VLSI patent sent back to the trial court is also of little value,” it said.

Lawyers for VLSI didn’t immediately respond to a request for comment.

Judges Alan D. Lourie and Timothy B. Dyk joined the opinion.

Intel is represented by Wilmer Cutler Pickering Hale and Dorr LLP. Irell & Manella LLP and MoloLamken LLP represent VLSI.

The case is VLSI Tech. LLC v. Intel Corp., Fed. Cir., 22-1906, aff’d in part, rev’d in part, and vac’d 12/4/23.