March 2022 Update: The Federal Circuit has issued an errata to this decision. Read about it here.
Addressing inter partes review (IPR) estoppel after the Supreme Court of the United States’ 2018 decision in SAS Institute, Inc. v. Iancu, the US Court of Appeals for the Federal Circuit overruled its decision in Shaw Industries Group v. Automated Creel Systems, stating that the only plausible reading of 35 U.S.C. § 315(e)(2) estops a party from raising all claims and grounds that reasonably could have been included in the party’s petition for IPR. The Court also rejected the district court’s two-tier damages model as contrary to customary patent damages calculations. California Institute of Technology v. Broadcom Limited, Case Nos. 20-2222; 21-1527 (Fed. Cir. Feb. 4, 2022) (Lourie, Linn, Dyk, JJ.) (Dyk, J., dissenting in part).
California Institute of Technology (Caltech) filed suit against Broadcom and Apple, alleging patent infringement directed to the generation and repetition of information in a wireless data transmission system. Wireless transmission systems generally use data repetition so that the transmitted information may be decoded even when data loss occurs. The patented circuitry discloses a form of irregular data repetition in which portions of the information bits may be repeated a varying number of times.
Apple filed multiple IPR petitions challenging the validity of the claims at issue. The Patent Trial & Appeal Board (Board) concluded in all cases that Apple failed to show that the challenged claims were unpatentable as obvious. At the district court, Apple and Broadcom raised new arguments of obviousness not asserted in the IPR proceedings. The district court granted Caltech’s motion for summary judgment of no invalidity, precluding Apple and Broadcom from raising arguments at trial that they reasonably could have raised in their IPR petitions.
At trial, the district court instructed the jury that “repeat” means “generation of additional bits, where generation can include, for example, duplication or reuse of bits.” Apple and Broadcom argued that the Broadcom chips (which were integrated into Apple devices) did not infringe the asserted claims because they did not repeat information at all. With respect to one of the asserted patents, the district court did not provide a jury instruction relating to its construction that the claim language “information bits appear in a variable number of subsets” requires irregular information bit repetition. The jury found infringement of all asserted claims. Apple and Broadcom filed post-trial motions for judgment as a matter of law (JMOL) and a new trial, both of which the district court denied.
The district court adopted Caltech’s proposed two-tier damages theory, explaining that Broadcom and Apple’s products were different and therefore possessed different values simply because they were “different companies at different levels in the supply chain.” The district court ultimately entered judgment against Broadcom for $288 million and against Apple for $885 million. Broadcom and Apple appealed.
Broadcom and Apple argued that the district court’s construction of “repeat” was inconsistent with the claim language and specification. The Federal Circuit disagreed, finding that the district court “carefully and fully considered both the language of the claims and that of the written description and faithfully applied our precedent” in reaching its construction. The Court also disagreed with Broadcom and Apple’s argument that the accused circuits did not “repeat” the information bits, finding that substantial evidence in the record supported the jury’s verdict.
The Federal Circuit next turned to the district court’s denial of Broadcom and Apple’s motion for JMOL. Broadcom and Apple argued that each of the AND gates in the accused products, when considered individually, did not irregularly repeat the information bits. The Court disagreed and, relying on the detailed testimony of Caltech’s expert witness, concluded that the overall architecture of the accused products did in fact implement irregular repetition.
Turning next to the district court’s refusal to provide a jury instruction for the “variable number of subsets” claim language, the Federal Circuit found error in the district court’s refusal and remanded for a new trial on infringement on the affected patent. Citing its 2004 holding in Sulzer Textil v. Picanol, the Court explained that it was the trial court’s duty to inform the jurors of the court’s claim construction and the jury’s obligation to apply the court’s interpretation.
Apple and Broadcom argued that the district court erred in granting summary judgment of invalidity and barring them under 35 U.S.C. § 315(e)(2) from presenting new grounds for invalidity not raised in their IPR petitions. Section 315(e)(2) precludes petitioners from raising invalidity grounds in a civil action that they “raised or reasonably could have raised during that inter partes review.” In its analysis, the Federal Circuit took a new look at its 2016 decision in Shaw Industries Group v. Automated Creel Systems. In Shaw, the Court held that only the grounds actually at issue in an IPR (i.e., those raised “during” the proceeding) were subject to estoppel, and that petitioners were not barred from later raising non-instituted grounds. In the present case, the Court noted that Shaw was decided before the Supreme Court’s 2018 decision in SAS Institute, and that at that time, the Board regularly instituted on only a subset of the grounds raised in the petition. With partial institution no longer permitted, the Court deemed its interpretation under Shaw to no longer be applicable.
The Federal Circuit thus took the opportunity to explicitly overrule Shaw and clarify that estoppel applies to all claims and grounds that could have been reasonably included in the petition. Because Broadcom and Apple were aware of the newly raised prior art references at the time of filing the IPRs, the Court concluded that it was proper for the district court to bar them from raising a new invalidity challenge based on these references.
Two-Tier Damages Model
Broadcom and Apple argued that Caltech’s two-tier damages model was flawed and unsupported by the record. Caltech’s experts argued that it would have negotiated with Broadcom at the “chip-level” and with Apple at the “device-level,” justifying different royalty rates. Caltech’s model created a carve-out so that chips sold to Apple from Broadcom would not be double-counted.
The Federal Circuit soundly rejected the district court’s adoption of Caltech’s damages model. The Court explained that under its 1983 decision in Stickle v. Heublein, once full recovery is obtained against one infringer of a particular device, at most, nominal additional damages may be awarded against another infringer. The Court added that under Glenayre Elecs. v. Jackson, “[a] party is precluded from suing to collect damages from [a buyer] when actual damages covering that very use have already been collected from the maker and seller of that product.” The district court attempted to distinguish those cases by arguing that they only apply to damages calculations with defendants having overlapping royalty bases. The Court rejected this conclusion as “wholly contrived” and “contrary to the customary way patent infringement disputes are ordinarily resolved.” Thus, the Court rejected the two-tier damages model and remanded for a new trial on damages.
Judge Dyk dissented from the majority’s holding that denial of JMOL was proper. His argument rested primarily on an apparent lack of evidence that any additional information bits were generated. Judge Dyk argued that the simple branching of the input to multiple AND gates did not meet this limitation, and, citing to expert testimony, that the AND gates simply allowed information bits to “flow through.” He stated that neither the majority’s opinion nor any Caltech arguments explained how the combination of these two noninfringing components resulted in infringement. Judge Dyk also argued that there was no basis for the jury to find infringement under the doctrine of equivalents because of lack of sufficient explanation by Caltech’s expert.