Background
The California Institute of Technology (Caltech) held patents covering irregular repeat-and-accumulate (IRA) codes — a form of error correction technology designed to improve the speed and reliability of wireless data transmission. Caltech alleged that certain Broadcom Wi-Fi chips and Apple products incorporating those chips (including iPhones, iPads, and MacBooks) infringed these patents. After trial, a jury found infringement and awarded Caltech approximately $270 million against Broadcom and $838 million against Apple, for a combined total exceeding $1.1 billion.
Both Broadcom and Apple appealed, challenging claim construction, infringement findings, the validity of the patents, and the damages award. Caltech cross-appealed on certain issues. A central dispute concerned IPR estoppel under 35 U.S.C. § 315(e)(2), which prohibits an IPR petitioner from later asserting in district court invalidity grounds that “reasonably could have [been] raised” in its IPR petition. Broadcom and Apple had filed IPR petitions challenging Caltech’s patents but had not raised all of the invalidity arguments they subsequently pressed in the district court proceeding.
The Court’s Holding
The Federal Circuit affirmed infringement of the primary Wi-Fi error-correction patents and affirmed that the defendants’ district court invalidity arguments were barred by IPR estoppel. On estoppel, the court held that § 315(e)(2) bars not only grounds actually raised in an IPR petition but also any invalidity ground the petitioner “reasonably could have raised” based on prior art that was available when the petition was filed. Defendants cannot strategically withhold prior art combinations from their IPR petition and then deploy them as invalidity defenses if litigation does not go their way.
However, the court vacated the $1.1 billion damages award and remanded for a new trial on damages. Caltech had proposed a “two-tier” royalty model that sought a lower chip-level royalty from Broadcom (the chip manufacturer) and a separately higher device-level royalty from Apple (whose products incorporated the chips). The Federal Circuit rejected this model as contrary to established patent damages principles: a patentee is entitled to one reasonable royalty for each infringing unit, and the two-tier approach effectively sought to collect double royalties for a single infringement. The court remanded for Caltech to present a proper damages theory consistent with those principles.
Key Takeaways
- IPR estoppel under § 315(e)(2) is broad: petitioners are barred from asserting in district court any invalidity ground based on prior art they knew about or reasonably could have discovered when filing their IPR petition, even if they chose not to include that ground in the petition.
- Defendants cannot strategically partition their invalidity attack by reserving some prior art combinations for district court in case their IPR fails — that approach is barred by estoppel.
- A two-tier royalty model that seeks a royalty from both a component manufacturer and a downstream device manufacturer for the same infringed patent is inconsistent with patent damages law.
- The Federal Circuit’s damages ruling illustrates the importance of presenting a single, coherent royalty theory that apportions value to the patented technology at the appropriate level in the supply chain.
Why It Matters
Caltech v. Broadcom was one of the largest patent cases in the semiconductor and wireless communication industries in years. The IPR estoppel ruling carried major implications for how defendants approach patent invalidity strategy: companies can no longer hold back prior art arguments for district court while simultaneously pursuing IPR review on other grounds. The decision forces defendants to front-load their prior art into the IPR petition or lose the right to raise it later in litigation.
The damages ruling also clarified an important boundary for patent owners seeking royalties from integrated supply chains. As products become more modular — with chip makers, device assemblers, and software platforms all potentially contributing to an infringing product — courts will insist on a single consistent damages theory that avoids double recovery. For the Wi-Fi and wireless standards communities, the case underscored the ongoing complexity of licensing standard-essential patents at different levels of the technology supply chain.