Google LLC v. Sonos, Inc. — Federal Circuit Reverses PTAB Invalidation of Google’s Hotword Detection Patents

Case
Google LLC v. Sonos, Inc.
Court
U.S. Court of Appeals for the Federal Circuit
Date Decided
June 9, 2026
Docket No.
2024-2119, 2024-2120
Judge(s)
Moore, Chief Judge; Lourie and Reyna, Circuit Judges (opinion by Moore, C.J.)
Topics
Inter partes review, patent validity, anticipation, claim construction, hotword detection

Full Opinion

Your browser cannot display this PDF inline.

Download the full opinion (PDF)

Background

Google and Sonos have been embroiled in wide-ranging patent litigation for several years. This appeal arises from a different front in that battle: here, Google is the patent owner, and Sonos attacked Google’s patents through inter partes review (IPR) — a procedure before the Patent Trial and Appeal Board (PTAB) that allows third parties to challenge the validity of issued patents on grounds of prior art.

The two patents at issue, U.S. Patent Nos. 10,134,398 and 10,593,330, cover a system for coordinating hotword detection — think “OK Google” or “Hey Siri” — across multiple voice-enabled devices in the same household. The core inventive concept is that when several devices simultaneously hear a hotword, they should exchange “confidence scores” with each other while still remaining in low-power sleep mode, so the device best positioned to handle the command wakes up while the others stay dormant. Without this coordination, every device in a room might spring to life at once.

Sonos filed IPR petitions arguing that the claims were anticipated by — meaning fully disclosed in — a single earlier patent known as Rosenberger (U.S. Patent No. 8,340,795), and alternatively that they were obvious over Rosenberger combined with a second reference called Basye. PTAB agreed, finding the challenged claims unpatentable. Google appealed to the Federal Circuit.

The Court’s Holding

A Federal Circuit panel unanimously reversed and remanded. Chief Judge Moore, writing for the court, concluded that the Board’s key anticipation finding lacked substantial evidence — the bedrock standard of review for factual determinations from PTAB.

The Board had relied on column 8 of Rosenberger, arguing it disclosed a device that exchanged confidence signals while remaining in its low-power “listening” mode. The Federal Circuit carefully read that passage and found it described the opposite: Rosenberger’s device exits its low-power mode before performing inter-device coordination. Two specific arguments Sonos advanced — that a “status light change” in column 8 indicated a lingering low-power state, and that the phrase “waking up” was causally linked to determining the device was in the best position to respond — were each rejected as misreadings. Rosenberger lists those two scenarios as independent alternatives separated by the word “or,” not as a cause-and-effect sequence.

Sonos also raised an alternative anticipation theory: that during inter-device coordination, the Rosenberger devices consumed less power than their normal “operating mode,” and that reduced consumption could itself satisfy the agreed-upon claim construction for “low power mode.” The Federal Circuit declined to resolve this theory in the first instance, instead remanding to PTAB to make the necessary factual findings. The Board had never addressed the alternative theory in its written decisions, and the Federal Circuit emphasized that its role is to review findings the Board actually made, not to make new ones itself.

Key Takeaways

  • Substantial evidence still matters: The Federal Circuit closely scrutinized the text of the prior art reference and refused to credit the Board’s reading when the patent itself said something different. Patent challengers must ensure their expert testimony maps precisely to the cited disclosure.
  • Alternative theories must be argued and decided at the Board level: Sonos had a potentially viable backup argument but it had never been adjudicated below. Because the Federal Circuit reviews — not makes — factual findings, that theory now goes back for a second look at PTAB.
  • The Google–Sonos patent war continues: This decision protects Google’s hotword-detection patents from this particular IPR challenge, but the conflict spans multiple dockets. Sonos still has its own patent assertions against Google pending in parallel proceedings.
  • Nonprecedential but instructive: The opinion provides a careful demonstration of how the Federal Circuit applies the “substantial evidence” standard when reviewing PTAB findings on anticipation.

Why It Matters

As smart home devices proliferate, the question of which company owns the underlying coordination technology becomes commercially significant. Google’s hotword patents, if upheld on remand, cover a fundamental aspect of how voice assistants work in multi-device environments — a market Google, Amazon, Apple, and others compete in aggressively. The ruling also underscores the high evidentiary bar challengers face when asserting anticipation at PTAB: a prior-art patent must disclose each claim limitation squarely, and vague inferences are not enough.

For patent practitioners, the decision is a reminder that PTAB proceedings are not a guaranteed shortcut to patent invalidation. Expert testimony that strays from what the reference actually says — or that relies on a causal relationship the reference describes as merely alternative — may not survive Federal Circuit review.

Leave a Comment

Scroll to Top