Ironburg Inventions Ltd. v. Valve Corporation - IPR Estoppel’s Discoverability Problem: When “Found In The Search” Isn’t Enough

Ironburg Inventions Ltd. v. Valve Corporation, Fed. Cir. No. 24-2088 (June 18, 2026).

The Federal Circuit reversed inter partes review (IPR) estoppel on both invalidity grounds Valve sought to assert at trial, finding the evidence of prior art discoverability was either underdeveloped or tainted by hindsight.

Background

Ironburg sued Valve in 2015 over a patent covering a video game controller with back-mounted controls (the Steam Controller). Valve filed a 2016 IPR petition, IPR2016-00948 (the '948 IPR), which resulted in a final written decision. Valve later tried to assert two invalidity grounds at trial that it had not raised in the '948 IPR: one based on Kotkin, and one based on the Willner-Koji-Raymond combination. Ironburg moved to estop both under § 315(e)(2).

The Legal Standard

Section 315(e)(2) bars a petitioner from asserting in district court any invalidity ground it "raised or reasonably could have raised" in the IPR. The operative test: whether a skilled searcher conducting a diligent search reasonably could have been expected to discover the ground. The burden rests with the party asserting estoppel (Ironburg), per the prior Ironburg I ruling.

Ground 1 — Kotkin: Reversed

The district court estopped the Kotkin ground solely because the CPC/IPC patent classifications containing Kotkin were included in Valve's pre-petition searches. The Federal Circuit held that was insufficient. The same classification search returned 26,333 references,far too many to constitute a meaningful discovery. The court held that when a classification search alone returns an unreviewably large number of results, something more is required (e.g., keyword narrowing, citation searching, or other filtering) before a reference can be deemed discoverable. That additional evidence was never developed.

Ground 2 — Willner-Koji-Raymond: Reversed

Ironburg relied on post-remand searches by a search company to establish that Raymond (the missing piece of the combination) was discoverable. The Federal Circuit found the evidence pervasively tainted by hindsight bias:

Searcher 1's early search string #3 used forward-and-backward citation searching without date restrictions, pulling in patents that cited Raymond only because another petitioner had previously raised it in its IPR. This court found this infected Searcher 1's entire subsequent methodology.

A later corrective declaration by Searcher 2 applied date filters only to one search string (#34), missing the broader contamination from string #3. The Fed. Cir. said, "too little, too late."

The Fed. Cir. found that Searcher 1's supplemental search was conducted expressly because Raymond hadn't yet been found. This is the definition of hindsight-driven searching.

Because Raymond was not shown to be discoverable, the entire Willner-Koji-Raymond ground escapes estoppel because § 315(e)(2) operates at the ground level, not the individual reference level, per Ingenico Inc. v. IOENGINE, LLC, No. 2023-1367, at 14 (Fed. Cir. May 17, 2025).

Still Unresolved Questions

The majority flagged two open questions it declined to resolve: (1) whether the skilled searcher inquiry is a question of law or fact, and (2) what "discovery" actually requires: does a searcher need to read and understand a reference, or merely locate it?

However, Judge Stark's concurrence offered a useful two-step framework:

1. Were the prior art references findable (not just accessible) by a diligent skilled searcher?
2. Would the skilled searcher reasonably have identified the invalidity ground from those references?

Judge Stark drew a clear line between accessibility and findability:Kotkin was accessible (in a searched classification) but not findable given the volume of results. He also noted a potential shortcut: when a reference is cited on the face of the challenged patent (as Willner was, undisputed), that may satisfy step one without more.

Case Disposition

Reversed and remanded. The district court must now allow Valve to assert both the Kotkin ground and the Willner-Koji-Raymond ground at trial.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.