In Ironburg Inventions Ltd. v. Valve Corp., 2026 WL 1755196 (Fed. Cir. June 18, 2026) (“Ironburg II”), the Federal Circuit reversed a judgment that had estopped an accused infringer from pressing two invalidity grounds under 35 U.S.C. § 315(e)(2). Writing for the panel, Judge Hughes held that a classification search returning an unreviewable mass of references does not, without more, prove that a “skilled searcher” would have found the asserted prior art. The Federal Circuit also searches reverse engineered to locate known references are not probative of what such a searcher would have discovered. The decision tightens the proof a patentee must marshal to invoke IPR estoppel, and Judge Stark’s concurrence maps the doctrinal work that remains.
Background
Ironburg owns U.S. Patent No. 8,641,525, directed to a handheld game controller with “inherently resilient and flexible” back controls. It sued Valve in 2015, accusing the Steam Controller of infringing several claims. In April 2016, Valve petitioned for IPR (the ‘948 IPR, IPR2016-00948), which the Board partially instituted and which yielded a final written decision in September 2017 cancelling several claims.
The estoppel fight, however, grew out of grounds Valve never petitioned. In January 2018, a third party — Collective Minds Gaming (“CMG”) — filed its own IPR, raising two new obviousness grounds built on four new prior art references. Valve amended its district-court invalidity contentions to add those grounds. Ironburg moved to estop them under § 315(e)(2), and in November 2019 the district court agreed, reasoning that “a skilled searcher could have been reasonably expected to find all of these references.”[1] A February 2021 jury then found willful infringement and awarded $4,029,533.93.Ironburg Inventions Ltd. v. Valve Corp., 64 F.4th 1274, 1284 (Fed. Cir. 2023), 64 F.4th 1297 (Fed. Cir. 2023) (“Ironburg I”). On remand, the district court again applied estoppel, and Valve appealed a second time.
The Skilled Searcher Standard
On the first appeal, Ironburg Inventions Ltd. v. Valve Corp., 64 F.4th 1274 (Fed. Cir. 2023), the court fixed two foundational points. First, it held the district court had misallocated the burden of proof, faulting Valve to show it could not reasonably have raised the grounds; instead, “the burden of proof rests with Ironburg to prove that these were grounds Valve ‘reasonably could have raised.'” Id. at 1297. Most importantly, it adopted the governing test: a petitioner “reasonably could have raised” any ground that “a skilled searcher conducting a diligent search reasonably could have been expected to discover.” Id. at 1298.[2] The Federal Circuit then remanded to district court to carry out the analysis.
On remand, the district court permitted limited discovery and again estopped both grounds (IPR Estoppel Order, 2024 WL 2091177 (W.D. Wash. May 3, 2024)). This time Ironburg pointed to searches it conducted for trial through Cardinal IP with results containing the references. Ironburg also pointed out Valve’s own 2014 pre-petition searches had also identified the references. The district court found the evidence persuasive. Valve appealed a second time, and the Federal Circuit reversed and remanded finding the proffered evidence failed to establish a skilled searcher would have discovered the references. Ironburg Inventions Ltd. v. Valve Corp., No. 2024-2088, 2026 WL 1755196 (Fed. Cir. June 18, 2026) (“Irongburg II”).
Classification Results alone don’t establish discovery. The Court’s first holding sharpened the skilled searcher test. § 315(e)(2) estops what reasonably is discovered by a skilled diligent searcher, not just possibly. The district court had concluded the references were discoverable simply because the patent classifications searches contained it. But those same classification strings returned 26,333 references. Id at *3. That, the Federal Circuit held, does not discharge the proponent’s burden that “something more is required” when a search returns an “unreviewable number”. Id. Diligent searchers do narrow via keyword and citation searching. The panel declined to hold that manual review is always required, but it endorsed the “normative force” of the point that “thousands of sources, not further narrowed in any way, likely cannot be considered discovered for estoppel purposes.”
Policing Hindsight Bias
The second holding polices hindsight. Ironburg II reaffirmed that “when evidence is introduced that sheds light on what a skilled searcher may have discovered only after the time of the relevant IPR petition, such evidence is likely irrelevant to the skilled searcher inquiry.” Id. at *4. Ironburg’s searcher had used forward-and-backward citation searching that would have captured later-arising references citing the very art at issue, and a corrective declaration applying date filters came only after the fact. Those corrections, the court concluded, were “too little too late” because keywords were “tainted” from the initial searches. Id. at *7. The searcher’s supplemental search fared no better: by the expert searcher’s own admission, it was “only undertaken in the first instance because of the need to locate Raymond—clear evidence of hindsight bias.” Id. at *8. The court reversed the estoppel finding and remanded for the district court to consider the validity grounds raised by Valve based on those references.
Judge Stark’s Concurrence: A Two-Step Framework
Judge Stark, who authored Ironburg I, concurred in Ironburg II to highlight both the contribution and the gaps. Drawing on Ingenico Inc. v. IOENGINE, LLC, 136 F.4th 1354 (Fed. Cir. 2025), he stressed that “a[n] [invalidity] ground is not the prior art asserted during an IPR,” and that the statute “speaks in terms of ‘ground[s],’ not prior art references.” Id. at 1365. He proposed a two-step framework. In his view, a patentee invoking the skilled searcher test “must prove at least the following: (i) the prior art reference(s) at issue were findable by a skilled searcher conducting a diligent search and (ii) the skilled searcher also reasonably would have been expected to discover, based on the reference(s) found, the invalidity ground the challenger seeks to assert in court.” Ironburg II, 2026 WL 1755196, at *9 (Stark, J., concurring). In Ironburg II, Judge Stark resolved only the first step, leaving the contours of “discoverability” and of the “something more” requirement for the district courts and the Board to develop.
Practical Takeaway
A patentee invoking IPR estoppel must build a clean, contemporaneous search record, not a litigation-driven reconstruction. Two lessons follow. First, document the narrowing. A classification search that returns tens of thousands of hits proves little standing alone; pair it with the keyword, citation, or manual-review steps a diligent searcher would actually take to reach a reviewable set and preserve that work. Second, guard against hindsight from the outset. Restrict every query to the petition date, avoid forward citation searches that sweep in later art, and be ready to explain why a search was run before the target reference was known. For accused infringers resisting estoppel, the mirror image holds: probe how the proponent’s searcher arrived at the references, and press for the admission that a search was undertaken only to locate art the proponent already had in view.
About Brooks Kushman P.C.
Founded in 1983, Brooks Kushman P.C. has built a national reputation as a premier intellectual property and technology law firm. We accomplish this with the understanding that the most effective IP solutions come from putting great minds together – our clients and our own. With offices across the country, we forge strong relationships with corporations, small to medium-sized businesses, and leading universities across the country.
Brooks Kushman counts a number of Fortune 100 Corporations across a variety of industries among its clients. Our attorneys have a deep understanding and broad range of experience in a variety of industries and technologies, including automotive, AI & data, automation, consumer electronics, manufacturing, medical device, computer technology, aerospace, chemicals, biotechnology, retail, food & beverage, green technology, fintech, and more. We are also recognized by leading legal publications and rankings, including, Best Lawyers, Law360, Intellectual Asset Management, Managing Intellectual Property, and World Trademark Review. For more information, please visit www.BrooksKushman.com.
[1] One of those references, Willner, appeared on the face of the asserted patent and was affirmed by the Federal Circuit as estopped because they were “discoverable”. This reference was used in the concurrence of Ironburg II as a guidepost to establishing what a skill searcher is expected to have ‘discovered”.
[2] Critically, the court rejected Ironburg’s reliance on CMG’s petition as proof, because the record said nothing about the diligence of CMG’s pre-petition search — if CMG had used “scorched earth” tactics, its success would be “irrelevant to a determination of what would have been discovered by an ordinarily skilled searcher acting with merely reasonable diligence.” Id. at 1299. With no other evidence, the court vacated and remanded.
