Ironburg Inventions Ltd. v. Valve Corporation: The Federal Circuit Conducts a “Searching” Review of the District Court’s “Skilled Searcher” Analysis

By Gabriella Hazan and Jonathan Herstoff

In its June 18, 2026 precedential opinion in Ironburg Inventions Ltd. v. Valve Corporation (“Ironburg II”) (authored by Judge Hughes, and joined by Judges Chen and Stark, with Judge Stark also authoring a separate concurring opinion), the Federal Circuit provided guidance as to what a patentee must show to meet its burden of proving that a particular piece of prior art was reasonably findable for an accused infringer’s use in an inter partes review (“IPR”) proceeding.¹  This guidance is relevant to whether a particular ground of invalidity “reasonably could have been raised” during an IPR to demonstrate the party is subject to IPR estoppel under 35 U.S.C. § 315(e)(2).  When an accused infringer is the petitioner in an IPR that leads to a final written decision, it is estopped under § 315(e)(2) from later asserting “that the claim is invalid on any ground that the petitioner raised or reasonably could have raised during that [IPR].”²  This case established that: (i) a prior-art reference is not reasonably discoverable simply because it turns up in a search that returns tens of thousands of results; and (ii) evidence tainted by hindsight bias (such as a search that uncovers references that: (1) were not available at the time of the IPR petition and question, and (2) cite to the prior-art references that the patentee alleges should have been discovered earlier) cannot satisfy the patentee’s burden of proving what a skilled searcher would have found at the time of the original IPR petition.³  The decision serves as a valuable reminder that a patentee must satisfy a demanding standard in order to invoke IPR estoppel successfully.

Ironburg Inventions Ltd. (“Ironburg”) owns U.S. Patent No. 8,641,525 (the “’525 patent”), directed to a handheld video game controller.⁴ Ironburg sued Valve for patent infringement, but before trial, Valve petitioned for an IPR to challenge the validity of the patent claims.⁵ The Patent Trial and Appeal Board (“PTAB”) instituted an IPR, which resulted in a final written decision in September 2017.⁶ A third party, Collective Minds Gaming Co. Ltd. (CMG) then filed an IPR petition challenging the ’525 patent in January 2018, raising two obviousness grounds that had not been raised in Valve’s IPR: one based on U.S. Published Patent Application No. 2010/0298053 (Kotkin), and another based on a combination of U.S. Patent No. 6,760,013 (Willner), Japanese Patent Application No. JP-A H10-020951 (Koji), and U.S. Patent No. 5,773,769 (Raymond).⁷

After CMG filed its IPR petition, Valve amended its invalidity contentions in the District Court to include the new grounds of invalidity that were raised in CMG’s IPR petition.⁸ Ironburg argued that IPR estoppel barred Valve from raising these grounds.⁹  The District Court agreed with Ironburg and barred Valve from raising the new grounds of invalidity, reasoning that “a skilled searcher” would have been able to find the prior-art references that CMG (but not Valve) relied on in their IPR petition.¹⁰ The case proceeded to a jury trial in January 2021, culminating in a verdict that Valve willfully infringed the surviving claims and an award of more than $4 million in damages.¹¹ The parties appealed.

This first post-trial appeal led to Ironburg I, which, most relevant here, focused on the burden of proof. The Federal Circuit held the District Court improperly placed the burden on Valve to demonstrate that IPR estoppel does not apply, rather than placing the burden on Ironburg to show that Valve reasonably could have raised CMG’s cited references in Valve’s IPR.¹²  On remand, based on prior-art searches conducted by Valve and Ironburg, the District Court again found that IPR estoppel barred Valve from relying on CMG’s cited references as grounds for invalidity.¹³ Valve again appealed, challenging the District Court’s conclusions with respect to the Kotkin, Koji, and Raymond references.

The Federal Circuit found that the District Court’s conclusions on remand were erroneous.  As an initial matter, the Federal Circuit held that it was error to find that Kotkin was reasonably discoverable merely because Kotkin appeared in International Patent Classification System (“IPC”) and Cooperative Patent Classification System (“CPC”) categories¹⁴ that Valve searched before filing its IPR petition.¹⁵ The Federal Circuit found that the District Court should have taken into account the fact that the classification searches yielded more than 26,000 references.¹⁶ The Federal Circuit concluded that a reasonable searcher would not be expected to review that many references, and in order to establish IPR estoppel, a patentee must submit “something more” than a simple classification search with an unmanageable number of results.¹⁷  Although the Federal Circuit did not set forth a definitive rule, the Federal Circuit suggested that a patentee potentially could meet its burden by showing that there was a reasonable next step a skilled searcher would have conducted that would have narrowed the results to a manageable level.¹⁸

The Federal Circuit also rejected the District Court’s reliance on evidence from a search firm that Ironburg retained on remand, holding that the searches the firm conducted were tainted by hindsight bias.¹⁹ First, the Federal Circuit found that some of the searches impermissibly captured references that first became available after Valve filed its IPR petition.²⁰ The Federal Circuit further found that certain later-arising results, which would not have been available to Valve, were used to guide follow-up searches.²¹  Accordingly, the Federal Circuit concluded that the District Court erred in relying on these searches when it decided the issue of IPR estoppel.²²  In addition, the Federal Circuit found that the District Court erred in relying on another search that uncovered the Raymond reference, based on deposition testimony revealing the additional search was motivated by the fact that the firm knew that there was a reference that it had not uncovered yet.²³  Therefore, the Federal Circuit reversed the District Court’s finding of IPR estoppel, and remanded with instructions that Valve be permitted to use Kotkin, Koji, Raymond, and the Willner-Koji-Raymond combination to make its patent-invalidity arguments on the merits.²⁴

This case makes clear that a patentee cannot show that a skilled searcher would have been expected to find a particular prior-art reference simply by showing that the reference appears in the results of a search. It will be interesting to see how this “something more” element is analyzed with the application of advanced search tools and the use of Artificial Intelligence.

This case also highlights the difficulties that patentees are likely to face in trying to establish IPR estoppel.  After this decision, it is difficult to see how IPR estoppel can be shown without resort to detailed testimony about what searches a “reasonable searcher” would conduct, and how a reasonable searcher would cull down the hits if there were a large number of search results.  All of these requirements will likely require at least an evidentiary hearing—if not a full trial.  Given the Federal Circuit’s emphasis on the fact that the classification searches turned up a large number of results, it would seem—somewhat counterintuitively—that accused infringers can more easily evade IPR estoppel when the patent at issue is in a crowded art.  It will be interesting to see whether the Federal Circuit will ultimately conclude that the skilled-searcher test will always require a patentee to show that a skilled searcher reasonably would have found a particular reference, or whether the proper focus is whether a skilled searcher reasonably should have discovered the particular information (perhaps from a different reference).  Although Ironburg II represents the Federal Circuit’s most detailed guidance on the skilled-searcher standard to date, there will be the need for more guidance in the future.  In addition to the issues discussed above, the decision also does not opine on issues such as: (i) what constitutes a number of search results that is “too large” for a skilled searcher to be expected to review manually and whether the answer to that question requires consideration of things such as the nature of the art and the search results, the amount of money at issue in the litigation, and the parties’ size and resources; (ii) what type of evidence is needed to show that a reasonable searcher would have conducted the search using particular search terms; and (iii) how a patentee is to prove that a search turning up a large number of results would have been narrowed by a skilled searcher in a particular way.  Additionally, because it was not at issue here, this decision does not address how the patentee is to separately show—after proving that a particular reference was reasonably discoverable—that a reasonable searcher would have discovered the particular ground of invalidity in the reference or in the combination of references. Whether such an inquiry requires testimony from a person of ordinary skill in the art is also an issue that remains unresolved.  The decision also declines to address whether the skilled-searcher inquiry is a question of law or a question of fact, which will determine the level of deference that a district court’s skilled-searcher findings will receive on appeal.