CAFC Clarifies "Skilled Searcher" Inquiry for IPR Estoppel

Need to know

In a recent precedential decision, the Federal Circuit held that IPR estoppel does not attach merely because prior-art references were publicly available or included in voluminous search results. Rather, to invoke estoppel, a patent owner must show that a skilled searcher conducting a diligent, non-hindsight-driven search reasonably would have discovered the prior-art references at issue.  

In Ironburg Inventions, Ltd. v. Valve Corp., No. 24-2088 (Fed. Cir. Jun. 18, 2026) (“Ironburg II”), the U.S. Court of Appeals for the Federal Circuit clarified the “skilled searcher” inquiry of the America Invents Act’s estoppel provisions. Valve filed an inter partes review (IPR) petition challenging Ironburg’s video game controller patent that reached a final written decision, then later sought to assert two additional grounds of invalidity in district court largely derived from prior-art references used in an intervening third-party IPR petition against the same patent. The district court held that Valve was estopped under 35 U.S.C. § 315(e)(2) because it reasonably could have raised those grounds in its original IPR. The Federal Circuit reversed and remanded, finding insufficient evidence that a diligent search by a skilled searcher would have uncovered the references at issue.  

The court leaned heavily on the “skilled searcher” standard it endorsed in Ironburg Inventions, Ltd. v. Valve Corp., 64 F.4th 1274 (Fed. Cir. 2023) (“Ironburg I”). Under that standard, an IPR petitioner “reasonably could have raised” grounds under § 315(e)(2) that “a skilled searcher conducting a diligent search reasonably could have been expected to discover.” Consistent with Ironburg I, the court explained that the statutory language of § 315(e)(2) requires a focus on what grounds reasonably — rather than possibly — could have been raised prior to the filing of an IPR petition.  

The court first addressed the additional ground based on the prior-art reference “Kotkin.” The district court found that Kotkin was reasonably discoverable, relying on evidence showing that Valve’s search vendor had run classification searches that produced thousands of results in which Kotkin appeared. The Federal Circuit found the district court’s analysis insufficient, explaining that a reference’s mere presence within an unreviewably large set of results does not establish that a skilled searcher would have located and recognized it. While the court stopped short of holding that manual review of search results is required in all instances to establish discovery, it acknowledged the “normative force” of Valve’s argument that thousands of search results, standing alone, likely cannot establish discoverability for estoppel purposes under § 315(e)(2).  

The court next addressed whether the district court properly accounted for hindsight bias in estopping Valve’s second invalidity ground, based on a combination of the prior art “Willner-Koji-Raymond” references. Ironburg’s search vendor evidence included data containing forward-and-backward citation searches and a supplemental search aimed at locating the Raymond reference. The Federal Circuit held that both searches were tainted by hindsight bias; the initial search pulled in later-arising patents and printed publications, and the supplemental search was motivated solely by a desire to find Raymond.

The court did not provide further clarification as to whether the skilled-searcher inquiry is a question of law reviewed de novo or fact reviewed for clear error because the error was apparent under either standard.

Takeaways

• Estoppel is not automatic for a piece of publicly available prior art. The public availability of a prior-art reference does not automatically trigger § 315(e)(2) estoppel. The patent owner bears the burden of showing that the reference was reasonably discoverable by a skilled searcher.  
• Volume cuts against discoverability. A petitioner opposing the application of estoppel in district court should document the size and un-narrowed nature of any search results a patent owner relies on; thousands of unreviewed hits are unlikely to establish that a reference “reasonably could have been" found.  
• Patent owners should guard against hindsight. Search evidence must reflect a diligent search as it would have been conducted before the IPR — not a reverse-engineered effort to locate a specific reference. Searches motivated by knowledge of the target reference are vulnerable to hindsight-bias challenges.