IP Intelligence

Home | IP Intelligence

After an accused infringer files an IPR, it may still try to assert additional prior art later in district court. Patentees often respond by invoking IPR estoppel, arguing that the accused infringer reasonably could have found and raised that art in the IPR. But the burden is on the patentee to prove that point, and a recent Federal Circuit decision makes clear that a showing that a reference existed in a searchable database, or appeared somewhere in a large search result, is not necessarily enough. The patentee must show a realistic, non-hindsight search path that would have led a skilled searcher to the asserted prior art before the IPR petition was filed. That requirement may make estoppel harder to prove for later-identified prior art.

core/paragraph

core/heading

In <a href="https://www.cafc.uscourts.gov/opinions-orders/24-2088.OPINION.6-18-2026_2711717.pdf" target="_blank" rel="noreferrer noopener">Ironburg Inventions Ltd. v. Valve Corp.</a>, Valve filed an IPR petition challenging Ironburg’s video game controller patent. After that IPR, Valve sought to rely in district court on invalidity grounds based on prior art that had not been included in its petition. Ironburg argued that Valve was estopped under 35 U.S.C. § 315(e)(2) because those grounds reasonably could have been raised earlier. The district court agreed.

The first appeal focused on the burden of proof. The Federal Circuit held that the district court had placed the burden on the wrong party. Valve did not have to prove that it could not reasonably have raised the grounds in the IPR. Ironburg, as the party asserting estoppel, had to prove that the grounds reasonably could have been raised. The Federal Circuit vacated the estoppel ruling and remanded.

On remand, Ironburg submitted additional search evidence, and the district court again found estoppel. The second appeal focused on whether that evidence satisfied the skilled-searcher standard. The Federal Circuit reversed, holding that Ironburg had not shown that a skilled searcher conducting a diligent search reasonably could have discovered the later-raised grounds before Valve filed its IPR petition.

The district court relied in part on evidence that one later-asserted reference appeared within patent classifications searched by Valve’s prior-art search firm. But the relevant search returned more than 26,000 references. The Federal Circuit held that the mere presence of the reference somewhere in those results did not establish reasonable discoverability.

The Federal Circuit did not require manual review in every case and did not fully define what “discovery” requires. But when a search returns an unreviewably large number of references, the patentee must show “something more.” A broad classification search may show that a reference was accessible. It does not necessarily show that the reference reasonably would have been found.

Hindsight Undermined the Later Search Evidence

Ironburg also relied on search evidence generated years after Valve’s IPR petition. The Federal Circuit rejected that evidence because it did not reliably show what a skilled searcher would have found at the relevant time.

Some of the later searching used citation information that included materials arising after Valve’s IPR petition. Those materials then influenced later search terms and classifications. The Federal Circuit concluded that this process could not show that the same references reasonably would have been found before the IPR petition was filed.

The Federal Circuit also rejected a supplemental search that appeared to have been conducted because a particular reference still had not been found. A search aimed at locating a known missing reference is different from an objective pre-petition search. To prove estoppel, the patentee must show that a skilled searcher, using information available before the IPR petition was filed, reasonably would have found the asserted art or ground without being guided by hindsight.

The opinion does not provide a checklist for proving estoppel. It does, however, identify the type of evidence that is likely to matter. A stronger record would likely include a conventional pre-petition search methodology, the search strings and sources a skilled searcher reasonably would have used, narrowing steps that reduce large result sets to a manageable universe, and controls to avoid hindsight.

The Federal Circuit also treated a reference cited on the face of the challenged patent as different from a reference that appeared somewhere in a broad classification search. That distinction may help patentees in some cases, but it also reinforces the limits of relying on theoretical searchability alone.

Judge Stark’s concurrence adds another layer. Section 315(e)(2) estops invalidity “grounds,” not merely prior-art references. In his view, a patentee must show both that the references were findable and that the invalidity ground based on those references reasonably would have been discovered. That distinction could be important for obviousness combinations, where finding individual references may not be the same as discovering the asserted ground.

The Federal Circuit addressed traditional prior-art search tools, including classification searches, keyword searches, citation searches, and manual review. The Federal Circuit did not address AI-assisted searching.

As AI tools become more common, patentees may argue that more art reasonably could have been discovered before an IPR petition was filed. Accused infringers will likely focus on what tools were reasonably available and reasonably used at the time, whether those tools would have produced reliable and reviewable results, and whether the proposed search path avoids hindsight. The relevant question should remain what a skilled searcher reasonably would have discovered before the IPR petition was filed, not what can be found later with better tools.

For accused infringers, Ironburg provides a useful response to broad estoppel arguments. Prior art is not subject to IPR estoppel merely because it existed in a searchable database or appeared somewhere in a large search set. For patentees, the decision underscores the need for evidence showing how a diligent skilled searcher would have found the art before the IPR petition was filed, without relying on hindsight. After Ironburg, searchable is not enough.

Searchable Is Not Enough: The Federal Circuit Raises the Bar for Estopping ‘New’ Prior Art