Introduction

The past year brought several changes to patent law. Those changes largely came from two institutions, both in the Washington, DC area: the US Patent and Trademark Office, the institution that grants patents and conducts post-grant administrative proceedings re-evaluating the validity of issued patents; and the US Court of Appeals for the Federal Circuit, the appellate court that hears all patent appeals in the United States. Recent trends at the PTO and Federal Circuit promise continuing developments in how post-grant proceedings are conducted, appellate scrutiny of jury verdicts, and the eligibility of artificial intelligence-related inventions.

Trends at the US Patent and Trademark Office

The past year brought a new presidential administration and, with it, significant changes at the US Patent and Trademark Office. Upon taking office on 20 January 2025, President Trump named Coke Morgan Stewart the Acting Director of the PTO. John Squires replaced Stewart on 22 September 2025, following his Senate confirmation as the permanent Director. Both Acting Director Stewart and Director Squires made significant changes to inter partes reviews (IPRs) – administrative proceedings used to challenge the validity of previously issued patents.

By statute, the PTO Director has broad authority over whether to institute an IPR. An IPR cannot be instituted unless the Director (or his delegee) finds a reasonable likelihood that at least one challenged patent claim will be proved unpatentable. But even if that merits threshold is met, the Director still may decline to institute an IPR as a matter of discretion.

Before 2025, IPRs were instituted frequently and, when instituted, usually led to invalidation of the challenged patents: about 65–75% of requests for IPR were granted, and in granted cases about 70–80% of claims were invalidated. But that changed dramatically under the new administration. For a stretch after Director Squires’s confirmation, the PTO instituted no IPRs at all – to the point that, by November 2025, “the USPTO’s new 0% institution rate” was making headlines (Dennis Crouch, “An Era of No: The USPTO’s New 0% Institution Rate”, PatentlyO (12 November 2025)). IPR institutions resumed by December, but in far smaller numbers than before. That dramatic decline in IPR institution rates was precipitated by a series of Director-driven changes within the PTO.

First, there was a shift in the PTO’s internal standards for assessing whether an IPR should be instituted. The prior administration had followed guidance issued by then-Director Kathi Vidal, which had sought to increase the availability of IPR proceedings. That guidance had, for example, narrowed the circumstances in which IPRs would be denied in light of parallel district court proceedings involving the same patent. In February 2025, however, Acting Director Stewart rescinded Director Vidal’s guidance. And in March, Acting Director Stewart issued her own guidance, which articulated additional factors that might support discretionary denial of an IPR petition. Those factors included the patent owner’s “settled expectations” of a patent’s validity, as measured by “the length of time the claims have been in force”. The impact of those changes was felt immediately in a series of discretionary denials where the challenged patents had been issued six or more years earlier. See, eg, iRhythm Techs., Inc. v Welch Allyn, Inc., IPR2025-00363, 2025 WL 1605274, at *1 (P.T.A.B. 6 June 2025); and Dabico Airport Sols. Inc. v AXA Power APS, No. IPR2025-00408, 2025 WL 1710857, at *1 (P.T.A.B. 18 June 2025).

Second, there was a shift in who decides whether institution is appropriate. Although the IPR statute gives the Director authority to decide whether to institute an IPR, previous Directors had largely delegated that task to the Patent Trial and Appeal Board, the entity responsible for trying and deciding IPRs after institution. Acting Director Stewart, however, announced that she would participate directly in the institution process, by personally deciding whether to deny institution for discretionary reasons. Once Director Squires took office, the role shift became even more stark. In October 2025, Director Squires took full control of reviewing IPR petitions for both merits and discretionary reasons to deny institution. IPR institution rates plummeted: from October to 1 December 2025, Director Squires did not institute a single IPR. While Director Squires has since instituted some IPRs, he has done so in just a handful of cases. Director Squires has also bucked past practice when announcing the grant or denial of institution. While the PTAB had often issued lengthy opinions explaining decisions whether to institute, Director Squires’s institution decisions now state only the outcome. Although Director Squires’s tenure is still new, if the trend continues, one can expect IPRs to remain scarce in 2026.

The sweeping changes to IPR institution practice have drawn numerous legal challenges – though, to date, those challenges have not been successful. Perhaps most notably, the rescission of former Director Vidal’s guidance and application of new discretionary denial factors has been the subject of multiple mandamus petitions to the US Court of Appeals for the Federal Circuit. The petitioners in those cases have argued, among other things, that:

• rescinding Vidal’s guidance without notice and comment violated the Administrative Procedure Act;
• applying the new guidance to already pending cases was arbitrary and capricious; and
• the PTO’s new policies violated IPR petitioners’ due process rights.

The Federal Circuit has so far uniformly denied relief: In re Motorola Sols., Inc., 159 F.4th 30 (Fed. Cir. 2025); In re SAP Am., Inc., Nos. 2025-132, 2025-133, 2025 WL 3096788 (Fed. Cir. 6 Nov. 2025); In re Google LLC, No. 2025-144, 2025 WL 3096849 (Fed. Cir. 6 Nov. 2025); In re HighLevel, Inc., No. 2025-148, 2025 WL 3527144 (Fed. Cir. 9 Dec. 2025); In re SanDisk Techs., Inc., No. 2025-152, Dkt. 46 (Fed. Cir. 9 Dec. 2025); and In re Cambridge Indus. USA Inc., No. 2026-101, 2025 WL 3526129 (Fed. Cir. 9 Dec. 2025). The Federal Circuit has held that challenges to the lack of notice-and-comment procedures must proceed in district court, not through mandamus to the Federal Circuit. It has also held that judicial review of the Director’s exercise of discretion in specific cases was barred by a provision of the IPR statute that makes institution decisions final and non-appealable. And while that provision does not bar constitutional challenges, the Federal Circuit has rejected such challenges, holding that the denial of an IPR petition does not implicate the petitioner’s due process rights.

Other legal challenges to the reshaping of IPR institution practice remain pending before the Federal Circuit. And on 5 January 2026, the Federal Circuit heard oral argument in a case, on appeal from a district court suit under the Administrative Procedure Act, raising the question whether institution criteria must be promulgated through notice-and-comment rule-making: Apple Inc. v. Squires, No. 24-1864 (Fed. Cir.). While that case involves even earlier PTO guidance, decisions in that and other cases are expected to shed further light on whether these recent changes to IPR institution practice will be sustained.

Trends at the Federal Circuit

Beyond those challenges to the PTO’s new IPR institution practices, the Federal Circuit issued a number of notable decisions arising from patent infringement litigation.

The Federal Circuit made headlines this year with its first en banc decision – decided by the full court, not just a three-judge panel–in a utility patent case since 2018. That case, EcoFactor, Inc. v Google LLC, 137 F.4th 1333 (Fed. Cir. 2025) (en banc), dealt with the admissibility of expert testimony regarding patent damages – a difficult area of law where the Federal Circuit has sometimes struggled to provide clear guidance to litigants.

The dispute in EcoFactor arose from an infringement suit over a smart thermostat patent. At trial, the plaintiff’s damages expert, David Kennedy, testified in favour of a particular royalty rate, and opined that the rate was justified by prior licence agreements between the plaintiff and third parties. Each of those prior licences, however, was for a lump-sum amount, whereas the royalty Kennedy proposed was a per-unit royalty. The jury found infringement and awarded USD20 million in damages. The defendant appealed, challenging the admissibility of Kennedy’s testimony under Federal Rule of Evidence 702.

A panel of the Federal Circuit initially upheld the damages award in a split decision, with the panel majority concluding that Kennedy’s analysis, even if imperfect, was reliable enough to be admissible. But the en banc Federal Circuit agreed to rehear the case, and on rehearing held that Kennedy’s testimony was inadmissible. Writing for the en banc majority, Chief Judge Moore held that neither Kennedy’s testimony nor other evidence offered by the plaintiff provided an adequate basis for translating the prior licences’ lump-sum royalties into a per-unit royalty like the one Kennedy had proposed. Accordingly, the majority held, Kennedy’s proposed royalty lacked the “sufficient facts or data” that Federal Rule of Evidence 702 requires for expert testimony to be admissible.

The en banc decision in EcoFactor represents the culmination of a trend of increased appellate scrutiny of patent damages awards. The Federal Circuit has increasingly sought to impose limits on experts’ ability to propose, and judges’ and juries’ leeway to accept, aggressive patent damages models. Among other things, the Court has developed an extensive body of law regulating “apportionment” of patent damages and requiring threshold comparability showings before experts may rely on comparable licences. Although the full implications of EcoFactor remain to be seen, the decision suggests the Federal Circuit is likely to continue its close scrutiny of patent damages awards.

Indeed, the Federal Circuit has increasingly scrutinised all facets of jury verdicts in patent cases, not just damages. Ordinarily, juries get broad deference: a jury verdict can be overturned only if no reasonable jury could have decided the case the way the jury did. But, in 2025, the Federal Circuit often found that demanding requirement met, or found other legal grounds for undoing jury verdicts.

The court tossed several jury verdicts of infringement, in cases with massive and small damages verdicts alike. For example, in Laboratory Corp. of America Holdings v Qiagen Sciences, LLC, 148 F.4th 1350 (Fed. Cir. 2025), a jury had found literal infringement and infringement under the doctrine of equivalents (DOE) and awarded USD4.7 million in damages. The Federal Circuit reversed. It ruled that the district court should have construed a key claim term and that the patentee’s DOE evidence was insufficient. In Finesse Wireless LLC v AT&T Mobility LLC, 156 F.4th 1221 (Fed. Cir. 2025), the Court rejected a verdict of more than USD160 million, determining that “no reasonable jury” could have found infringement of either asserted patent. Other cases have similarly rejected jury infringement verdicts. See, eg, Wonderland Switzerland AG v Evenflo Co., Inc., 162 F.4th 1346 (Fed. Cir. 2025) (rejecting DOE theory as a matter of law and vacating infringement verdict); and Colibri Heart Valve LLC v Medtronic CoreValve, LLC, 143 F.4th 1367 (Fed. Cir. 1 July 2025) (ruling that prosecution history estoppel barred DOE and vacating USD106 million damages award).

The Court has also closely scrutinised jury findings of no invalidity, even though review of such findings is “doubly” deferential given the presumption of validity that attaches to issued patents. When a jury finds that an accused infringer falls short of proving invalidity, that finding can be reversed only if the only reasonable conclusion a jury could reach is that the accused infringer proved invalidity by clear and convincing evidence.

Again, however, the Federal Circuit found that doubly high bar met in 2025. In a pair of unrelated cases issued just weeks apart, the Court threw out two jury findings that the patent claims in those cases were supported by adequate written descriptions: Duke Univ. v Sandoz Inc., 160 F.4th 1305 (Fed. Cir. 2025); and Seagen Inc. v Daiichi Sankyo Co., Ltd., 160 F.4th 1322 (Fed. Cir. 2025). In both, the Court dissected the record and found that no reasonable jury could have arrived at the jurors’ conclusions.

The trend is expected to continue into 2026. Appellate issue preservation will thus remain essential, for patent owners and accused infringers alike. Defendants will want to raise in district court any grounds that might be vehicles for undoing adverse jury verdicts on appeal. And to help maintain trial victories on appeal, plaintiffs may consider providing juries multiple means for finding in their favour, offering the Federal Circuit options to affirm on different factual grounds. Given the Federal Circuit’s apparent scepticism of jury verdicts – especially large ones – dotting every “i” and crossing every “t” could prove essential to preserving a judgment on appeal.

Developments in Artificial Intelligence

Artificial intelligence had a widespread impact on nearly every aspect of society in 2025. Patent law is no exception. One area seeing significant development is the eligibility of AI-related inventions for patent protection. Under a long-standing (if sometimes confusing) line of case law, “abstract ideas” are not eligible for patenting, but inventions directed to abstract ideas may be patentable if they add a sufficient inventive step beyond the abstract idea itself. Both the Federal Circuit and the PTO have recently considered how that doctrine applies to AI.

In Recentive Analytics, Inc. v Fox Corp., the Federal Circuit concluded that patents claiming the use of generic machine learning techniques for the specific task of event scheduling and the creation of network maps were directed to abstract ideas and ineligible for patent protection: 134 F.4th 1205 (Fed. Cir. 2025). The Court noted that improvements to machine learning techniques could be eligible for patenting, but found the claims in Recentive did not disclose patentable improvements beyond the abstract ideas themselves.

The PTO has also spoken on the eligibility of AI inventions. In Ex parte Desjardins, a PTAB panel led by Director Squires overturned an examiner’s eligibility rejection for a claim related to training machine learning models: No. 2024-000567, 2025 WL 3095778 (P.T.A.B. 26 Sept. 2025). Director Squires concluded that the claims recited an improvement to how a machine learning model operates, which he found sufficient to confer eligibility. Director Squires also said that “[c]ategorically excluding AI innovations from patent protection in the United States jeopardizes America’s leadership in this critical emerging technology” (ibid at *5). To him, other patentability requirements – novelty, non-obviousness, enablement and written description – are the appropriate tools to limit a patent to its proper scope and should be the focus of examination. Following Desjardins, Director Squires updated the Manual of Patent Examining Procedure – the PTO’s guidance document for patent examiners – to reflect Desjardins’s reasoning.

On 28 November 2025, the PTO issued revised guidance on how it will evaluate AI assistance in developing an invention. While AI cannot be named as the inventor on a patent, the new guidance says that AI systems may provide tools and services to human inventors who conceive of an invention. Under the revised guidance, when a natural person conceives of an invention with the assistance of AI, the PTO applies the traditional “conception” test for inventorship and may award a patent to a human inventor who satisfies that test.

The PTO has also considered how AI may be used to assist examination of patent applications. In October, the PTO announced the launch of an Artificial Intelligence Automated Search Pilot Program (dubbed ASAP!). Under that opt-in pilot programme, the PTO will send applicants the results of an AI-assisted, automated prior-art search before examination begins. While applicants are not required to take action in response, this early notice may allow them to identify and address potential patentability issues before the PTO formally considers their applications.

Few technologies have evolved as quickly, dramatically and often unpredictably as artificial intelligence. Perhaps the one thing we can be sure of is that AI will continue to develop – and that courts and the PTO will continue striving to keep up.