All insights
Annual Report9 min read

The State of the Unified Patent Court: 2025 Annual Report

UPC 2025 statistics: 741 classified decisions, ~883 new cases, preliminary injunction grants near 57%, infringement findings near 63%, and the year doctrine took hold.

Our platform classified 741 UPC decisions and orders issued in 2025 — up from 451 in 2024, and more than the Court's first two calendar years combined. That single number is the cleanest summary of where the Unified Patent Court now stands: not a pilot, not a curiosity, but a working court producing rulings at industrial pace. The headline figure for filings belongs to the Court itself, which reported roughly 883 new cases by the end of May 2025 and kept growing through the year. This is our full-year read on the UPC 2025 statistics — official caseload where the official numbers are authoritative, our classified-decision corpus where the question is who actually won.

A word of calibration first. The outcome rates below rest on small denominators — merits decisions still number in the dozens, not the hundreds — so we state every denominator plainly and let the counts speak for themselves.

UPC 2025 statistics: the caseload kept climbing

The Court opened 2025 with momentum and never lost it. By the end of May 2025 the UPC had recorded around 883 new cases since it began operating in June 2023 — 183 more than in January of the same year — including 457 infringement actions filed to date. The docket has also internationalised: most cases now run in English (55%), with German down to roughly 38%. The institution visibly strained to keep up. The Munich Local Division stood up a second panel in May 2025, and the Court of Appeal added a third panel in October to absorb its rising workload.

Our own corpus tells the same growth story from the output side. Counting every action type tied to a case number — appeals, revocation counterclaims, procedural sub-applications and all — the records we hold rose from 1,091 in 2024 to 1,238 in 2025. That raw count is deliberately broader than the Court's "new cases" tally, so we don't present it as filings; but the trajectory is unambiguous. The more telling figure is decisions: our platform classified 741 rulings and orders issued in 2025, of which 477 were procedural and 264 attached to substantive matters — the connective tissue of a court now disposing of disputes in volume, not adjudicating its first test cases.

The outcome base rates that emerged

This is the part of the UPC 2025 statistics that practitioners pay for, and the part where denominators matter most. Across all the decisions our platform has classified since the Court opened, three base rates have stabilised:

Metric (cumulative, our classified corpus)RateBasis
Preliminary injunction granted~57%49 granted / 86 decided preliminary injunction motions
Infringement found (on the merits)~63%45 infringed / 71 merits infringement decisions
Validity challenged successfullymajority36 revoked + 10 partially revoked vs. 8 maintained / 13 amended
CoA reversal on the merits (official)~33%2 of 6 merits appeals overturned

The preliminary injunction figure deserves a footnote on denominators. Our ~57% is a cumulative cut across all 86 classified preliminary injunction motions since the Court opened; the 2025-only slice (17 granted of 28 decided, ~61%) runs slightly higher, and the divisional spread is wide — Munich stricter, Düsseldorf markedly more grant-friendly. The direction is the story: the UPC is a venue where a well-prepared applicant has a real, better-than-even chance of an injunction, and the odds have edged up as the Court has settled in.

On validity, our corpus reads the contest from the defendant's side: across the merits decisions we classified, validity is attacked successfully in a large share — 36 patents fully revoked and 10 partially revoked, against just 8 maintained as granted. The Court's own published framing, two years in, puts it in round numbers from the action's side: of infringement actions, just over 50% end with the patent valid and infringed, about 25% not infringed, and about 25% invalid. The lesson for patentees is unchanged from 2024: winning on infringement is necessary but not sufficient, because the counterclaim for revocation is where a meaningful share of cases is actually lost.

The appellate layer reinforces the point. Of the six merits decisions the Court of Appeal has ruled on, two were overturned — roughly 33% — and across all case types the CoA has reversed just over a quarter of the first-instance decisions it has reviewed. First-instance outcomes are firm, but not final.

Division concentration: German gravity, and a new centre of mass

Where the work lands has barely shifted at the top, but the shape underneath it has. Of the 741 decisions we classified in 2025, the three big German local divisions — Munich, Düsseldorf and Mannheim — accounted for exactly half (372 decisions). Munich remains the centre of gravity for new filings, leading the German divisions on infringement actions, ahead of Düsseldorf, Mannheim and Hamburg.

The genuinely new feature of the 2025 map is the Court of Appeal, which by our count issued 157 decisions in 2025 — about 21% of the year's total output, on par with Munich's local division. That is what a maturing court looks like: cases have now travelled the full distance to Luxembourg in volume, and the CoA has become a primary node rather than an occasional backstop. The rest of the docket spread across at least 17 active divisions, with The Hague, Paris, Milan and Hamburg forming the next tier — a reminder that the UPC is genuinely pan-European even as German divisions dominate the headline counts.

Technology mix: telecom and pharma still set the agenda

The sectors driving the UPC docket in 2025 were the ones you would expect from a court built for high-value, cross-border disputes. Among the 2025 decisions our platform classified and tagged by technology, telecommunications led with roughly 201, pharmaceutical and medical followed at about 173, and computing and AI accounted for around 77 — the same three-way contest, at sector level, that has defined the Court since it opened. Telecom's dominance is inseparable from the SEP and FRAND wave now working through the system; pharma and medical-device disputes supplied many of the year's landmark validity fights. Measurement and testing, biotechnology and semiconductor cases filled out the next band. The composition matters strategically: the UPC's emerging doctrine is being forged disproportionately on telecom and life-sciences patents, so the standards it sets will hit those industries first and hardest.

The case law matured — the year doctrine took hold

If 2024 proved the UPC could function, 2025 was the year it started laying down doctrine that will outlast any single dispute. Four developments stand out.

Long-arm jurisdiction. In Fujifilm v. Kodak, the Düsseldorf and Mannheim divisions confirmed the UPC could rule on infringement of the UK designation of a European patent against a German-domiciled defendant — and granted an injunction reaching the UK, a non-member state. Built on the CJEU's BSH Hausgeräte v. Electrolux foundation, this "long arm" is now the most consequential jurisdictional question the Court faces, and it is heading to the CJEU on referral.

The first permanent SEP injunction. In Philips v. Belkin, the Court issued its first-ever permanent injunction in a standard-essential-patent case, finding Philips' wireless-charging patent valid and infringed. Notably, Belkin's FRAND defence failed for lack of market power — Qi is not the only charging standard — a holding that shapes how SEP holders and implementers will frame the FRAND inquiry going forward.

Inventive step and sufficiency, from the top. On 25 November 2025, the two CoA panels handed down coordinated landmark rulings in Amgen v. Sanofi and Meril v. Edwards, laying the foundations of the UPC's own doctrine on inventive step and sufficiency of disclosure. The Court endorsed a "holistic" approach to inventive step — read as less rigid than the EPO's problem-solution method — and in doing so revived Amgen's PCSK9 patent that the Munich Central Division had revoked. After more than two years of borrowing, the UPC now has an inventive-step test of its own.

FRAND-rate-setting, still open. Late in the year the CoA declined to stay Sun Patent Trust v. Vivo pending the question of whether the UPC can itself set FRAND rates, and the InterDigital v. Amazon skirmish — where the Mannheim division signalled willingness to issue an "anti-interim-licence injunction" — carried the SEP venue contest to year-end. The Court has not yet said whether it will price licences; that it is being asked tells you how central it has become to global SEP strategy.

What to watch in 2026

  • The CJEU answers on long-arm reach. The referral arising out of Fujifilm v. Kodak will define how far the UPC's injunctions travel — and how foreign courts, the UK's especially, respond. This is the dominant jurisdictional story of the year.
  • Does the UPC set FRAND rates? The Sun Patent Trust v. Vivo line should reveal whether the Court will fix global licence terms itself, with direct consequences for where SEP disputes are filed.
  • Base rates with bigger denominators. As merits decisions accumulate, the preliminary injunction and infringement rates above will firm up — and the gap between divisional preliminary injunction regimes will either narrow toward a single UPC standard or harden into forum strategy.
  • PMAC opens, and fees bite. The Patent Mediation and Arbitration Centre is slated to begin operating in 2026, and court fees rose on 1 January — two separate nudges toward earlier settlement in a docket already heavy with consensual exits (we classified 107 settlements and withdrawals among 2025 decisions alone).

2025 was the year the Unified Patent Court stopped proving it could work and started deciding how European patents will be litigated. The numbers say the venue has arrived; the case law says the rules are being written now.

Want the division cuts, outcome base rates and SEP case flow behind this report? Explore the platform, or read our companion pieces below.

Sources

Annual ReportCaseloadBase Rates

Related insights

See the data behind the analysis.

Every article here is built on UPClytics — cross-case UPC analytics on divisions, judges, firms, and outcomes.