The UPC Marks the Edge of Its Own Jurisdiction in Sandoz v. Novo Nordisk

UPC Milan (IT) Local Division

In a procedural order dated October 22, 2026, the Milan Central Division of the Unified Patent Court (UPC) delivered a short ruling in Sandoz v. Novo Nordisk (UPC_CFI_2221/2026) that deserves more attention than its procedural posture suggests. The operative holding is a single sentence: the UPC has no jurisdiction to entertain a claim seeking a declaration that a composition, independently of a specific patent right, was obvious at a given priority date. Behind that line sits a question the court will keep meeting as it matures: what, exactly, is this court for, and where does its competence stop?

A Routine Revocation with Three Unusual Riders

The underlying action is ordinary enough. Sandoz sought revocation of Unitary Patent EP 3 689 365, Novo Nordisk’s composition comprising semaglutide for the treatment of type 2 diabetes, administered subcutaneously once weekly at 1.0 mg, principally for lack of inventive step and added matter. The semaglutide portfolio is among the most heavily litigated in Europe, and this patent alone is the subject of numerous European Patent Office (EPO) oppositions, several brought through straw men. A single-action revocation at the UPC, if successful, would clear protection across every Contracting Member State at once.

What made the filing notable were three further requests. Request 5 asked the court, under what Sandoz called its “inherent jurisdiction,” to declare that the semaglutide composition was obvious at the priority date—not that this patent is invalid, but that the subject matter itself was obvious, free-standing and untethered from any particular right. Requests 6 and 7, as Sandoz clarified at the September hearing, were aimed at what it called the “divisional game”: preventing Novo from using divisional filings and strategic withdrawal to keep the subject matter in play while avoiding a ruling on the merits.

Novo met all three with a preliminary objection under Rule 19 of the Rules of Procedure, arguing that the UPC is a court of conferred, attributed, jurisdiction, not inherent jurisdiction, as it may hear only the actions enumerated in Article 32(1) of the UPC Agreement and grant only the remedies the Agreement provides. On Request 5, the court agreed. An abstract declaration of obviousness is neither the revocation of an identified patent nor a declaration of non-infringement. It is not on the Article 32 menu at all.

A Familiar Destination by an Unfamiliar Road

For a U.S. audience, the result will feel recognizable from an unexpected direction. Article III forbids advisory opinions, and the Declaratory Judgment Act bites only where there is an actual controversy of sufficient immediacy and reality, the standard the Supreme Court restated in MedImmune, Inc. v. Genentech, Inc. A request to declare a composition “obvious” in the air, divorced from an identified patent and a concrete dispute between adverse parties, would fail that test for want of a justiciable controversy. The UPC arrived at a structurally identical place by a different route: not constitutional standing, but the principle that a treaty court holds only the competence its treaty confers. Two systems, two doctrines, one instinct: courts resolve disputes about rights, not free-floating propositions about the state of the art (that comparison is analysis, not a claim that the court reasoned in those terms).

The Half That is Not Patentee-Friendly

It is tempting to file this order under “encouragement for patentees,” and the refusal of a portfolio-wide, third-party-sweeping obviousness declaration is genuinely that. But read the order whole and the gloss thins. The same logic that denies Sandoz an abstract ruling supports the court’s competence over Requests 6 and 7. A court entitled to hear a revocation action must be able to preserve the subject matter of that action; otherwise a patentee could neutralize a live challenge simply by surrendering the patent or withdrawing the application and re-emerging through a divisional. The court therefore confirmed it has competence to determine a revocation action pending before it, notwithstanding withdrawal of the patent, and to take steps to prevent withdrawal of the application at the EPO while the UPC action is on foot. That is not comfortable reading for originators who treat divisional cascades and timed withdrawals as routine portfolio tools.

Competence is Not Entitlement, and That Distinction is the Point

Here, the order is careful, and practitioners should be too. Competence is not entitlement to relief. The court decided only that it may entertain Requests 6 and 7, not that it will grant them. Admissibility, the merits, the scope of any order, and the conditions for a protective measure were all expressly reserved to the main proceedings. The court also indicated that the material before it did not amount to sufficient evidence of misconduct on Novo’s part to justify relief at this stage. In other words, the gate is open; the challenger still has to walk through it, and on the present record it had not. Reading competence as a near-win on the “divisional game” would badly overstate what was decided.

There is a quieter lesson about legal culture, too. Request 5 borrowed the common-law vocabulary of “inherent jurisdiction”, the reservoir of powers a court draws on to do justice and control its own process. The UPC, a treaty court with civil-law foundations, declined to treat that reservoir as a source of substantive competence. It will exercise ancillary powers to protect proceedings properly before it; it will not manufacture heads of jurisdiction the Agreement did not confer. Incidental power to preserve an action, yes; standalone competence to opine, no. That distinction is likely to recur well beyond pharmaceuticals.

What to Take from Milan

For challengers, the single-action, portfolio-wide shortcut is closed: obviousness must still be litigated patent by patent, declaration by declaration. That is precisely the asymmetry, cost, delay, and the whack-a-mole of divisional thickets, that Sandoz was trying to escape, and it remains. The consolation is real but partial: the court has signalled it will not let a patentee manipulate the existence of a right to defeat a concrete challenge, where abuse can actually be shown. For patentees, the comfort is narrower than “the UPC will not entertain broad attacks.” It will not entertain abstract ones. It will, on a proper record, scrutinize withdrawal and divisional tactics deployed to dodge a ruling. Permission to appeal was granted, so the Court of Appeal may yet refine where the line falls.

The headline from Milan is not that patentees won or challengers lost. It is that the UPC, still defining itself, chose a principled and recognizably mainstream boundary, i.e., it is a court of conferred jurisdiction that decides disputes about rights, not a forum for advisory pronouncements about the prior art. Everything that matters next, whether the “divisional game” amounts to abuse, and what the court should do about it, now turns on the merits, where competence was only ever the price of admission.