On Thursday 9 October 2026, EPO Board of Appeal 3.3.04 revoked Novo Nordisk’s European patent EP 3 448 416, one of the patents behind the semaglutide products sold as Ozempic and Wegovy. The board reached that conclusion contrary to its own preliminary opinion, and the hearing turned principally on formal grounds — the comparison between the claimed invention and the formulation actually described in the patent, and the status of a study published before the application was filed.
Reporting by JUVE Patent indicates the opposition was led by Hamm & Wittkopp for Stada and Kraus & Lederer for Teva, with Potter Clarkson acting for Novo Nordisk. Under the patents that survive, market exclusivity for semaglutide products could now lapse as early as 2031; EP 416 would have run to 2037.
Why this one stings more than a normal revocation
Patents are revoked at the EPO every week. What makes this decision unusually expensive is a portfolio decision taken years earlier: there was no divisional application pending in this family. When the parent fell, nothing was left behind it.
A pending divisional is, in practice, insurance. It keeps a second set of claims alive before the examining division while the parent is attacked in opposition, so an adverse appeal outcome costs you one patent rather than the whole position. The window to buy that insurance is narrow and it is unforgiving: under Rule 36 EPC a divisional may only be filed while the parent is still pending, and pendency ends the day before the European Patent Bulletin mentions the grant. Approve the Rule 71(3) text without having decided the divisional question and the option is simply gone — there is no restoration, no late filing, no discretionary remedy.
A preliminary opinion is not a result
The second lesson is about how much weight to put on a board’s preliminary opinion. It is a provisional, non-binding view issued to focus the oral proceedings, and boards do change their minds once the arguments are heard — as happened here. Treating a favourable preliminary opinion as a result, and standing down contingency planning on the strength of it, is a risk that compounds when there is no divisional in reserve.
What IP owners should take from this
- For any family protecting material revenue, decide the divisional question before approving the grant text — not when an opposition lands.
- Track pendency as a hard deadline in your docket, not as a note on the file.
- Model what the portfolio looks like if the strongest patent is revoked. If the answer is a cliff rather than a step, the portfolio is thinner than the patent count suggests.
- Read a favourable preliminary opinion as an improvement in odds, never as an outcome.
PerspireIP works on exactly these questions — portfolio analysis to find where a single revocation would create a gap, invalidity searching for parties on the other side of an opposition, and patent docketing that treats pendency as the irreversible deadline it is.