Semaglutide Patent Expiration: How a Molecule Patent Actually Ends

I used to think a patent worked like a carton of milk: there is a date stamped on it, and after that date the thing stops being yours alone. For semaglutide, the molecule in Ozempic, Wegovy and Rybelsus, the stamped date was March 20, 2026. Novo Nordisk filed the core compound patent family on March 20, 2006, a patent runs 20 years from filing (this is the one part of patent law that fits on a napkin), and so in countries without an extension the compound patent just ended that day. It did end, in India, China and Brazil among other places.

In the United States, semaglutide patent expiration works differently, and the stamped date is wrong. The patent expired in 2026. Well, no. The US compound patent, US 8,129,343, was originally due to run out in 2026, but after an adjustment and an extension it now runs to December 5, 2031, per the FDA Orange Book. Novo Nordisk, on the other hand, says in its 2025 Form 20-F that the US compound expiry for Ozempic, Rybelsus and Wegovy is 2032. Secondary US patents are reported to run to 2038 (one law firm summary says 2033), and one estimate puts a US generic Rybelsus at May 2, 2034. The European Union granted a Supplementary Protection Certificate to 2031, and Japan’s protection was extended to about 2031 too.

When the semaglutide patent expires, and why there are several answers

The reason there is no single date is that “the semaglutide patent” is not one thing. There is the compound patent, which covers the molecule itself. Then there are formulation patents, which cover particular ways of turning the molecule into a medicine, and device patents, which cover the pen you inject it with. Each has its own filing date and its own end.

This is like the end of a Marvel movie. The credits roll, half the theater leaves (the patent lawyers never leave), and the people who know better stay seated, because there is always another scene.

What the compound patent covers, and what it does not

The compound patent is the big one, in the sense that it covers semaglutide, the molecule. When it ends, a competitor may make the molecule. That is basically the whole of what it gives you. It does not let you copy any particular tablet or pen if those are covered by their own patents.

Here is the cleanest example. Rybelsus is a semaglutide tablet, and getting a peptide through the stomach is hard, so Novo uses an absorption enhancer called SNAC. The website Generic Guide reports that “a semaglutide oral tablet formulation patent covering the SNAC absorption enhancer in a specific range runs to 2031 in India, blocking generic oral semaglutide (Rybelsus equivalent) even after the injectable compound patent expired in March 2026.” Two Indian manufacturers then gave the Delhi High Court undertakings that their tablets fall outside that concentration range. I assume the hearing went something like: “Your Honor, our tablet contains SNAC. Just not that SNAC.”

Pens work the same way. In Canada, generic semaglutide became available from mid-May 2026 (the reason Canada got there early is its own story, told elsewhere on this site), while pen device patents there are reported to run to at least 2033.

Here is a made-up scenario, purely made up, to show where this goes. A generic company gets the molecule in 2026, but the pen is patented, so it designs a new pen. The cap is patented, so it designs a new cap. The click the pen makes when it doses is patented, so the new pen is silent. The box has a patented fold. Eventually the generic arrives as a silent, capless, unboxed object that a patient must assemble from a kit. No real drug is like this, I think.

How a 20-year patent becomes a 26-year one

In the US, two separate mechanisms push the compound date out. A Patent Term Adjustment compensates for delays at the patent office, and one source reports Novo received 1,114 days of it for its semaglutide drug substance patents. A Patent Term Extension compensates for time lost waiting for drug approval, and DrugPatentWatch reports that US 8,129,343 “received a 1,040-day patent term extension from the USPTO.” Together, they are why 2026 became 2031.

(I tried to add those days to March 20, 2026 and land on December 5, 2031, and could not make it work, which says something about either the sources or my arithmetic, and the smart money is on my arithmetic.) How exactly the two figures combine, and why Novo says 2032 when the Orange Book says 2031, is not settled by the sources here.

The extension has limits. Under 35 U.S.C. 156, an extension cannot leave more than 14 years of exclusivity after approval, and only one patent per active ingredient product may be extended. So Wegovy and Rybelsus could not get a second extension on the same compound patent. It is like a library card that lets you renew one book, once, and never past a fixed date. You pick carefully. For comparison, Eli Lilly’s tirzepatide compound patent got a 129-day extension.

Other countries made different choices. Brazil’s regulator, ANVISA, refused Novo’s extension request, and the first generic was approved in May 2026. China has no extension that applies, because Ozempic was approved there in April 2021, and the patent law amendment that created extensions took effect June 1, 2021, without reaching back. Missed it by weeks. China does have data protection running to about April 2027, and as of October 2026 it had approved zero generics, though at least 17 Chinese manufacturers are in late trials or applications.

There is also pediatric exclusivity, a separate add-on that sometimes tacks months onto a drug’s protection. Whether it attaches to any semaglutide date is not something these sources address, so I will not pretend it is settled.

The Orange Book and the 30-month line

The Orange Book is the FDA’s list of patents a brand company says cover its approved drug. Listing matters because of the 1984 Hatch-Waxman Act, which lets a brand company trigger an automatic 30-month stay of a generic application’s approval by listing patents there. So a listed patent is not just a patent. It is a patent with a waiting period attached, sort of like a queue where anyone already standing in it can make the person behind them wait two and a half years.

DrugPatentWatch’s article on this is titled “Own the Market: How Brand Companies Strategically List Patents in the FDA Orange Book,” and I feel like the word “strategically” is carrying a lot of weight in that headline (I’m sure it just means thoughtfully).

Three ways to read all these dates

There are, roughly, three ways to read this.

The first reading is that the compound patent is the real date and everything else is noise. To be fair, there is a lot to this. In India, Brazil and Canada, once the molecule was free, generic injectables showed up within months. The formulation and device patents slowed specific products but did not stop the category.

The second reading is that the real date is whatever the extension rules say, which makes this a story about legislatures. Brazil refused an extension and got generics in 2026. China’s extension rule arrived a few weeks too late. The US and EU granted extensions and are at 2031 or so. The molecule is identical everywhere; the end dates track policy. That seems like a reasonable thing for policy to decide, I suppose.

The third reading is the least comfortable one. It says the molecule’s expiry is not really the event at all. In the US, after the compound patent ends in 2031 (or 2032), there are secondary patents to 2033 or 2038, a tablet estimate of 2034, and an Orange Book that can turn each listed patent into a 30-month wait. On this reading a drug does not go off patent. It goes off patent one piece at a time, and the order of the pieces is set by lawyers, and the people waiting for the price to fall cannot tell from the outside which piece is the last one.

The carton does have a date on it. So do the cap, the spout and the carton itself, and in America the cap might be good until 2038.