For decades, most enzymes used in medicine, lab kits and factories have been borrowed from nature, then tweaked until they did the job. The tweaking can eat up years.
A small company in Lithuania would rather skip the borrowing altogether.
Biomatter, based in Vilnius, announced an $18 million Series A on November 12, 2026. Astanor led the round, and three earlier backers, UVC Partners, Inventure and Metaplanet, came back for more. Its client list already includes AstraZeneca.
A Series A is usually a startup’s first large round from professional investors, raised once the business can show it is growing. Biomatter’s was oversubscribed, which means investors offered more money than the company planned to take.
Laurynas Karpus starts with the reaction
Enzymes are proteins that speed up chemical reactions. Drugmakers use them in production, diagnostic firms build tests around them, and farms and chemical plants depend on them too.
Laurynas Karpus co-founded Biomatter in 2018, when it still went by Biomatter Designs, and he runs it as CEO. In 2021 the founders published a paper in Nature Machine Intelligence that was among the first to show AI models could invent working enzymes unlike anything found in nature.
The seed round followed in August 2024: €6.5 million, co-led by UVC Partners and Inventure.
Biomatter’s method flips the usual order. It looks first at the reaction a customer needs, then designs a protein for that exact job.
Computer models handle the design. Biomatter’s own lab tests what they produce, and every measurement goes into a private dataset the models learn from on the next project.
That loop has cut its programs down to weeks.
From one-off projects to multiyear programs
Biomatter says it has delivered more than 50 custom enzymes to over 25 large customers in Europe, the US and Japan. Besides AstraZeneca, the named clients include the German chemical group BASF, Sanofi, Kirin Holdings, Neogen and Tereos. Several started with a single project and grew into multiyear deals covering a handful of enzymes.
You can already buy one of the results. Biomatter worked with Thermo Fisher Scientific to make its Invitrogen SuperScript V reverse transcriptase hold up better against oxidation. Reverse transcriptase is the enzyme scientists use to copy RNA into DNA. The improved version ships without cold packs and keeps at room temperature for up to six months.
Another project went after mRNA vaccine production. Biomatter designed a methylase, an enzyme that attaches a tiny chemical tag called a methyl group, that works without SAM. Natural methylases can’t run without that additive.
The decision to build in Vilnius
The biggest share of the money goes into a building. Biomatter is putting up a 2,000 m² enzyme design and testing facility in Vilnius, roughly 21,500 square feet, scheduled to open in 2027. The logic comes straight from how the company works. Its models only get better when real experiments feed them new results, so more lab space means more enzymes measured, more data flowing back into the system and more of the design-and-test cycle happening in one place. Biomatter says the facility should take programs that now run for weeks and finish them in days. A company that already turned years into weeks now wants to turn weeks into days, and it is pouring concrete to back that up.
The rest pays for a bigger commercial team and a stronger US presence. About half of Biomatter’s customers are already American, so the expansion follows the clients.
“Our goal is to make a custom enzyme as easy to order as a catalog product, including for reactions that don’t have one today.”
Laurynas Karpus, CEO and co-founder of Biomatter
Astanor brings a food-and-farming angle. Founded in 2017, it invests mostly in agrifood technology and holds about 50 companies across Europe, North America and beyond. Agriculture is one of Biomatter’s target markets, so the pairing makes sense.
Hold on, an AI company spending on a building?
Yes. And that’s the part I like most.
Most AI pitches stop at the software. Biomatter is putting its biggest bet on lab space, because its models are only as useful as the experiments behind them.
The proof is already out in the world, too. An enzyme that lets a lab kit skip the cold chain sits on Thermo Fisher’s shelves. Names like Sanofi are on the client list, and several customers have stretched one project into years of work.
That’s harder to fake than a demo.
The 2027 opening in Vilnius
The Vilnius facility is the next milestone, with its opening set for 2027. Alongside it, Biomatter plans to grow its US commercial team and keep adding to its computational platform and enzyme dataset, the steps laid out in the company’s announcement.
