Boltz, the London-based AI lab spun out of MIT, has raised $110 million in a Series A round led by Bessemer Venture Partners, betting that drug discovery is turning into a computing problem — and that the world's biggest drugmakers will pay for answers by the token.

The round, announced on November 18, brings Boltz's total funding to $138 million, less than a year after the company emerged from stealth. Existing investors Andreessen Horowitz, Amplify Partners, Zetta Venture Partners and Obvious Ventures all returned, joined by new backers Greycroft, F-Prime Capital, Hummingbird and Benchling.

What makes the raise stand out is that Boltz is already selling. Six of the world's 20 largest pharmaceutical companies — including Pfizer, GSK, Takeda and Bristol Myers Squibb — are paying customers, and usage of its paid models has grown roughly fiftyfold since June.

From free models to paid tokens

Boltz began by giving its work away. Its first model, Boltz-1, was released open source in November 2024 and quickly became a standard tool for predicting how proteins fold, drawing comparisons with DeepMind's AlphaFold. Boltz-2 followed in June 2025, adding predictions of how tightly a molecule grips its target.

The models have now been downloaded more than five million times and are used by over 100,000 scientists, according to Bessemer. But the business sits in the two proprietary models Boltz launched on June 16: BoltzMol-1, which hunts for small-molecule drug candidates, and BoltzProt-1, which designs proteins. Both are sold through an API priced by consumption — "molecular tokens" — so a drugmaker pays for the molecules it asks the models to design rather than for software licences.

The early results give the pricing model something to stand on. Boltz says BoltzMol-1 found confirmed hits on six of ten difficult targets while testing only 28 to 51 compounds per target — a fraction of what a traditional screen would churn through. BoltzProt-1 nearly tripled its predecessor's hit rate on nanobody design.

Six of the top 20 pharmas are already customers

That traction is the core of the investment case. Drug discovery typically takes years and costs millions per programme, with most candidate molecules failing long before they reach a patient. If a model can shrink the search from hundreds of thousands of compounds to a few dozen promising ones, the economics of the whole pipeline change.

Boltz is a public benefit corporation, and its founders — chief executive Gabriele Corso, chief technology officer Jeremy Wohlwend and Saro Passaro, all MIT researchers from professor Regina Barzilay's group — say the mission is to put molecular-design tools in the hands of every scientist. The company builds no drug pipeline of its own and does not compete with its customers.

A crowded race for pharma's AI budget

The money will fund more researchers and more computing hardware, as Boltz tries to keep pace with a fast-growing field. Greycroft, a new investor in the round, estimates the AI drug-discovery market will grow roughly eightfold by 2030.

The competition is well funded. Alphabet's Isomorphic Labs raised $5 billion earlier this month in the biggest biotech venture round on record. Boltz's answer is focus: it sells the models rather than digging for drugs itself — and when paying customers increase usage fiftyfold in five months, the company argues, the science has stopped being a demo and started being infrastructure.