Boltz raises $110M to expand paid AI molecular design
Boltz has raised a Bessemer-led $110 million Series A to expand molecular-design tools sold by usage. Its open-source research brought broad adoption; the commercial test is whether paid computation delivers repeatable gains in laboratory discovery.
- Boltz says it raised $110 million in a Bessemer-led Series A to fund compute, data and hiring.
- The company reports partnerships with six of the 20 largest pharmaceutical companies and a 50-fold increase in paid molecular-token usage since June.
- Open-source alternatives remain available; the paid products must justify their cost through better experimental results and easier deployment.
Boltz, a London-based artificial intelligence company spun out of the Massachusetts Institute of Technology, has raised $110 million in Series A funding led by venture investor Bessemer Venture Partners. It builds tools that predict molecular interactions and help scientists select or design potential medicines for laboratory testing, selling computational usage while customers retain their discoveries.
The company said in its funding announcement that the money will support compute, data and team expansion. It reports partnerships with six of the world's 20 largest pharmaceutical companies, naming Pfizer, GSK, Takeda and Bristol Myers Squibb. The announcement does not disclose revenue, contract values or a valuation, or define the ranking behind that top-20 comparison.
The financing funds the expansion from widely adopted open research into proprietary commercial products. That research gives Boltz an established audience, but its own open-source models also give scientists an alternative to paying for the newest tools.

The Boltz team in an office photo published with its Series A announcement. Credit: Boltz. Source: Boltz.
From open research to proprietary design tools
Boltz began as a company in 2025, following work at MIT's Jameel Clinic for Machine Learning in Health. Its co-founders are chief executive Gabriele Corso, whose work includes the DiffDock molecular-docking model; chief technology officer Jeremy Wohlwend, who led the Boltz-1 structure-prediction model; and head of research Saro Passaro, who built Boltz-2 for joint structure and binding-affinity prediction, according to Bessemer's investment explanation.
An MIT account published in December 2024 identified Corso and Wohlwend, then graduate students, as lead Boltz-1 developers, alongside research affiliate Passaro and faculty collaborators. MIT said their experiments matched the accuracy of Google DeepMind's AlphaFold3, a model for predicting molecular structures, on a diverse set of complex biomolecular structure predictions. That comparison concerns structure prediction, rather than drug-development outcomes.

