跳到正文
Tech.eu· John Reynolds·· 10 小時前AI 評分57

倫敦 AI 藥物研發初創 Boltz 完成 $110M A 輪融資

London-based Boltz raises $110M Series A

AI 導讀

倫敦 AI 藥物研發初創 Boltz 完成 $110m A 輪融資,Bessemer Venture Partners 領投。這是 Boltz 年初完成 $28m 種子輪後的新一輪融資。公司表示將把資金用於算力和數據投入,以及擴大團隊。

正文

London-based AI drug discovery startup Boltz has raised a $110m Series A funding round, less than a year after emerging out of stealth.

Boltz is an AI research startup that develops AI models to speed up drug discovery. It was spun out of MIT in the US but is now located in London, employing 40 AI researchers in the capital's King's Cross district.

Bessemer Venture Partners led the round, with existing investors including Andreessen Horowitz, Amplify Partners, Zetta Venture Partners and Obvious Ventures participating.

New investors include Greycroft, F-Prime Capital, Hummingbird and Benchling, joining Nvidia, which also announced an investment earlier this year during Jensen Huang’s visit to the UK. It follows Boltz raising a $28m seed round at the start of the year.

The startup builds software which helps pharmaceutical and biotech researchers design potential drugs and decide which candidates to test in labs.

Its AI models can predict what protein and drug candidates look like when bound together, estimate how strongly they bind, and propose new molecules designed to bind to a target. It develops open-source AI models and other models through its own platform and API.

At MIT, the co-founders released open-source AI models that matched the performance of closed-source models like DeepMind’s AlphaFold3. These models are now part of the default setting for early-stage drug discovery.

Boltz has announced collaborations with Pfizer, GSK, Takeda and Bristol Myers Squibb.

Revenue is generated through partnerships and providing access to its platform and models.

It says it will use the funding to invest in compute and data, and expand its team.

Boltz said: ‘We believe we are still just at the beginning of an exponential trajectory in model capabilities. Our upcoming models are already showing substantial improvements in the quality of the molecules they generate. Our long-term goal remains a world where a scientist can go from a therapeutic hypothesis to a human-ready molecule without leaving their computer. Reaching that goal will require further advances in the models and substantial investment in compute and data.”

來源:Tech.eu · tech.eu