Anthropic buys Coefficient Bio
What happened
Anthropic is reportedly acquiring stealth biotech AI startup Coefficient Bio for about $400 million in stock, expanding the company’s push into drug discovery and life sciences. The purchase shows frontier‑model firms are broadening from model APIs into vertical application bets in regulated fields. (startupfortune.com)
Why it matters
Anthropic has absorbed the tiny, stealth team behind Coefficient Bio into its Healthcare & Life Sciences unit; the startup’s co‑founders are Nathan C. Frey and Samuel Stanton, both of whom previously led machine‑learning work inside Genentech’s Prescient Design group. ( ) Coefficient Bio was only formed late last year (reported as September 2025), operated in stealth with under ten employees, and had no public product when Anthropic moved to acquire the team. ( ) The group focused on building biology‑specific AI systems — often called biological foundation models — which are large machine‑learning models trained on biological data (for example, DNA or protein sequences, lab assay results, and structural information) so they can predict how molecules will behave or suggest experimental steps. ( ) Anthropic plans to fold Coefficient’s work into the same team that launched “Claude for Life Sciences,” a product line that connects Claude (Anthropic’s family of language models) to lab platforms and literature sources so the model can read papers, pull experimental data, and assist with research planning; that business unit is led by Eric Kauderer‑Abrams. ( ) Nathan Frey’s publication record and conference awards show the technical direction: he won an ICLR Outstanding Paper award for work on generative modeling of proteins and has published on “lab‑in‑the‑loop” antibody design systems that close the loop between model proposals and actual lab tests, while Samuel Stanton’s profile lists Bayesian and experimental‑design work applied to drug discovery at Genentech. ( )
Key numbers
- Anthropic is reportedly acquiring stealth biotech AI startup Coefficient Bio for about $400 million in stock, expanding the company’s push into drug discovery and life sciences.
- ( ) Coefficient Bio was only formed late last year (reported as September 2025), operated in stealth with under ten employees, and had no public product when Anthropic moved to acquire the team.
Sources
Quick answers
What happened in Anthropic buys Coefficient Bio?
Anthropic is reportedly acquiring stealth biotech AI startup Coefficient Bio for about $400 million in stock, expanding the company’s push into drug discovery and life sciences. The purchase shows frontier‑model firms are broadening from model APIs into vertical application bets in regulated fields. (startupfortune.com)
Why does Anthropic buys Coefficient Bio matter?
Anthropic has absorbed the tiny, stealth team behind Coefficient Bio into its Healthcare & Life Sciences unit; the startup’s co‑founders are Nathan C. Frey and Samuel Stanton, both of whom previously led machine‑learning work inside Genentech’s Prescient Design group. ( ) Coefficient Bio was only formed late last year (reported as September 2025), operated in stealth with under ten employees, and had no public product when Anthropic moved to acquire the team. ( ) The group focused on building biology‑specific AI systems — often called biological foundation models — which are large machine‑learning models trained on biological data (for example, DNA or protein sequences, lab assay results, and structural information) so they can predict how molecules will behave or suggest experimental steps. ( ) Anthropic plans to fold Coefficient’s work into the same team that launched “Claude for Life Sciences,” a product line that connects Claude (Anthropic’s family of language models) to lab platforms and literature sources so the model can read papers, pull experimental data, and assist with research planning; that business unit is led by Eric Kauderer‑Abrams. ( ) Nathan Frey’s publication record and conference awards show the technical direction: he won an ICLR Outstanding Paper award for work on generative modeling of proteins and has published on “lab‑in‑the‑loop” antibody design systems that close the loop between model proposals and actual lab tests, while Samuel Stanton’s profile lists Bayesian and experimental‑design work applied to drug discovery at Genentech. ( )
