OpenAI’s push into biology began before this week’s deal. In April, it introduced GPT-Rosalind, a specialised model intended for hard problems in biology, chemistry and drug discovery, and has since worked with pharmaceutical groups including Moderna and Novo Nordisk.¹ The company’s target is formidable: it says a new medicine can take 10 to 15 years to develop.¹

Anthropic has been advancing on a parallel track. It acquired AI biotech startup Coefficient Bio for roughly $400 million in April, according to The Information; in June it launched Claude Science, linking scientific databases, analytical tools and specialised agents.¹ In September, Anthropic said Claude had helped identify an enzyme system resembling CRISPR—an assertion Bloomberg reported some scientists considered overblown.¹

Then, on Tuesday, OpenAI acquired four-year-old Orchestra Bio, whose agents handle the unglamorous but consequential mechanics of drug development: monitoring studies, tracking finances and timelines, flagging delays, and resolving clinical-trial scheduling issues.² Terms were not disclosed; Orchestra had reportedly raised about $20 million by 2025.²

OpenAI will combine Orchestra’s team and technology with Rosalind and its Workbench tool, which is designed to help researchers plan experiments and analyse results.² Greg Brockman cast the deal as part of an “extremely exciting and promising emerging application” of the company’s models: accelerating drug development.Greg Brockman (gdb)³

That is the optimistic reading: AI agents could turn scientific leads into better-coordinated decisions and, ultimately, treatments that reach patients sooner. The harder-edged reading is competitive. The leading labs are no longer merely selling general-purpose models; they are buying their way into regulated, high-value industries. For smaller biotech AI startups, the Orchestra outcome offers a warning as well as an exit: the giants may become partners—or absorb them.²