Summary: Anthropic acquired Coefficient Bio in an all-stock deal reported on April 3, 2026 at a little over $400 million. The nine-person startup built AI drug R&D planning and regulatory-strategy tooling, and its team was set to join Anthropic’s Health Care Life Sciences division. The article says the price reflects scarce biology workflows and domain data, signaling faster vertical integration between frontier labs and biopharma.
Coefficient Bio now belongs to Anthropic. As relayed by R&D World on April 3, 2026, drawing on reporting from Newcomer and The Information, the all-stock deal values the company at a little over $400 million; OncoDaily analyzed it on April 4. Behind it is a New York startup from September 2025 — fewer than ten people building AI drug R&D tooling that treats biology as native ground.
Measured against 2026 norms the sum is small; the direction is what counts. It is Anthropic’s first biotech acquisition, and the contest between frontier labs has shifted: less about general models, more about who can pull biology workflows into a product.
Nine People, Eight Months, $400 Million
The shape of the deal fits its moment: all stock, just north of $400 million, roughly 0.1% dilution against the $380 billion valuation Anthropic carried out of its February Series G. Coefficient Bio dates to September 2025; most of its people came from Prescient Design, Genentech’s unit for computational drug discovery, and lead backer Dimension — a health-tech investor — owned about half the equity. Eight months from founding, nine employees, $400 million — a textbook capability purchase. What is bought is a team and the workflows it built, with no revenue involved.
Who Founded It, and What They Made
Samuel Stanton, a data science PhD from NYU, helped build Cortex at Prescient Design — a deep-learning architecture for drug discovery put together in modules — and co-wrote Beignet, an open-source standard describing how molecules are represented. Nathan C. Frey, a Penn materials science PhD, took an ICLR 2024 Outstanding Paper award, is a 2026 Termeer Fellow, led a cross-disciplinary Prescient team building biological foundation models, and set up its NVIDIA collaboration.
The Information reports the platform can already let AI write drug R&D plans, keep clinical regulatory strategy in order, and spot new drug opportunities. Both founders are Prescient Design alumni — the Genentech unit created to bring machine learning to antibody design, the kind of domain depth that does not come together quickly inside a frontier lab. Announcing the deal, Stanton wrote: “We’re ushering biopharma into the Intelligence Age.”
Landing Inside Anthropic’s Life Sciences Division
The team will sit inside Anthropic’s Health Care Life Sciences division. Its head, Eric Kauderer-Abrams, has already stated the goal: make Claude “hands down the best model for everything in biology.” None of this starts cold. Claude for Life Sciences shipped in October 2025 with PubMed, Benchling, and ClinicalTrials.gov wired in, and Bun and Vercept were bought earlier still. Coefficient is the first biotech purchase — and OncoDaily notes OpenAI’s automated-researcher plan adds competitive urgency. Piece the moves together and Anthropic appears to be buying the layer that makes a general model a pharma co-worker: it draws up plans, holds regulatory memory, and sorts through candidates.
The Migration Pulling Pharma People Toward AI Labs
Behind it is a reversed flow of computational biologists. Genentech eliminated at least 489 positions in 2025, part of that talent landing at AI-native startups like Xaira Therapeutics. In January 2026, NVIDIA and Eli Lilly unveiled a $1 billion partnership on an AI drug-discovery lab. Capital runs the same way: Breakout Ventures finished a $114 million fund in March betting on where AI meets biology, and Dimension, per reports, is gathering about $700 million for a third fund.
Three Signals for the AI Biopharma Sector
Three things to watch. First, at over $40 million a head, the price secures workflows and domain data rather than models — biology’s moat lives in process and data, not in the coming round of weights. Second, biotech developers should see Claude-ecosystem drug R&D tooling arrive faster: literature search, molecular representation, and regulatory strategy all extend the Claude for Life Sciences base. A research-quality angle follows: once Coefficient’s planning and regulatory tooling runs inside Claude, drug R&D plans become artifacts a model drafts and a human audits — the pattern agentic coding tools long ago set for software. Third, the vertical-integration contest our year-opening outlook flagged is accelerating: pharma either builds AI capability of its own or stands by as people and tooling head for the frontier labs.