Key Takeaways:
Anthropic acquired Coefficient Bio, an 8-month-old stealth AI biotech startup, for just over $400 million in stock, bringing fewer than 10 ex-Genentech computational biologists into its healthcare division.
Coefficient Bio's team, led by co-founders Samuel Stanton and Nathan C. Frey, built AI models for protein and antibody design that achieved 97-100% expression success rates in lab validation, winning an ICLR 2024 Outstanding Paper Award.
The deal represents roughly 0.1% dilution against Anthropic's $380 billion post-money valuation and positions the company against DeepMind's AlphaFold in the race to own the full AI drug discovery pipeline.
Anthropic acquired Coefficient Bio in an all-stock deal valued at just over $400 million, The Information reported on April 2, with TechCrunch and Newcomer confirming the close. The startup was formally founded roughly eight months ago, had fewer than 10 employees, no publicly known product, and no disclosed revenue. What it had was a team of ex-Genentech computational biologists with rare credentials in AI-driven drug discovery.
Co-founder Nathan C. Frey previously led multidisciplinary teams at Prescient Design, Genentech's computational drug discovery unit, working on biological foundation models and lab-in-the-loop autonomous systems. His co-authored paper on protein discovery using a method called discrete walk-jump sampling (dWJS) won an ICLR 2024 Outstanding Paper Award, generating functional antibodies with 97-100% expression success rates and 70%+ binding affinity matching or beating known benchmarks. The team joins Anthropic's healthcare and life sciences division led by Eric Kauderer-Abrams.
Against Anthropic's $380 billion post-money valuation from its February 2026 Series G, the acquisition represents roughly 0.1% dilution. The financial impact is minimal. The strategic signal is not.
Anthropic has been building its life sciences stack methodically. In October 2025, it launched Claude for Life Sciences with connectors to Benchling, PubMed, Open Targets, and ChEMBL. In January 2026, it added Claude for Healthcare with HIPAA-ready clinical trial tools. Coefficient Bio gives the company something those products couldn't: proprietary generative models purpose-built for designing functional proteins and antibodies at scale, not just analyzing existing research.
The acquisition positions Anthropic against DeepMind's AlphaFold, which predicts protein structures but does not generate new ones. Coefficient Bio's approach generates candidate molecules and validates them in the lab. The difference between structure prediction and generative design is the difference between reading a blueprint and drawing one.
The broader context is the AI industry's vertical expansion beyond chatbots and coding tools. OpenAI closed a $122 billion funding round this week to build infrastructure. Anthropic is spending $400 million on biology. The frontier labs are no longer competing on model benchmarks. They're competing on which real-world domains they can own.
People Also Ask
Q: What is Coefficient Bio? A: Coefficient Bio was a stealth AI biotech startup founded in August 2025 by ex-Genentech researchers. It built AI models for protein and antibody design, drug R&D planning, and clinical regulatory strategy before being acquired by Anthropic.
Q: How much did Anthropic pay for Coefficient Bio? A: Anthropic paid just over $400 million in stock, representing roughly 0.1% dilution against its $380 billion post-money valuation from the February 2026 Series G funding round.
Q: What is discrete walk-jump sampling? A: dWJS is a generative AI method for designing functional protein and antibody sequences. It won an ICLR 2024 Outstanding Paper Award and produced antibodies with 97-100% expression success rates in lab validation.
Q: How does this compare to DeepMind AlphaFold? A: AlphaFold predicts existing protein structures. Coefficient Bio's approach generates new functional proteins and validates them experimentally. Anthropic is adding generative design to complement structure prediction in the AI drug discovery pipeline.
Sources: