- Chai Discovery raised a $400M Series C, nearly tripling its valuation to $3.8 billion just seven months after a $130M round at $1.3 billion.
- Index Ventures led the round; OpenAI, Thrive Capital, Sequoia Capital, Kleiner Perkins, and Menlo Ventures are all on the cap table.
- Pfizer, Eli Lilly, and Novartis have signed on, with Pfizer’s deal granting Chai access to its proprietary data plus its Chai-3 model.
- The new Chai-3 model roughly doubles hit rates for molecular targets, yet no AI-designed drug has won regulatory approval to date.
For a decade, “AI for drug discovery” was mostly a pitch-deck promise. Chai Discovery just turned it into a $400 million check. The San Francisco startup closed a Series C that nearly tripled its valuation to $3.8 billion, seven months after its previous round, with OpenAI on the cap table and Pfizer, Eli Lilly, and Novartis already signing deals. Co-founder and CEO Joshua Meier’s framing is blunt: “AI drug discovery has moved from promise to deployment.”
Inside the $400M Round
A valuation that tripled in seven months
Index Ventures led the Series C, joined by Kleiner Perkins, Sequoia Capital, and Dimension. New backers Bain Capital Ventures, Battery Ventures, Baillie Gifford, BDT & MSD, Sapphire Ventures, and Avra Capital came in alongside existing investors including Thrive Capital, OpenAI, Oak HC/FT, Menlo Ventures, and General Catalyst. The round pushes Chai’s total funding to roughly $630 million.
The pace is the real headline. Chai raised $130 million at a $1.3 billion valuation in December 2025; seven months later it is worth $3.8 billion. Valuations that nearly triple in under a year usually signal that investors see commercial pull, not just technical promise, and Chai’s growing roster of pharma customers is the evidence they are pointing to.
Business Insight — OpenAI’s presence on the cap table is a strategic tell. The frontier-model labs increasingly want exposure to vertical AI applications where their models create defensible value, and biology is one of the largest untapped markets for that thesis.
How Chai-3 Designs Molecules
Searching a quintillion possibilities
Chai builds frontier AI models that predict and reprogram how biochemical molecules interact. Its focus is antibodies, the body’s custom-built locks for identifying invaders. The catch is scale: there are roughly a quintillion possible antibody configurations, and the traditional approach meant screening millions of candidates one at a time. Chai’s system runs rapid simulations against a disease target and curates that vast space down to high-probability candidates, rather than guessing.
The company calls its latest model, Chai-3, a “step-change” over Chai-2, reportedly doubling success rates on molecular-interaction targets to an implied 35% to 40% hit rate, with gains in binding affinity and broader antibody design. That is the metric pharma partners care about, because a higher hit rate compresses the most expensive, time-consuming stage of early discovery.
Business Insight — Hit rate is Chai’s real product. Doubling it doesn’t just improve accuracy; it changes the unit economics of R&D, which is why a software model can command deals from companies that spend billions on wet-lab experiments.
Why Big Pharma Is Signing Up
Pfizer, Lilly, and Novartis all in
Chai’s commercial traction is what separates it from the many AI-biology startups still selling potential. It struck a landmark licensing agreement with Pfizer that grants access to Chai-3 plus an AI model trained on Pfizer’s proprietary data, signed a customer agreement with Eli Lilly, and formed a formal collaboration with Novartis. Three of the world’s largest drugmakers validating one platform in a single year is rare.
For pharma, the appeal is leverage: rather than build frontier AI in-house, they rent Chai’s models and pair them with their own molecular libraries. For Chai, those deals are both revenue and a data moat, since training on partners’ proprietary datasets makes the next model harder for rivals to match. It is the classic enterprise-AI flywheel, applied to biology.
Business Insight — Watch the deal structure, not just the logos. Whether Chai captures value through flat software licenses or milestone-and-royalty terms tied to drugs that reach the clinic will determine if this is a healthy SaaS business or a call option on future blockbusters.
The Catch: No Approved Drug Yet
The Phase II cliff still stands
The sobering context: despite roughly $20 billion poured into generative-AI drug discovery, no AI-designed drug has been approved. More than 173 AI-originated programs are now in clinical development, up from around two dozen in 2023, and 15 to 20 are expected to reach trials in 2026. The bottleneck is biology, not chemistry: AI-driven candidates post strong Phase I pass rates of 80% to 90%, but that collapses to roughly 40% in Phase II, in line with traditional methods.
In other words, Chai’s models are demonstrably good at the design problem, but the industry has not yet proven AI shortens the part of the journey that costs the most and kills the most candidates. A $3.8 billion valuation prices in the belief that better molecular design eventually translates into better clinical outcomes, a link that remains unproven at scale.
Business Insight — The category’s next milestone isn’t a bigger round; it’s a first approval. The AI-drug startup that gets a molecule through Phase III will reset the market’s willingness to pay, and until then, valuations like Chai’s rest on conviction more than proof.
Related
- Novo Nordisk Just Handed OpenAI the Keys to Its Drug Lab
- Alphabet Just Quietly Bet $2.1B on Curing Every Disease
- The Startup Making AI 20x Cheaper Just Raised $800M
- Anthropic Just Bet $200M on One Question About Work
Sources
- SiliconANGLE — Chai Discovery nabs $400M Series C as AI-designed antibodies reach Big Pharma
- Endpoints News — Chai Discovery gets $400M, tripling valuation from seven months ago
- Fierce Biotech — Chai brews up $400M Series C to fuel AI used by Lilly, Novartis and Pfizer
AI Biz Insider · AI Business EN · aibizinsider.com
댓글 남기기