🔬The BioAI Phase Shift - Matthew McPartlon & Neil Patil, Chai Discovery
Latent Space RJ Honicky
Chai Discovery says pharma tools are finally good enough to trust. That’s why a 2-year-old startup is sitting at the center of big drug deals.
Based on reporting by Latent Space, RJ Honicky — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
January’s JPM Pharma conference in San Francisco was supposed to be about the usual deal-making circus. Instead, a new class of AI tools deals stole part of the show, and Chai Discovery ended up right in the middle of it. The company is backed by OpenAI and now valued at $4 billion, which is a strange place for a 2-year-old to land, but the bigger story is that pharma seems to have moved from curiosity to actual spending.
For years, AI-for-pharma companies often drifted toward building their own drug programs. That made sense. To sell software to drugmakers, you first had to prove the software worked. And if you already had strong targets or clinical evidence, it was often easier to raise money around a specific asset, or strike a milestone-heavy licensing deal, than to pitch a platform on faith.
What changed, according to Chai’s founders, is that the models got good enough for drug design teams to trust them. That matters because better tools can speed up the whole discovery loop: more candidates reach lab and animal tests faster, toxicity can be screened earlier, delivery can be improved, and the odds of something making it to clinic go up. It also opens the door to things that are painfully hard in a lab, like designing precise antibodies or bi-specific antibodies that bind two proteins at once.
The Chai pitch is not just that the models are better. It’s that better models shorten iteration time enough to change the work itself. Instead of treating AI as a neat efficiency layer, Chai is aiming at something closer to engineering: a workflow where you can get a molecule that’s already close to right, right out of the gate.
That’s why Chai has spent heavily on partnerships with pharma companies. Working that closely, the team says, helps shape the product around real research needs rather than hypothetical ones. It also shows up in the interface: the company is building tools more like CAD or graphics software than a chatbot, because nobody designing molecules wants to chat with a box when they can move things around directly. Since June, Chai has added deals with Lilly, Novartis, and argenx, plus an expansion of its Eli Lilly program.
The episode’s core argument is blunt: once the model is good enough, the best product starts to matter. In pharma, that’s not a cute slogan. It’s a budget line.
My take — AI-written commentary, not fact-checked reporting
This is the part where the AI crowd stops pretending every industry wants a chatbot. Pharma wants fewer excuses, better molecules, and less theater. The industry has spent years paying for promises; now it’s paying for tools that actually change what can be designed. Dry, but sensible — which is usually how the real money moves.
Read more about this at: Latent Space