Chai Discovery raises $400M at $3.8B valuation to bring AI-designed drug molecules to Eli Lilly, Novartis, and Pfizer

Chai Discovery raises $400M at $3.8B valuation

Chai Discovery has closed a $400 million round at a $3.8 billion valuation, backed by Index Ventures, Kleiner Perkins, Sequoia Capital, and Dimension Capital as lead investors. The company builds AI models for molecular design, with Eli Lilly, Novartis, and Pfizer among its active customers.

The pitch is that drug discovery is undergoing the same compression that hit software development — except the feedback loop is measured in years and the payoff runs to blockbuster-drug scale. Chai's models work at the atomic level, placing individual atoms in three-dimensional space to construct candidate molecules, which makes the architecture fundamentally different from language or video models. The company trains from scratch in PyTorch, sometimes down to the kernel level, and generates proprietary training data in-house alongside protein sequencing and structure datasets captured via high-resolution microscopy.

“We raised $400,000,000 at a $3,800,000,000 valuation. We're super lucky to be partnered with Index, Kleiner Perkins, Sequoia, and Dimension... In July of last year, we put out a paper called CHAI2 — 'Zero Shot Antibody Design in a 24 Well Plate' — literally getting at the point where you could design these molecules without needing to fine tune them on a lot of data. And now these technologies are actively being deployed in companies like Eli Lilly, Novartis, Pfizer.”

The 2025 inflection

A year ago, success rates for AI-designed molecules were around 0.1%, and the technology was largely confined to academic curiosity. In June 2025, Chai published a paper called CHI2, subtitled "Zero Shot Antibody Design in a 24 Well Plate," demonstrating that models could design viable candidates from a simple target prompt without fine-tuning on custom data. Joshua Meier describes a presentation at a pharma company where a scientist recognized one of Chai's output molecules and said he had spent five years of his career trying to find a binder to that exact target.

Meier argues 2026 is when these tools move from side experiments to the core of entire discovery programs. Pharma companies are unlikely to publicize this loudly — IP sensitivity and competitive advantage mean adoption will lag disclosure — but the companies deploying Chai now are, in Meier's framing, among the most scientifically rigorous in the world.

Business model

Chai's current commercial model is platform licensing: pharma partners pay to deploy Chai's technology broadly across their drug portfolios, and those revenues fund better model development. The top pharma companies each spend $5–15 billion per year on R&D, which Meier notes exceeds annual semiconductor R&D spend, making the addressable market durable regardless of royalty structures.

Royalty and joint-venture arrangements, the traditional biotech deal structure, are not part of the current model but Meier leaves them open as partner-driven options. For now, the company positions itself as the platform layer underneath a pharma company's discovery engine rather than a co-developer with a stake in specific drugs.

Stack and team

Chai describes itself as simultaneously a frontier research lab, a product company, and a science company. The comparison Meier uses is training Claude while also building Cursor — the model research and the product engineering have to advance in parallel because the feedback loop from working scientists shapes what the models need to do next.

The $400 million gives Chai the capital to keep both tracks running. With Index's Nina, Kleiner's Ilya, Sequoia's Pat Grady, and the Dimension team all on the cap table from a single round, the investor lineup is unusual enough that Meier calls it rare even by the standards of well-funded AI companies.