Summary: Chai Discovery will connect its generative design software to Eli Lilly’s TuneLab program, while argenx will use its platform for de novo antibody research. The January 2026 profile says the startup began in OpenAI’s offices and closed a $130 million Series B at a $1.3 billion valuation in December 2025. An investor expects first-in-class medicines from these models to enter clinical trials by the end of 2027.
TechCrunch ran a lengthy profile on January 16, 2026, following Chai Discovery — the San Francisco startup that began as four people working out of OpenAI’s offices and, in a bit over a year, turned itself into one of the most sought-after names in AI drug development. Seven days before the piece appeared, the company’s own announcement (January 8, via Business Wire) laid out the newest moves: Eli Lilly will hook Chai’s generative design software into its TuneLab drug-discovery program, and the Netherlands-based pharma argenx will run de novo antibody research on the Chai platform.
What deserves attention here is not the size of any deal but the validation behind it. Most “AI for science” companies are still at the publish-papers-and-chase-benchmarks stage; Chai has already placed its models with several of the companies best positioned to put them to the test. “2025 was the year we proved AI could transform preclinical drug discovery. 2026 will be the year of deployment,” cofounder Joshua Meier summed up on LinkedIn.
Born Under OpenAI’s Roof
OpenAI runs through Chai’s backstory. Cofounder Josh Meier was already on OpenAI’s research and engineering team in 2018, and cofounder Jack Dent first met him in Harvard computer science classes. Roughly six years before the company existed, Sam Altman messaged Dent — at that point a Stripe engineer — to float whether Meier would do a proteomics spinout with OpenAI. Meier read the technology as not yet ready, turned the offer down, headed to Facebook and helped build ESM1, described as the first transformer protein-language model, then logged three years at Absci.
In 2024 the roles reversed: “Josh and I reached back out to Sam,” Dent says. Chai came to life with OpenAI among its earliest seed investors, and the team set up initially inside OpenAI’s offices in San Francisco’s Mission neighborhood. “They were kind enough to give us some office space,” Dent recalled. The backstory doubles as the thesis: point frontier-model engineering at proteins and antibodies, a domain rich in data and poor in tooling.
Fundraising at Speed: Hundreds of Millions in Roughly Twelve Months
Chai’s fundraising traces execution compounding at full tilt. In December 2025 the company closed a $130 million Series B at a $1.3 billion valuation, and barely twelve months after founding, total raised stood in the “hundreds of millions.” General Catalyst is a major backer, with managing director Elena Viboch among the deal’s champions.
The sums on the pharma side make a useful contrast. Days after the Chai collaboration went public, Lilly committed to a separate $1 billion partnership with Nvidia to set up an AI co-innovation lab in San Francisco. Betting on an outside software startup while building its own compute lab tells you where big pharma believes the constraint sits: at the point where models meet data, not in raw compute.
The Deal at the Center: Inside Lilly’s TuneLab
The Lilly partnership is what the whole story turns on. Chai’s generative design models will be wired into TuneLab, Lilly’s in-house AI/ML drug-discovery program led by Aliza Apple. Apple’s framing puts the value proposition plainly: “By combining Chai’s generative design models with Lilly’s deep biologics expertise and proprietary data,” the goal is better molecules designed faster.
On the engineering side, the flagship algorithm Chai-2 is aimed at antibodies — the proteins that fight illness. Dent insists that “every line of code in our codebase is homegrown” and the architectures are “highly custom,” with no fine-tuning of open-source LLMs involved. The better mental model is a computer-aided design suite for molecules, not a chatbot vendor. Lilly and argenx are not alone on that list — Pfizer is on it too.
Putting Dates on It: From Deployed Models to the Clinic
The investors have said the timeline out loud. Viboch holds that there are “no fundamental barriers to deployment of these models in drug discovery,” sees pharma partnerships arriving across 2026, and anticipates “by the end of 2027 seeing first-in-class medicines enter into clinical trials.”
A demanding schedule, by any standard. Antibody design was never mainly about generating a plausible structure — manufacturability, immunogenicity, and what each pass of wet-lab validation costs determine which candidates survive. Generative models shrink the distance between design and feedback, but the wet lab keeps its own tempo. Whether the thesis holds will depend on pharma partners pushing proprietary data back into the platform until a flywheel turns. For engineers tracking AI for science, this is the textbook case: once model capability gaps close, the next winners get picked by who holds the deployment venues and the proprietary data.