Nvidia and Eli Lilly announced an AI co-innovation lab on January 12, 2026 at the J.P. Morgan Healthcare Conference. The two companies said they will jointly invest up to $1 billion over five years in talent, infrastructure and compute. The lab is based in the San Francisco Bay Area, with work expected to begin in South San Francisco in early 2026. As of November 13, 2026, we found no announcement of a formal opening. Coverage in June still described the work as "expected to begin".
Key Facts
| Item | Detail |
|---|---|
| Announced | January 12, 2026 |
| Investment | Up to $1 billion over five years, from both companies together |
| Split between the companies | Not disclosed |
| Location | San Francisco Bay Area. Work to begin in South San Francisco |
| Staffing | Lilly scientists working alongside Nvidia model builders and engineers. No headcount given |
| Software platform | Nvidia BioNeMo |
| Compute | Infrastructure intended to use Nvidia's Vera Rubin architecture |
| Manufacturing tools | Nvidia Omniverse libraries and RTX PRO Servers for digital twins of production lines |
| Opening date | Not announced |
The $1 billion is a ceiling on joint spending. It is not a payment from one company to the other, and it is not a licensing deal with milestones like most entries in the AI pharma deal tracker.
What the Lab Is For
- First project. A continuous learning system that connects Lilly's automated wet labs with computational dry labs, so experiments can run around the clock with a scientist overseeing them.
- Models. New foundation models for biology and chemistry.
- Later work. Clinical development, manufacturing and commercial operations, including robotics and digital twins of manufacturing lines.
In the release, Nvidia chief executive Jensen Huang said AI's "most profound impact will be in life sciences". Lilly chief executive David Ricks said the combination "could reinvent drug discovery as we know it".
The Supercomputer Behind It
The lab builds on a Lilly-owned supercomputer announced with Nvidia on October 28, 2025, which Lilly calls an AI factory and has since named LillyPod.
| Item | Detail |
|---|---|
| Announced | October 28, 2025 |
| System | Nvidia DGX SuperPOD built from DGX B300 systems, described as the first of its kind |
| GPUs | 1,016 Nvidia Blackwell Ultra |
| Stated AI performance | More than 9,000 petaflops |
| Go-live | Late February 2026, according to R&D World coverage of a ribbon-cutting |
| Use by June 2026 | Running large open-weight language models on site for staff with no token or budget limits, per a post by Lilly's chief AI officer reported by R&D World |
Lilly describes it as the most powerful supercomputer in the pharmaceutical industry. Lilly's chief AI officer, Thomas Fuchs, told R&D World in January 2026 that it is "still quite modest" next to the largest AI training clusters.
Lilly's Other AI Deals
| Date | Partner | Terms | Scope |
|---|---|---|---|
| January 2024 | Isomorphic Labs | $45M upfront, up to $1.7B in milestones | Small molecules against multiple targets |
| June 2024 | OpenAI | Not disclosed | Antimicrobials for drug-resistant bacteria |
| September 2025 | TuneLab, Lilly's own platform | Not applicable | Lets biotech companies use selected Lilly models trained on its proprietary data |
| Early January 2026 | Nimbus Therapeutics | $55M in upfront and near-term milestones, total up to about $1.3B | Oral obesity drug |
| January 2026 | Chai Discovery | Not disclosed | Biologics design and a model built only for Lilly |
| March 29, 2026 | Insilico Medicine | $115M upfront, total about $2.75B | Licence to preclinical oral drugs and joint programmes |
The release also says Nvidia's Clara open models are planned as part of a future TuneLab workflow. For why the "up to" figures in this table are much larger than the upfront payments, see biobucks. For Isomorphic's progress since 2024, see Isomorphic Labs funding and clinical trial status.
Other Nvidia Pharma Work
On the same day as the Lilly announcement, Nvidia announced work with Thermo Fisher Scientific to put its AI into laboratory instruments, according to GEN. Nvidia is also among the organisations named at the launch of OpenAI's GPT-Rosalind model, covered on the OpenAI and Novo Nordisk page.
Methodology
Lab facts and quotes come from the joint Nvidia and Lilly press release of January 12, 2026. Supercomputer specifications come from R&D World's reports of October 2025, January 2026 and June 2026. The Thermo Fisher item is from GEN. Deal terms are sourced on the deal tracker page. Status as of November 13, 2026.