TL;DR
Nvidia and Eli Lilly announced a joint co-innovation lab in the San Francisco Bay Area that will focus on foundation models for AI-driven drug discovery, backed by up to $1 billion of investment over five years. The effort will use Nvidia's BioNeMo software and the new Vera Rubin accelerators, and aims to create a continuous learning system linking lab work with computational modeling.
What happened
At JPMorgan Healthcare, Nvidia and Eli Lilly unveiled plans to establish a co-innovation research lab in the San Francisco Bay Area, committing as much as $1 billion across five years to develop foundation models for AI-assisted drug discovery. The partnership will combine Lilly's biologists and chemists with Nvidia's software engineers and model developers to build models on Nvidia's open-source BioNeMo platform and run training on the new Vera Rubin GPU systems. Nvidia described the lab's initial priority as building a continuous learning setup that tightly connects Lilly's experimental wet labs with computational dry labs to enable around-the-clock experimentation. The companies also intend to explore AI applications beyond discovery, including clinical development, manufacturing and commercial operations; Nvidia's Omniverse Robotics was cited as a potential tool for optimizing production. Vera Rubin hardware promises a performance increase over prior-generation chips, but broad availability is not expected until later in the year.
Why it matters
- Combining pharma domain expertise with large-scale AI infrastructure could shorten model training cycles and accelerate hypothesis testing.
- A continuous learning pipeline linking experiments and models may allow more automated, iterative research workflows that run outside normal lab hours.
- Access to high-performance GPUs such as Vera Rubin could materially increase compute capacity for large foundation models in biology and chemistry.
- If successful, the approach could be applied across drug development, clinical trials and manufacturing, potentially altering multiple stages of pharmaceutical R&D.
Key facts
- Financial commitment: up to $1 billion over five years to the joint lab.
- Announcement venue: revealed at the JPMorgan Healthcare conference.
- Location: a co-innovation lab planned for the San Francisco Bay Area, opening later this year.
- Software platform: work will use Nvidia's BioNeMo open-source framework for biology and chemistry models.
- Compute hardware: the collaboration will harness Nvidia's new Vera Rubin accelerators.
- Initial technical goal: build a continuous learning system to link wet-lab experiments with computational models for 24/7 experimentation.
- Vera Rubin performance: Nvidia says the new platform delivers roughly fivefold performance improvements over prior-generation Blackwell GPUs.
- Vera Rubin availability: chips are not expected in significant quantities until the second half of the year.
- Existing Lilly compute: Eli Lilly previously deployed a Blackwell Ultra SuperPOD with 1,016 B300 GPUs.
- Scope: lab will also explore AI use cases in clinical development, manufacturing, and commercial operations, including possible use of Omniverse Robotics for plant optimization.
What to watch next
- Timing and scale of Vera Rubin hardware deliveries and whether the lab receives early access; not confirmed in the source.
- Whether the planned continuous learning system becomes operational and how it will integrate wet- and dry-lab workflows.
- Publication of models, preprints or reproducible results from the lab's research outputs; not confirmed in the source.
Quick glossary
- Foundation model: A large AI model trained on broad data that can be adapted to many downstream tasks through fine-tuning or prompting.
- GPU (Graphics Processing Unit): A processor designed to accelerate parallel computations, commonly used to train and run large AI models.
- BioNeMo: An open-source Nvidia framework for building and training deep learning models targeted at biological and chemical applications.
- Continuous learning system: An approach where models are incrementally updated as new data arrives, enabling ongoing adaptation without retraining from scratch.
- Omniverse Robotics: Nvidia’s platform for simulating and coordinating robots and automation workflows, used to prototype manufacturing processes and logistics.
Reader FAQ
How much are Nvidia and Eli Lilly committing to the lab?
The partnership is backed by up to $1 billion over five years.
Where and when will the lab open?
The co-innovation lab is planned for the San Francisco Bay Area and is expected to open later this year.
Which software and hardware will the lab use?
Work will use Nvidia's BioNeMo software platform and the new Vera Rubin GPU accelerators.
Will the lab run experiments continuously overnight?
Nvidia says the aim is a continuous learning system that links wet and dry labs to enable 24/7 experimentation.
Are the new Vera Rubin GPUs available now?
The source says the chips are unlikely to be available in significant numbers until the second half of the year.
AI + ML Nvidia, Eli Lilly just say yes to making drugs together, using Vera Rubin GPUs If penicillin was discovered on moldy bread, who's to say the next miracle…
Sources
- Nvidia, Eli Lilly just say yes to making drugs together, using Vera Rubin GPUs
- NVIDIA and Lilly Announce Co-Innovation AI Lab to …
- Nvidia, Eli Lilly commit $1B to AI drug discovery lab
- NVIDIA Kicks Off the Next Generation of AI With Rubin
Related posts
- PC shipments to stall in 2026 as AI demand soaks up memory supply
- Anthropic’s cutoff of third‑party coding clients risks customer fallout
- Building a 25 Gbit/s Testbench Workstation for the SCION Association