NVIDIA Eli Lilly AI Lab to Speed Drug Discovery

NVIDIA Corporation (NVDA) and Eli Lilly and Company (LLY) on Jan. 12, 2026 unveiled the NVIDIA Eli Lilly AI lab, a co-innovation facility in South San Francisco that will co-locate scientists to accelerate AI-driven drug discovery using BioNeMo and the Vera Rubin architecture.

Co-Innovation Commitment and Timeline

Announced at the J.P. Morgan Healthcare Conference, the companies committed up to $1 billion over five years (2026–2030) to fund talent, infrastructure, and compute resources. Both firms will allocate dedicated, incremental resources to the effort. The lab will bring together Lilly’s experts in biology, chemistry, and medicine with NVIDIA’s AI researchers and engineers. Operations are expected to begin in early 2026, targeting an opening by March 31. The collaboration will extend beyond discovery to clinical development, manufacturing optimization, supply-chain modeling, and commercial operations, linking research with downstream production and commercialization.

Platforms, Hardware, and R&D Focus

NVIDIA expanded its BioNeMo platform into a full open development environment, adding models for RNA three-dimensional structure prediction, ReaSyn v2 molecular synthesis reasoning, and toxicity prediction. The platform includes open datasets and training recipes to support end-to-end AI model development. In 2025, NVIDIA released more than 650 open models and 250 datasets across biology, chemistry, and medical imaging to ease model building for life-science groups.

The lab will operate on NVIDIA’s Vera Rubin architecture and DGX Cloud capacity, incorporating Omniverse, RTX PRO Servers, and the Isaac and Jetson platforms to support robotics and digital-twin simulations. Its initial technical goal is a closed-loop discovery framework that links agentic wet labs with computational dry labs. This scientist-in-the-loop, continuous AI-assisted experimentation aims to generate large-scale, high-quality lab data for training foundational and advanced models in biology and chemistry. The hardware and software choices are designed to sustain continuous training and rapid iteration within the co-located facility.

Lilly AI Assets and Use Cases

Lilly will contribute complementary data and deployment capabilities. TuneLab, launched in September 2025, offers federated AI models to roughly 1,300 biotech partners through integrations with Benchling and Revvity. Lilly has invested more than $1 billion in proprietary data generation linked to these platforms. The company’s "Lilly AI factory" supercomputer, announced in October 2025, uses over 1,000 NVIDIA Blackwell chips to support large-scale model training integrated into the lab’s workflows.

Manufacturing and automation are immediate focus areas. The lab will use digital twins for line stress-testing, supported by Omniverse and RTX PRO Servers. An autonomous lab partnership with Thermo Fisher Scientific was announced concurrently. A Multiply Labs case study demonstrated a 70% reduction in cost per dose and a 100-fold increase in throughput per square foot in cell-therapy manufacturing, illustrating potential productivity gains.

What’s Next

The lab is expected to begin operations in early 2026, with a target opening by March 31. It is designed to complement Lilly’s AI factory and TuneLab platforms, accelerating continuous cycles of experimentation, model training, and manufacturing optimization. This collaboration signals a significant expansion in compute demand for pharmaceutical R&D and advances Lilly’s AI-driven drug discovery strategy.

"AI is transforming every industry," said Jensen Huang, NVIDIA’s founder and chief executive.