Overview
- Eli Lilly will own and operate an Nvidia DGX SuperPOD using DGX B300 systems to power a dedicated AI factory for drug discovery and development.
- The buildout is slated to finish in December, with the supercomputer going online in January according to company timelines.
- The system will use more than 1,000 Blackwell Ultra GPUs to train large biomedical foundation models on millions of experiments to broaden and speed early research.
- Select proprietary models will be shared through Lilly TuneLab using federated learning, allowing biotechs to benefit from Lilly-trained models without directly exchanging sensitive data.
- Lilly and Nvidia highlight applications spanning imaging, digital twins for biomanufacturing, and AI agents for R&D workflows, while noting no AI-designed drugs are yet on the market.
 
  
 