.svg){kind=spacer}
Details
- Eli Lilly and NVIDIA have unveiled the world's first NVIDIA DGX SuperPOD with DGX B300 systems, powered by 1,016 Blackwell Ultra GPUs and capable of delivering over 9,000 petaflops of AI performance—making it roughly 7 million times more powerful than Lilly's Cray supercomputers from 1992.
- This dedicated "AI factory" is designed to train biomedical foundation and frontier models, leveraging over $1 billion in experimental data that Lilly has amassed across 150 years of pharmaceutical research.
- The infrastructure will enable a broad array of AI applications, including novel drug discovery using NVIDIA BioNeMo, precision medicine through genomics analysis, advanced clinical trial operations, manufacturing optimization with digital twins, and continuous autonomous workflows in labs.
- Lilly is making these AI capabilities accessible through TuneLab, a federated learning platform now featuring both Lilly models and NVIDIA's Clara foundation models, ensuring partner collaboration without compromising data privacy via NVIDIA FLARE.
- This investment positions Lilly as the top AI-ready pharmaceutical company and supports its $50 billion U.S. expansion, including a planned $4.5 billion facility in Indiana expected to create 13,000 jobs.
Impact
This milestone cements Lilly’s position as a leader in the race for AI-driven drug discovery, outpacing competitors by scale and proprietary data. With rivals like Novo Nordisk also investing in similar supercomputing platforms, Lilly's deployment of the largest pharma-owned AI infrastructure sets a new bar for the industry. The move exemplifies the shift toward computational research in biopharma, promising accelerated timelines, reduced costs, and enhanced U.S. competitiveness in biotechnology.
