Details
- Eli Lilly launched LillyPod on February 26, 2026, in Indianapolis, the most powerful AI factory wholly owned by a pharmaceutical company, delivering over 9,000 petaflops with 1,016 NVIDIA Blackwell Ultra GPUs assembled in four months.
- Involves Eli Lilly executives including Diogo Rau (EVP CIO), Thomas Fuchs (Chief AI Officer), Tim Coleman (CTO), and NVIDIA partnership; supports genomics, protein diffusion, small-molecule models, and Lilly TuneLab platform.
- Powers large-scale AI training for drug discovery via NVIDIA DGX SuperPOD, Spectrum-X networking, and BioNeMo models; enables simulation of billions of molecules in a 'dry lab,' surpassing wet lab limits of ~2,000 ideas/year.
- Builds on prior Lilly-NVIDIA collaborations like AI factory project and announced co-innovation lab with $1B investment over five years focused on real-world lab data and federated learning via TuneLab.
- LillyPod handles 700TB data with 290TB GPU memory; targets 100% renewable energy by 2030; first pharma deployment of DGX B300, positioning Lilly ahead in in-house AI vs. cloud-reliant peers.
Impact
LillyPod accelerates AI-driven drug discovery by enabling in-silico testing of billions of molecules, slashing timelines and costs in a R&D-intensive industry racing against compute limits. It sets a benchmark for pharma adopting full-stack NVIDIA infrastructure, potentially widening Lilly's edge over rivals like those using cloud AI. Combined with TuneLab's ecosystem and upcoming co-innovation lab, it fosters collaborative model improvement while prioritizing data privacy.
