Eli Lilly and Company unveiled that it is building the most powerful supercomputer owned and handled by a pharmaceutical company, in collaboration with NVIDIA. The supercomputer will work on an "AI factory," a specialized computing infrastructure that manages the entire AI lifecycle from data ingestion as well as training to fine-tuning and high-volume inference.
The supercomputer uses the world’s first NVIDIA DGX SuperPOD with DGX B300 systems. With over 1,000 B300 GPUs working together on a single high-speed network, everything, from GPUs to storage, communicates in one unified fabric.
The new supercomputer and AI system let scientists learn and try things much faster. They’ll be able to train AI on millions of experiments to explore possible medicines — greatly broadening how and what drugs can be discovered.
Many of these AI models will be offered on Lilly TuneLab, a shared AI/ML drug-discovery platform made for the wider biopharma community. TuneLab will keep growing its model collection, including adding workflows that use select NVIDIA Clara open-source models.
The company plans to adopt the supercomputer to shorten development cycles and simultaneously help get medicines to people faster. New scientific AI agents can support researchers in reasoning, planning and collaborating across digital and physical ecosystem.
Executive Statement
According to Diogo Rau, executive vice president and chief information and digital officer at Lilly, Lilly's mission is to make life science better for people around the world, and today that requires excellence not just in science but also in technology. They don't believe any other company in their industry is doing what we do at this scale. As a 150-year-old medicine company, one of their most powerful assets is decades of data. With purpose-built AI models and AI, they can set a new scientific standard that accelerates innovation to deliver medicines to more patients, faster.