Eli Lilly and computing giant Nvidia are teaming up to build a new supercomputer that the pair say will be the most powerful owned and operated by a pharmaceutical company.
Lilly will use the new tool to fuel an “AI factory,” the Indianapolis pharma said in an Oct. 28 release, a computing infrastructure that can take in data, train models on experiments performed by Lilly scientists and generate inferences.
The supercomputer will be housed within Lilly’s existing facilities and run entirely on renewable energy, the company said.
Lilly plans to use its new computing power for molecule discovery research and to shorten development cycles, according to the release, and it will also make some of its models available on its TuneLab platform. TuneLab allows early-stage biotechs to access Lilly’s models in return for providing training data for further model creation.
"Lilly is shifting from using AI as a tool to embracing it as a scientific collaborator," Thomas Fuchs, Lilly’s senior vice president and chief AI officer, said in the release. "This isn't just about speed, but rather interrogating biology at scale, deepening our understanding of disease and translating that knowledge into meaningful advances for people served by Lilly medicines as well as the broader life sciences ecosystem."
Some of the uses for the supercomputer-enabled AI tool include AI agents that can help scientists with reasoning, advanced medical imaging and digital twins that can team up with Lilly’s robotic systems to speed up manufacturing, the Big Pharma said.
The supercomputer will be the world’s first Nvidia DGX SuperPOD with DGX B300 systems, Lilly said, and will run on a single network powered by more than 1,000 graphics processing units.
Though Lilly’s may end up being the most powerful, the drugmaker is hardly the first pharma to tap Nvidia for a new supercomputer. Novo Nordisk signed on to use a Nvidia-made supercomputer in Copenhagen in June, while the California chipmaker is also behind the U.K.’s most powerful supercomputer, Cambridge-1, built in 2020 in collaboration with GSK, AstraZeneca and the National Health Service.