Nvidia (NVDA) announced on Tuesday that drugmaker Eli Lilly (LLY) is building an AI supercomputer running on more than 1,000 Blackwell Ultra GPUs. The chip giant said this will help cut down on drug discovery times and “enable breakthroughs in genomics, personalized medicine, and molecular design at industrial scale.”
Unveiled at Nvidia’s GTC in Washington, D.C., the setup, which Nvidia refers to as an AI factory, will train biomedical foundation and frontier models for the development of new drugs. According to Nvidia, some of those models will be available to third-party companies that take advantage of Lilly’s TuneLab AI platform, which will also run Nvidia’s Clara open foundation models for healthcare.
“Our foundation models are spawning new possibilities for our chemists, helping them uncover new motifs and configurations of atoms that were out of reach with traditional methods,” Lilly chief AI officer Thomas Fuchs said in a statement.
“AI gives us the means to accelerate progress toward both developing and delivering better, more personalized and targeted medicines,” he added.
Nvidia said Lilly will build its AI system on its Nvidia DGX SuperPOD and will allow Lilly to run large language models that can decrease the amount of time it takes to write up clinical trials and process medical imaging.
The deal with Nvidia is part of Lilly’s $50 billion effort to grow its US manufacturing and R&D presence, which includes four new facilities, plus a planned $4.5 billion lab in Indiana.
Nvidia said Lilly can also use its Nvidia Isaac and Isaac Sim offering to roll out robots as part of its production pipeline and develop AI agents using Nvidia’s NeMo software that can run 24/7 and conduct experiments to develop new potential treatments.
The pairing shows the breadth of Nvidia’s reach across various industries, ranging from Big Tech cloud platforms to automotive companies and more.
The announcement comes just weeks before the AI leader is set to report its third quarter earnings and follows a litany of similar GPU deals.
But Nvidia is also contending with rising competition from the likes of AMD (AMD), as well as threats from its own customers, such as Amazon (AMZN) and Google (GOOGL, GOOG), which use both Nvidia chips and their own AI processors.
On Monday, Qualcomm (QCOM) jumped into the AI data center race with its own AI accelerators, which the company said further expands the competitive landscape for Nvidia.
Still, the chip developer is largely expected to hold on to the pole position in the AI fight thanks to its massive first-mover advantage and CUDA software that allows developers to squeeze the necessary processing power out of its chips.
