Eli Lilly and Co. spent about four months building the LillyPod supercomputer. (IBJ photo/Michelle Kaufman)
Indianapolis-based Eli Lilly and Co. on Wednesday unveiled what the company calls “the most powerful supercomputer owned and operated by a pharmaceutical company.”
Lilly and Nvidia Corp., a Silicon Valley-based artificial intelligence chipmaker, teamed to develop the supercomputer called LillyPod that is housed at Lilly’s Indianapolis corporate campus.
LillyPod, which occupies 3,750 square feet in a space that is 30,000 square feet and the size of a football field, uses 1,016 Nvidia B300 GPUs — specialized processors called graphics processing units — designed for AI, machine learning and crunching massive amounts of data at high speeds. The noise created by the supercomputer’s cooling systems reaches more than 100 decibels, and each computer panel creates about 30 mph of windspeed when opened.
Lilly Chief AI Officer Thomas Fuchs said LillyPod will accelerate manufacturing and help the company discover and develop drugs at a faster pace. The machine, announced in October, will power an “AI factory” that brings together Lilly’s proprietary data with advanced AI for capabilities such as analyzing genome sequences and improving clinical trials.
It will also help scientists identify small molecules that could lead to oral medicines and large molecules that could lead to genetic medicines for rare diseases, Fuchs added.
“This supercomputer is truly a scientific instrument,” he said. “It will help us to look deeper, broader, faster at all types of scientific questions. We are going to use it to produce medicines faster and better.”
LillyPod will also help the company with other tasks, such as by providing scientific AI agents that will help researchers with reasoning, planning and collaboration. Other tasks will include using advanced medical imaging to better understand diseases and using AI to improve manufacturing processes.
According to Nvidia, the supercomputer will enable large-scale biomedical foundation and frontier AI models, with select models made available on Lilly TuneLab — an AI and machine-learning platform that provides biotech companies with access to drug-discovery models built on $1 billion worth of Lilly’s proprietary data.
In a separate endeavor, Lilly and Nvidia announced last month they plan to invest up to $1 billion in infrastructure and research to support a new lab near San Francisco focused on using AI technology to accelerate drug discovery. The AI co-innovation lab is expected to open early this year and place Lilly scientists and researchers alongside Nvidia AI engineers.
To explain what it means to have 1,016 GPUs of computing power, Lilly Chief Information and Digital Officer Diogo Rau said a single GPU can perform 2 million math problems in one second. And to put it another way, the LillyPod “has the equivalent computing power of taking every single iPhone on the planet and putting them together in one room,” he said.
“When we say that LillyPod will do things that no human could ever do, we absolutely mean it,” Rau said. “We are going to make breakthroughs in genomics and molecular design and peptide discovery, single cell biology and so many other areas.”
Rau also compared LillyPod to the Cray-2 computer system that Lilly installed in the 1980s. A single GPU, he said, is 7 million times as powerful as the Cray-2.
“And then we’ve got 1,016 of them,” he said. “We discovered so much from the Cray-2 supercomputer, and now we’ve got 7 million of those in a chip and 1,016 of those chips. The power is just, no matter how you look at it, it’s mind-blowing.”
Rau added that LillyPod will be a supercomputer for all purposes across Lilly and not just a departmental machine. One task, which will be powered by eight GPUs, will be to scan, index and learn all the available human research that is relevant to Lilly’s work.
“It’s going to study 37 million pieces of literature, so an enormous volume,” Rau said. “It is really hard to fathom just how powerful those machines are.”
Copyright © 2026 All Rights Reserved.