BMS, NVIDIA to build 'most powerful' AI factory

Bristol Myers Squibb has said it is the first life sciences company to buy a new AI supercomputing infrastructure blueprint from NVIDIA, saying it will deliver a "step change" in computational power and efficiency.

The US pharma group is deploying an NVIDIA DGX SuperPOD, which is based on the chip giant's Vera Rubin NVL72 system – a next-generation AI and supercomputing platform, specifically designed for complex and autonomous agentic AI workflows and scientific computing named after acclaimed dark matter researcher Vera Cooper Rubin.

According to BMS, the system will deliver "the most advanced and most energy-efficient NVIDIA infrastructure in life sciences" and will be used to scale proprietary AI models, truncate drug discovery timelines, and advance its "hybrid intelligence" vision, in which AI scientists work alongside researchers working on new medicines.

The deal will equip the company with the industry's most powerful centralised AI supercomputer, which BMS said will operate with up to 10 times greater performance per megawatt than predecessor systems and allow it to "pursue larger and more sophisticated AI workloads without a proportional increase in energy consumption."

In a clear sign of the AI arms race underway in the pharma sector, BMS's announcement comes just a few months after Eli Lilly made similar claims on the back of a partnership with NVIDIA, saying it did not believe any other company was operating at the scale of its project.

"BMS has made a deliberate bet on AI, and we are beginning to see it pay off in our pipeline and operations," said Greg Meyers, chief digital and technology officer at BMS, which indicated it has been building up to this project for nearly three years.

"Expanding our compute capabilities with NVIDIA gives our researchers and teams across the business the scale they need to keep BMS at the leading edge of what AI can do for drug discovery & development," he added. The financial terms of the deal have not been disclosed.

Drug discovery using AI is being trumpeted as a way to shorten the time to lead candidate selection, reduce costs, and improve success rates, thanks to its ability to process large-scale datasets, uncover patterns, and generate predictions that can be deployed in the discovery of new targets and drug design.

BMS says its investment in AI has already started to pay off, with AI agents that automate target identification and validation saving its scientists weeks of manual work and freeing them to focus on hypothesis testing and other scientific decisions.

AI already informs the design of every small molecule programme at BMS, and the majority of its large molecule programmes, in what it calls a 'predict first' approach that comes ahead of any lab work.

"Drug discovery is a sequence of decisions made under uncertainty, and better decisions come from better evidence, faster," said Robert Plenge, BMS's chief research officer.

"This infrastructure lets us learn from every experiment and every clinical readout to sharpen the next hypothesis, allowing BMS scientists to spend less time on manual work and more time on the questions that require human judgment," he added.

"The goal isn't speed for its own sake; it's raising the probability that each programme we advance is the right one."