Bristol Myers Squibb announced Monday that it is purchasing the most advanced AI computing system from chip manufacturer Nvidia, with the goal of accelerating how the company finds and develops new medicines.
The pharmaceutical giant said it will become the first life sciences company to acquire an Nvidia DGX SuperPOD built on Nvidia’s Vera Rubin platform. Nvidia introduced the Vera Rubin architecture earlier this year as the next step beyond its current lineup of AI computing systems.
The price of the deal was not made public. Company executives noted the purchase builds on an earlier, smaller SuperPOD system Bristol Myers had previously obtained from Nvidia — a system that is roughly two to three generations older than the new Vera Rubin-based hardware.
Across the pharmaceutical industry, companies are pouring more money into AI infrastructure in hopes of pinpointing drug targets more quickly and increasing the likelihood that experimental treatments will hold up in clinical trials.
Robert Plenge, the company’s chief research officer, explained that the upgraded system will allow Bristol Myers to evaluate far more potential drug candidates during the early phases of development. “Maybe before we could do 10 and now we can do dozens,” he said.
Plenge added that AI tools are already trimming the time it takes to prepare medicines for clinical testing by 20% to 30%, and he believes that figure could climb as high as 50% in the years ahead. He also noted that one experimental treatment for sickle cell disease currently in early-stage clinical development likely would never have been identified without the help of AI-driven research.
Greg Meyers, the company’s chief digital and technology officer, said the decision to invest in the new system was partly driven by the rapidly expanding computing needs that come with deploying larger AI models throughout the company’s research operations. Bristol Myers currently uses AI across all of its small-molecule drug programs and in most of its large-molecule efforts.
Meyers also highlighted that the new system offers significant energy efficiency gains compared to older hardware. “When you host these things, you have to pay an electric bill,” he said. “Think of it as 10 times more compute capacity per watt spent … Electricity is not getting cheaper.”
