Bristol Myers Squibb Deploys NVIDIA AI Supercomputer for Drug Discovery

Bristol Myers Squibb announced Monday it will deploy a second NVIDIA DGX SuperPOD built on DGX Vera Rubin NVL72 systems, significantly expanding the computing power available to its research organization. The pharmaceutical company says the new infrastructure delivers up to ten times greater performance per megawatt than the systems it replaces.

The move extends a partnership that began approximately three years ago when Bristol Myers Squibb first deployed an NVIDIA DGX SuperPOD for research and development operations. Financial terms of the latest deployment were not disclosed, according to details first reported by Quartz.

Compute demand drives infrastructure expansion

Greg Meyers, Bristol Myers Squibb's chief digital and technology officer, told Reuters the investment responds to growing computational demands as the company deploys larger AI models across its research operations. "When you host these things, you have to pay an electric bill," Meyers said. "Think of it as 10 times more compute capacity per watt spent. Electricity is not getting cheaper."

The company's existing system is already running at capacity. Erin Davis, vice president of research business insights and technology, told NVIDIA's blog that Bristol Myers Squibb is "in production with some very large-scale predictions around large molecules" and building proprietary foundational models that require substantial GPU resources.

Accelerating drug candidate screening

Robert Plenge, Bristol Myers Squibb's chief research officer, said the additional computing power will enable the company to screen far more drug candidates in early development stages. "Maybe before we could do 10 and now we can do dozens," Plenge told Reuters.

Plenge noted that AI tools have already reduced the time needed to produce medicines for clinical testing by 20% to 30%. He suggested that figure could reach 50% over the next several years. As a specific example, he pointed to a sickle cell disease drug candidate now in early-stage trials that AI-enabled research helped identify.

Unified global infrastructure

The new cluster will support development of AI foundation models trained on Bristol Myers Squibb's proprietary data and will power agentic workflows across oncology, hematology, cardiovascular, immunology, and neuroscience programs. The system will also leverage BioNeMo, NVIDIA's platform for biological AI.

Both supercomputer systems will be combined into a unified environment accessible from every Bristol Myers Squibb site globally, the company said.

Why it matters

The deployment represents a significant bet on AI infrastructure by a major pharmaceutical company at a time when computational demands for large language models and biological simulations are outpacing traditional hardware capabilities. The 10x performance-per-watt improvement addresses a critical constraint: energy costs that scale with model size. If Bristol Myers Squibb achieves its projected 50% reduction in development timelines, the infrastructure investment could materially impact time-to-market for new therapies across multiple disease areas.

Bristol Myers Squibb has been expanding AI capabilities on multiple fronts, including a separate deal with Anthropic to provide its workforce of more than 30,000 access to the Claude AI model. NVIDIA also maintains AI computing partnerships with Eli Lilly and Amgen, according to Bloomberg.

Details were first reported by Quartz.