PRINCETON, N.J. — Bristol Myers Squibb said it will expand its artificial intelligence computing infrastructure through a broader collaboration with NVIDIA, deploying a DGX SuperPOD powered by DGX Vera Rubin NVL72 systems.
The pharmaceutical company said the new infrastructure will give its scientific and computational teams greater capacity to train proprietary AI models, analyze research data and support drug discovery and development programs across oncology, hematology, cardiovascular disease, immunology and neuroscience.
According to Bristol Myers Squibb, the Vera Rubin architecture can provide as much as 10 times greater performance per megawatt than its predecessor, allowing the company to run larger and more complex AI workloads without a proportional increase in energy consumption.
The investment builds on nearly three years of collaboration between the companies. Bristol Myers Squibb previously deployed NVIDIA DGX SuperPOD infrastructure to support its research and development operations.
“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 Bristol Myers Squibb. “We’re committed to translating AI into real outcomes for patients which requires infrastructure built to match that ambition. 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.”
Bristol Myers Squibb said it is already using AI agents to automate parts of target identification and validation, reducing weeks of manual work for scientists and allowing them to focus on hypothesis testing and other research decisions.
The company also uses what it calls a “Predict First” approach, in which AI-generated predictions inform experimental design before laboratory work begins. Bristol Myers Squibb said AI contributes to the design of every small-molecule development program and most of its large-molecule programs.
The company is seeking to create an integrated AI-supported research system spanning target identification through clinical proof of concept. The goal is to use data from experiments and clinical studies to improve decision-making throughout the development process.
“Drug discovery is a sequence of decisions made under uncertainty, and better decisions come from better evidence, faster,” said Robert Plenge, Executive Vice President and Chief Research Officer at Bristol Myers Squibb. “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. The goal isn’t speed for its own sake; it’s raising the probability that each program we advance is the right one.”
The expanded infrastructure will also support Bristol Myers Squibb’s model of “hybrid intelligence,” in which AI systems perform complex and data-intensive tasks while scientists oversee research direction, interpret results and make decisions requiring human expertise.
The Vera Rubin computing cluster will support the development of foundation models trained on decades of Bristol Myers Squibb’s proprietary scientific data. The company may also use capabilities from BioNeMo, NVIDIA’s biological AI platform, to support molecular design, biological research and other specialized workflows.
“BMS has built decades of extraordinary scientific knowledge across some of the most complex areas of human disease,” said Rory Kelleher, Senior Director of Business Development, Healthcare and Life Sciences at NVIDIA. “With NVIDIA Vera Rubin and BioNeMo Agent Toolkit, BMS has the ability to transform that enterprise scientific data into proprietary intelligence and give agents domain-specific tooling that help scientists explore biology, design molecules and evaluate hypotheses at unprecedented scale — reinventing how medicines are discovered and developed.”
