Bristol Myers Squibb today announced it will expand its compute infrastructure to deploy an NVIDIA DGX SuperPOD with DGX Vera Rubin NVL72 systems, which the company said will be the most powerful and energy-efficient single-owned NVIDIA infrastructure in life sciences.
The expanded Vera Rubin infrastructure will give BMS’s scientific and computational teams access to a step-change in both computational power and efficiency, with the Vera Rubin architecture delivering up to ten times greater performance per megawatt than its predecessor.
The investment builds on nearly three years of collaboration that began when BMS first deployed NVIDIA DGX SuperPOD infrastructure to support its Research and Development activities, and marks the next step in expanding that foundation to match the growing scope and sophistication of BMS's AI-driven scientific programs across oncology, hematology, cardiovascular, immunology and neuroscience.
BMS’s investment in AI-powered research has already begun to change how the company discovers and develops medicines, and these early results span multiple areas of drug discovery. For example, AI agents that automate target identification and validation save BMS scientists weeks of manual work, freeing them to focus on hypothesis testing and the highest-value scientific decisions. From there, through the company’s “Predict First” approach, a methodology where AI-generated predictions inform experimental design before work begins at the bench, AI informs the design of every small molecule program and the majority of the company’s large molecule programs. These are just two examples that reflect BMS’s broader approach: building an integrated, AI-powered learning system that spans target identification through clinical proof of concept, helping scientists make higher-confidence decisions at every stage.
The expanded agreement advances BMS’s vision of hybrid intelligence, a model of scientific work where AI co-scientists and human researchers operate in close coordination, with AI systems handling the execution of complex, data-intensive tasks while scientists focus on direction, interpretation, and the decisions that require deep human expertise. The Vera Rubin cluster will serve as the computational backbone for this vision, supporting the development of next-generation foundation models trained on BMS's decades of proprietary data and, where the science calls for it, drawing on domain-specific capabilities from BioNeMo, NVIDIA's platform for biological AI, while powering the agentic workflows that allow researchers to evaluate hypotheses at a scale that would have been unachievable just a few years ago.
