Bristol Myers Squibb is deploying a second NVIDIA DGX SuperPOD with eight of the most powerful AI systems in life scienc
Bristol Myers Squibb announced the deployment of its second NVIDIA DGX SuperPOD, built on eight DGX Vera Rubin NVL72 systems that represent the most powerful and energy-efficient AI cluster in life sciences. Called the "SuperDuperPOD" by Erin Davis, vice president of research business insights and technology at BMS, the system delivers up to 10x the performance per megawatt compared to the infrastructure it replaces.
The core philosophy behind the expansion is democratizing access. Rather than limiting a supercomputer to a small group of researchers, BMS is opening it to every scientist in the organization. "No one has to wait, and no one is told they have a limit," Davis explains. The eight rack-scale systems, each comprising NVIDIA Vera CPUs and Rubin GPUs, will provide a unified AI platform including NVIDIA BioNeMo Agent Toolkit for biological AI, enabling predictions, model training, and agentic workflows across the full drug discovery pipeline.
BMS has already operated a DGX SuperPOD for about three years with meaningful results. AI-enabled target identification saves scientists weeks of manual work. The team has used AI to expand its library of CELMoD compounds, molecules engineered to selectively degrade cancer-causing proteins with applications in blood cancer treatment. AI is also applied in lead optimization through a methodology called "Predict First," which uses design predictions to prioritize synthesis of molecules with the highest probability of success, ensuring laboratory experiments are aligned with the most promising candidates.
Payal Sheth, senior vice president of therapeutic discovery sciences at BMS, emphasizes the goal of translating AI capabilities into measurable impact. The new system will combine with the existing DGX SuperPOD into a single unified environment accessible from every BMS site globally. Barriers from past acquisitions and the need for deep computational expertise are being replaced with AI-native tooling managed through NVIDIA Mission Control. Researchers will be able to initiate complex predictions in plain English.
The unified system creates a cumulative learning loop across BMS's global operations. Datasets from research in Lawrenceville, New Jersey, feed models that teams in San Diego can draw on. This represents a fundamental shift from earlier in Sheth's career when every project was treated discretely with learnings that did not compound into an intelligence framework. Today, every experiment, clinical readout, and partnership compounds into higher-conviction scientific decisions.
Agentic workflows further enhance the architecture by breaking down silos. Agents can learn from decisions across programs and departments, creating what Davis calls "a huge game-changer." Scientists gain access to well-vetted, fully trained virtual scientists with BMS knowledge built in, effectively multiplying individual capabilities. Human instincts are not replaced but augmented with quantitative insights and predictions. The new system has detailed resource allocation already planned across modalities from small and large molecule design to clinical applications to digital twins. Davis confirmed to BMS Chief Digital and Technology Officer Greg Meyers that she could fully saturate the SuperDuperPOD.