Bristol Myers Squibb Doubles Down on AI Drug Discovery with Second NVIDIA DGX SuperPOD
Key Takeaways
- ▸BMS deployed a second NVIDIA DGX SuperPOD with 8x Vera Rubin NVL72 systems, providing 10x performance per megawatt over legacy infrastructure
- ▸Three years of DGX operations have proven measurable impact: AI-enabled target ID saves weeks of work, CELMoD compound library expanded to new disease targets, and lead optimization methodology prioritizes molecules for synthesis
- ▸Opening compute access to all BMS scientists—removing researcher wait times and compute limits—shifts focus from resource logistics to high-value scientific decisions
Summary
Bristol Myers Squibb announced the deployment of its second NVIDIA DGX SuperPOD, built on eight DGX Vera Rubin NVL72 systems—the most powerful AI cluster in life sciences—to democratize access to computational resources across its research organization. After three years of operating its first SuperPOD with measurable results, BMS is scaling AI applications across its entire drug discovery pipeline, from target identification and CELMoD compound expansion to lead optimization using their "Predict First" methodology that prioritizes high-probability molecules for experimental validation.
The new infrastructure, featuring NVIDIA Vera CPUs and Rubin GPUs, delivers 10x the performance per megawatt compared to the systems it replaces and includes access to NVIDIA's BioNeMo Agent Toolkit for biological AI. By eliminating compute bottlenecks and researcher wait times, BMS is transforming drug discovery from a resource-constrained process to what VP Erin Davis calls "Limitless Compute"—enabling AI-enabled target identification that saves scientists weeks of manual work and accelerates the discovery of treatments across cancer, brain health, and other therapeutic areas.
- NVIDIA's BioNeMo Agent Toolkit enables agentic workflows across the full drug discovery pipeline, from predictions and model training to multi-parameter optimization
Editorial Opinion
NVIDIA's partnership with BMS demonstrates the transformative potential of accessible AI infrastructure in life sciences, where computational bottlenecks have historically constrained discovery timelines. By democratizing access to massive compute and agent-based workflows, BMS is modeling how enterprises can shift from incremental AI pilots to organization-wide integration—proving that the constraint isn't the technology, but enabling scientists to use it. This signals a broader industry inflection where pharmaceutical R&D becomes compute-first and AI-native, likely reshaping competitive advantage in drug discovery.
