Bristol Myers Squibb is expanding its AI infrastructure through a deeper partnership with NVIDIA, deploying DGX SuperPOD systems built on NVIDIA’s Vera Rubin architecture to accelerate AI-driven drug discovery across its research portfolio. The two companies describe the deployment as creating the most powerful and energy-efficient single-owned NVIDIA compute infrastructure in the life sciences sector. The upgraded systems will reportedly enable BMS to train larger foundation models on proprietary scientific and clinical datasets while improving computational efficiency.
The deployment includes NVIDIA’s BioNeMo platform, which provides AI models and tools for molecular design, biological reasoning and hypothesis generation. BMS said the infrastructure will support its “Predict First” strategy, in which AI-generated predictions guide experimental design before laboratory validation. Financial terms were not disclosed.
BMS joins a growing number of pharmaceutical companies investing heavily in dedicated AI infrastructure for drug discovery, following similar initiatives by Eli Lilly and other large drugmakers. The NVIDIA deployment complements BMS’s recent agreement with Anthropic to deploy Claude Enterprise across research and corporate operations, reflecting a broader strategy to combine specialized scientific AI with enterprise-wide generative AI tools.
While the impact on drug development timelines remains to be demonstrated, the investment underscores how proprietary AI infrastructure is increasingly viewed as a strategic asset in pharmaceutical R&D, alongside data and drug discovery expertise.
Access the AllSci platform to explore the science behind the news.