Bristol Myers Squibb (BMS) is enhancing its computational capabilities by expanding its collaboration with Nvidia through the deployment of a DGX SuperPOD featuring DGX Vera Rubin NVL72 systems. This upgrade positions BMS to possess the most advanced Nvidia infrastructure dedicated to life sciences, building on nearly three years of partnership with Nvidia.
The new Vera Rubin architecture significantly boosts performance, delivering up to ten times greater efficiency per megawatt compared to previous systems. This advancement enables BMS to tackle more complex AI tasks in drug discovery without a proportional rise in energy consumption, supporting various scientific programs including oncology and immunology.
BMS is already integrating AI into its workflows, particularly through two key initiatives. AI agents are being utilized for target identification and validation, which streamlines processes and allows scientists to concentrate on critical hypothesis testing. Additionally, the "Predict First" methodology leverages AI predictions to inform experimental design, enhancing the decision-making process in both small and large molecule programs.
The overarching vision for BMS involves a hybrid intelligence model where AI and human researchers collaborate closely. The expanded infrastructure will serve as a backbone for developing next-generation AI models trained on BMS's extensive proprietary data. This strategic move aims to enhance the accuracy of drug discovery decisions, as highlighted by BMS executives who emphasize the importance of better evidence and faster insights in navigating the complexities of drug development.