Overview
- Bristol Myers Squibb announced Monday it is adding an eight-system DGX Vera Rubin NVL72 SuperPOD to its existing AI cluster to boost computing for drug discovery.
- The company says the Vera Rubin systems deliver up to about 10 times more compute performance per megawatt than the infrastructure they replace, addressing rising electricity and scale needs.
- BMS will combine the new SuperPOD with its older cluster into a single global data plane so datasets and models are shared across sites and programs.
- Researchers will gain broad access through AI-native tooling such as NVIDIA Mission Control and BioNeMo Agent Toolkit to run natural‑language predictions and agentic 'virtual scientist' workflows.
- BMS executives report existing AI work has already cut time to produce candidate medicines by roughly 20%–30%, and they expect the expanded, more efficient compute to widen chemical searches and speed early-stage testing.