Bristol Myers Squibb is purchasing an Nvidia DGX SuperPOD to expand its AI-driven drug discovery and development capabilities.
Key facts
- •Bristol Myers Squibb is the first life sciences group to purchase a DGX SuperPOD based on Nvidia's Vera Rubin architecture.
- •The new system comprises eight DGX Vera Rubin NVL72 units designed for training models and running scientific predictions.
- •The company reports that AI-enabled research has already reduced the time to produce clinical trial candidates by 20% to 30%.
- •The new infrastructure will be integrated with existing systems to allow global teams to access shared datasets and model outputs.
- •The Vera Rubin system is expected to provide 10 times more computing capacity per megawatt than the current infrastructure.
Bristol Myers Squibb has announced the purchase of an Nvidia DGX SuperPOD based on the new Vera Rubin architecture to support its drug discovery and research operations. The pharmaceutical company will be the first life sciences organization to acquire this specific system, which it plans to integrate with its existing infrastructure to accelerate research and development.
By the numbers
Expanding Computational Capacity
The new cluster will consist of eight DGX Vera Rubin NVL72 systems, combining Vera central processing units and Rubin graphics processing units. This infrastructure will be used to train proprietary models and perform large-scale predictions for compounds and proteins. The company intends to link this new system with its current SuperPOD, which has been in operation for approximately three years, to create a shared environment accessible to researchers worldwide.
AI in Drug Development
Bristol Myers Squibb utilizes a "Predict First" approach, which employs model-generated predictions to prioritize molecules for synthesis and laboratory testing. According to company executives, AI tools have already reduced the time required to identify and produce candidates for clinical trials by 20% to 30%, with potential for a 50% reduction in the future. The new system will also provide access to Nvidia’s BioNeMo Agent Toolkit for tasks such as protein-structure prediction and molecular docking.
Operational Efficiency
The company expects the new system to deliver up to 10 times the performance per megawatt compared to the infrastructure it replaces. By utilizing Nvidia Mission Control for workload management and providing researchers with natural-language interfaces for computing tasks, Bristol Myers Squibb aims to increase accessibility to its computing resources across its global research sites.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by Artificial Intelligence News.
