BMS is expanding its AI platform to support next‑generation foundation models trained on its proprietary data, potentially leveraging NVIDIA’s BioNeMo for domain-specific biology AI and agentic workflows to scale hypothesis testing while humans guide interpretation.

The moves reflect BMS’s broader strategy to leverage advanced AI and partnerships to accelerate drug discovery, development, and operational efficiency.

The Vera Rubin cluster underpins development of next‑gen foundation models and may integrate BioNeMo for biology-focused capabilities, enabling scalable AI‑driven hypothesis testing.

The expansion follows an initial DGX SuperPOD started about three years ago, with the goal of speeding target identification from weeks to days.

Leadership emphasizes translating AI into patient outcomes, reducing manual work, and increasing the likelihood that advancing programs are the right ones.

Executives stress turning AI capabilities into measurable scientific impact, broadening access beyond a small group and easing bottlenecks in computing capacity.

Predict First — AI-generated predictions inform experimental design before lab work and guide both small‑molecule and most large‑molecule development projects.

The initiative aims to transform AI usage in drug discovery, turning insights into tangible scientific outcomes like broader target exploration and faster decisions.

Overall objective is to accelerate discovery and development with faster, more confident AI‑assisted workflows while prioritizing energy efficiency and scalability.

Robert Plenge, BMS Chief Research Officer, says the goal is to raise the probability that each program is the right one, not just to move quickly.

NVIDIA infrastructure is described as the most advanced and energy-efficient in life sciences, delivering up to ten times more performance per watt than predecessors.

The article is exclusive to STAT+ subscribers, with access limited to paying readers.