Pharmaceutical giant Bristol Myers Squibb (BMS) is making a major investment in artificial intelligence by acquiring Nvidia’s next-generation DGX SuperPOD built on the Vera Rubin architecture. The new computing infrastructure is designed to accelerate drug discovery, train larger AI models, and expand AI capabilities across the company’s global research operations.
As AI becomes increasingly central to pharmaceutical research, companies are investing heavily in high-performance computing to process massive biological datasets, identify promising drug candidates, and reduce development timelines.
A Major Upgrade in AI Infrastructure
Bristol Myers Squibb will become the first life sciences company to purchase a DGX SuperPOD powered by Nvidia’s Vera Rubin platform. The new deployment consists of eight DGX Vera Rubin NVL72 systems that combine Nvidia’s latest CPUs and GPUs into a large-scale AI computing cluster.
The company plans to use the infrastructure to train proprietary AI models and perform predictions across research involving proteins, chemical compounds, and other biological data.
Rather than replacing its existing hardware, the new SuperPOD will work alongside BMS’s current Nvidia infrastructure, creating a shared computing environment that researchers around the world can access. The expanded platform will also allow significantly more scientists to utilize AI resources without the bottlenecks experienced on the company’s current systems.
AI Is Becoming Central to Drug Development
According to Bristol Myers Squibb, artificial intelligence now contributes to virtually every small-molecule drug program and the majority of its large-molecule research efforts.
AI is being used throughout the drug discovery process, including:
- Identifying new biological targets
- Optimizing lead compounds
- Predicting large-molecule behavior
- Training internal foundation models
The company says AI-powered target identification has already reduced certain research tasks by several weeks, while enabling scientists to evaluate a much larger pool of potential drug candidates before moving into laboratory testing.
The “Predict First” Approach
One of Bristol Myers Squibb’s primary AI strategies is a workflow it calls Predict First.
Instead of synthesizing large numbers of chemical compounds and evaluating them experimentally, researchers first use AI models to predict which molecules are most likely to meet the desired characteristics. Only the strongest candidates are then selected for laboratory synthesis and testing.
This approach helps reduce unnecessary experiments, allowing researchers to concentrate resources on compounds with the highest predicted probability of success. The company has also used AI to expand its research into CELMoD compounds, which target disease-causing proteins associated with cancer and other illnesses.
Faster Clinical Development
Beyond discovering new drug candidates, Bristol Myers Squibb is also using AI to shorten the time required to prepare medicines for clinical trials.
Company executives estimate AI has already reduced parts of this process by approximately 20% to 30%, with ambitions of reaching reductions of up to 50% in the future. Executives even cited an experimental sickle cell disease treatment that they believe would not have been discovered without the assistance of their AI systems.
Nvidia BioNeMo Strengthens Research Capabilities
The new infrastructure will provide researchers with access to Nvidia’s BioNeMo Agent Toolkit, a collection of AI tools built specifically for biological research.
BioNeMo supports tasks such as:
- Protein structure prediction
- Molecular generation
- Molecular docking
- Genomics analysis
- Sequence analysis
The platform also enables multiple computational tools to work together within a single research workflow, allowing scientists to streamline complex biological analyses while continuing to validate AI-generated results through human review.
Connecting Global Research Teams
A key goal of the new AI infrastructure is to improve collaboration across Bristol Myers Squibb’s worldwide research organization.
The unified computing environment allows datasets, trained models, and experimental findings generated at one research site to be immediately available to scientists working elsewhere. Researchers will also be able to submit certain AI workloads using natural-language prompts instead of specialized computing commands, making advanced computing resources more accessible across the organization.
Greater Performance With Improved Efficiency
Despite delivering dramatically more computing power, the Vera Rubin-based infrastructure is also expected to improve energy efficiency.
According to Bristol Myers Squibb and Nvidia, the new eight-system cluster can deliver up to ten times more computing performance per megawatt than the infrastructure it replaces, helping reduce the growing energy demands associated with large-scale AI workloads.
Looking Ahead
Bristol Myers Squibb’s investment highlights how artificial intelligence is becoming an essential component of modern pharmaceutical research rather than simply an experimental technology.
By combining next-generation AI infrastructure with advanced biological modeling, predictive analytics, and global collaboration tools, the company aims to accelerate drug discovery, improve research efficiency, and bring promising therapies to patients more quickly. As AI models continue to evolve, investments in high-performance computing like Nvidia’s Vera Rubin platform are likely to become a defining advantage for pharmaceutical companies competing to develop the next generation of medicines.
