Bristol Myers Squibb Expands AI Supercomputer for Drug Discovery

Bristol Myers Squibb (BMS) is significantly expanding its artificial intelligence capabilities with the deployment of a second NVIDIA DGX SuperPOD, strengthening its efforts to accelerate drug discovery and advance the development of next-generation medicines.

The pharmaceutical company announced that the new system, built on eight NVIDIA DGX Vera Rubin NVL72 systems, will become one of the most powerful and energy-efficient AI computing platforms in the life sciences industry. The investment builds on BMS’ existing AI infrastructure and reflects the company’s strategy of integrating artificial intelligence across every stage of research and development rather than limiting its use to specialized teams.

According to BMS, the new AI cluster will provide researchers across its global organization with access to a unified computing platform capable of running advanced biological models, predicting drug behavior, training foundation models, and supporting AI-driven research workflows.

Erin Davis, Vice President of Research Business Insights and Technology at Bristol Myers Squibb, said the company’s goal is to democratize access to high-performance computing.

“Instead of equipping a small group of researchers with access to the supercomputer, we’re opening it up to literally every scientist,” Davis said. “No one has to wait, and no one is told they have a limit.”

The new infrastructure combines NVIDIA Vera central processing units with Rubin graphics processing units and is designed to deliver up to 10 times more performance per megawatt than the computing systems it replaces. It will also support the NVIDIA BioNeMo Agent Toolkit, enabling scientists to perform biological AI modeling and automate parts of the drug discovery process.

BMS has already been using AI extensively during the past three years through its first DGX SuperPOD, with the company reporting measurable improvements in research productivity.

Artificial intelligence has been used to accelerate target identification, reducing tasks that previously required weeks of manual analysis. The company has also applied AI to expand its library of CELMoD compounds, a class of molecules designed to selectively degrade disease-causing proteins, particularly in blood cancers.

In addition, researchers are using AI during lead optimization through what BMS calls its “Predict First” strategy.

Payal Sheth, Senior Vice President of Therapeutic Discovery Sciences at BMS, explained that AI predictions help scientists prioritize which molecules should be synthesized and tested in the laboratory.

“We use predictions as a way to prioritize synthesis of molecules with multi-parameter optimization,” Sheth said. “This helps eliminate molecules that are unlikely to meet our desired properties and allows laboratory resources to focus on candidates with the highest probability of success.”

The increasing use of AI has dramatically increased computing requirements within the company’s research organization.

“We’re saturated,” Davis said. “We’re already running very large-scale prediction models for complex molecules, and we’re building our own foundation models. That requires a significant amount of GPU computing power.”

The new AI infrastructure will be combined with BMS’ existing DGX SuperPOD to create a single computing environment accessible from research sites worldwide. According to the company, scientists will be able to launch complex computational analyses using natural language rather than requiring advanced programming or computational expertise.

The unified platform is also intended to improve knowledge sharing across global research teams.

Sheth said data generated by one research group can now be used to strengthen AI models that support scientists working on entirely different drug programs.

“The compute infrastructure connects all of our scientists together and ensures our learnings become institutional knowledge,” she said. “Every experiment contributes to future discoveries instead of remaining isolated within individual projects.”

BMS is also exploring the use of AI agents that can work across research programs, analyze data, and support scientific decision-making.

According to Davis, these agentic workflows could eliminate traditional organizational silos by allowing AI systems to draw insights from multiple disease areas simultaneously.

“When scientists have access to well-trained virtual scientific assistants with BMS knowledge built in, they effectively become an entire research team,” she said.

Despite the rapid advances in AI, company leaders stressed that artificial intelligence is intended to complement rather than replace human expertise.

Sheth noted that scientific judgment remains essential, with AI serving to enhance decision-making through more comprehensive data analysis and predictive insights.

Davis echoed that view, emphasizing that researchers will continue to guide the scientific process while AI expands what individual scientists can accomplish.

The company has already planned applications for the new computing platform across small- and large-molecule drug discovery, clinical development, digital twin technologies, and other research areas.

“We didn’t build this simply to have the biggest computing platform,” Davis said. “This system is designed to support every stage of the drug discovery process.”

As pharmaceutical companies increasingly adopt artificial intelligence to accelerate research, Bristol Myers Squibb’s latest investment highlights the growing role of advanced computing in transforming how new medicines are discovered, optimized, and developed for patients worldwide.