Bristol Myers Squibb Partners with Nvidia AI Technology to Accelerate Drug Discovery Process

Bristol Myers Squibb (BMS) has announced the acquisition of an Nvidia DGX SuperPOD built on the chipmaker's cutting-edge Vera Rubin architecture, marking a significant advancement in AI-powered pharmaceutical research and development. This strategic investment positions BMS as the first life sciences organization to deploy a DGX SuperPOD based on Vera Rubin technology, which Nvidia unveiled earlier this year as the next evolution of its AI computing systems.

🖥️ Expanding AI Computing Infrastructure

The newly acquired cluster comprises eight DGX Vera Rubin NVL72 systems, with each rack-scale unit integrating Nvidia Vera central processing units (CPUs) and Rubin graphics processing units (GPUs). This powerful infrastructure will enable BMS to:

• Train proprietary AI models for drug discovery
• Execute complex predictions across research programs
• Process scientific data involving compounds, proteins, and biomolecules

While financial details remain undisclosed, this acquisition represents a substantial upgrade from BMS's existing Nvidia infrastructure. Company executives characterized their current SuperPOD as two to three generations behind the Vera Rubin architecture, highlighting the technological leap this investment represents.

The existing DGX SuperPOD, operational for approximately three years, will be integrated with the Vera Rubin system to create a unified computing environment accessible from BMS research facilities worldwide.

📈 Addressing Growing Computational Demands

Greg Meyers, BMS's Chief Digital and Technology Officer, emphasized that computing requirements have surged as the pharmaceutical giant deploys increasingly sophisticated AI models throughout its research organization. Erin Davis, Vice President of Research Business Insights and Technology at BMS, noted that the current infrastructure is operating at full capacity, driven by:

• Large-scale molecular predictions involving complex biological molecules
• Development of internal foundation models
• Expanded access requirements across research teams

Davis emphasized that the new system will democratize access to computational resources, stating: "The new system will not be limited to a small group of computational researchers." BMS plans to eliminate the waiting periods and access restrictions currently constraining researchers, enabling broader organizational utilization.

🔬 AI-Driven Drug Discovery Applications

BMS has integrated artificial intelligence into virtually all aspects of its drug development pipeline. According to the company, AI informs the design of every small-molecule program and the majority of large-molecule initiatives. Key applications include:

Robert Plenge, BMS's Chief Research Officer, highlighted the transformative impact: "Maybe before we could evaluate 10 candidates and now we can assess dozens during early development stages."

🎯 The "Predict First" Methodology

BMS has pioneered a computational screening approach called "Predict First," which leverages AI-generated predictions to filter out molecules lacking required properties before synthesis selection. This methodology enables:

• Pre-synthesis evaluation of thousands of potential compounds
• Identification of molecules with optimal multi-parameter combinations
• Focused laboratory resources on high-probability candidates

Payal Sheth, Senior Vice President of Therapeutic Discovery Sciences, explained: "We use predictions as a way to prioritize synthesis of molecules with multi-parameter optimization. This ensures precious laboratory experiments are aligned with progressing molecules that have the highest probability of success."

💊 Real-World Impact: CELMoD Compounds and Clinical Applications

BMS has successfully applied AI to expand its library of CELMoD compounds—engineered molecules designed to selectively degrade cancer-causing proteins. The company is currently investigating these compounds for:

• Blood cancers (hematologic malignancies)
• Additional oncological applications
• Other disease areas under investigation

The modeling work enabled researchers to examine additional protein targets and potential compounds before committing to experimental validation, significantly accelerating the discovery process.

BMS has reduced the time required to produce medicines for clinical trials by 20-30%, with projections suggesting potential reductions of 50% in coming years.

Plenge cited an experimental sickle cell disease treatment currently in early clinical development as a prime example, noting that this therapeutic candidate "probably would not have been discovered without the company's AI tools."

Note: These timeframe reductions refer to candidate identification and production for clinical testing, not subsequent trial performance.

🧰 Advanced Toolkits and Capabilities

The Vera Rubin system provides BMS researchers with access to Nvidia's BioNeMo Agent Toolkit, a comprehensive platform for biological and drug-discovery applications. BioNeMo offers:

Despite these advanced capabilities, BMS executives emphasized that human researchers will continue to review model outputs and make final decisions regarding which compounds or programs should advance through the development pipeline.

🌐 Global Research Connectivity and Data Integration

BMS is implementing tools designed to reduce the specialist knowledge required to initiate complex computing tasks. Researchers will be able to launch certain prediction requests using natural-language instructions, democratizing access to advanced computational resources.

The unified infrastructure will be managed through Nvidia Mission Control, which provides:

• Cluster provisioning and configuration
• Infrastructure monitoring and diagnostics
• Workload management and scheduling

This integrated environment enables seamless data sharing across global research sites. For example, datasets generated from programs at BMS's Lawrenceville, New Jersey facility can be immediately incorporated into models utilized by research teams in San Diego, California.

Sheth explained: "The compute infrastructure is what connects all of our scientists together and ensures that our learnings are institutionalized."

The two SuperPODs will operate through a common data environment, providing teams at different locations access to:

• Shared datasets from ongoing research
• Model outputs and predictions
• Experimental results and findings
• Clinical readout data
• Information from research partnerships and collaborations

⚡ Energy Efficiency and Sustainability

A significant advantage of the Vera Rubin architecture is its superior energy efficiency. Meyers highlighted the economic and environmental benefits:

Meyers emphasized the practical implications: "When you host these things, you have to pay an electric bill. Think of it as 10 times more compute capacity per watt spent… Electricity is not getting cheaper."

This efficiency gain translates to both reduced operational costs and a smaller environmental footprint, aligning with corporate sustainability objectives while delivering enhanced computational capabilities.

🎯 Strategic Resource Allocation

BMS plans to allocate the new computing capacity across multiple research domains:

• Small-molecule design and optimization
• Large-molecule design including biologics and complex proteins
• Clinical research applications and trial support
• Digital-twin applications for predictive modeling

While BMS did not provide specific details about the planned digital-twin initiatives or the precise capacity allocation for each research area, the diverse application portfolio demonstrates the versatility and strategic importance of this infrastructure investment.

🚀 Looking Forward

While BMS has not disclosed a specific deployment timeline or hosting location for the new Vera Rubin system, this acquisition represents a major milestone in pharmaceutical AI adoption. As the first life sciences organization to implement this next-generation architecture, BMS is positioning itself at the forefront of AI-driven drug discovery.

The integration of advanced computational infrastructure with democratized access, global connectivity, and energy-efficient performance creates a comprehensive platform for accelerated pharmaceutical innovation. This investment underscores the pharmaceutical industry's recognition that AI and high-performance computing are no longer optional tools but essential capabilities for competitive drug development in the modern era.

Photo by Chidera Faustina Okeke