Bristol Myers Squibb has announced the deployment of its next-generation artificial intelligence infrastructure, marking a significant scaling of its computational capabilities for drug discovery. The pharmaceutical giant is integrating a second NVIDIA DGX SuperPOD, constructed from eight DGX Vera Rubin NVL72 systems, into its existing research ecosystem. This deployment, internally termed the SuperDuperPOD, delivers ten times the performance per megawatt of its predecessor and establishes the most powerful AI cluster currently utilized in the life sciences sector. The primary objective of this expansion is the democratization of high-performance computing across the organization. Erin Davis, vice president of research business insights and technology at BMS, emphasized that the unified platform removes traditional access barriers, enabling every scientist to initiate complex predictive workflows without wait times or resource limits. The infrastructure will be managed through NVIDIA Mission Control and leverages the NVIDIA BioNeMo Agent Toolkit to facilitate biological AI, model training, and agentic workflows across the entire drug development lifecycle. BMS has operated AI-enabled infrastructure for approximately three years, yielding measurable efficiencies. The company currently utilizes machine learning to accelerate target identification, saving researchers weeks of manual labor, and to optimize its CELMoD compound library, which targets the selective degradation of cancer-causing proteins. A methodology termed Predict First now guides lead optimization, allowing computational models to predict molecular behavior and prioritize synthesis. This approach filters out low-probability candidates early, aligning laboratory experiments with molecules most likely to succeed in clinical trials. Payal Sheth, senior vice president of therapeutic discovery sciences, noted that the new architecture transforms isolated project data into a compounding intelligence framework. By establishing a single data plane accessible from all global sites, datasets generated in one location immediately inform models used in another. This closed-loop system ensures that experimental results, clinical readouts, and cross-functional partnerships continuously refine predictive accuracy. The expansion is closely tied to BMS strategic focus on brain health, an area Davis cited as personally motivated by her family experience with Alzheimer disease. The infrastructure is explicitly allocated to support symptom remediation research, alongside small and large molecule design, digital twins, and clinical applications. Rather than treating AI as a supplementary tool, BMS is positioning it as a foundational research driver. Davis described the new environment as a force multiplier for human expertise, where agentic systems handle routine computational heavy lifting while scientists focus on high-value scientific decision-making and experimental design. Internal assessments indicate that BMS current production-level predictions and foundational model development have already saturated previous compute capacities. With the Vera Rubin deployment, the company has secured ample resources to scale these operations. Leadership maintains that the hardware is merely an enabler for a broader cultural shift toward data-driven discovery. As the infrastructure transitions to full production, BMS expects the integrated platform to accelerate therapeutic timelines, reduce experimental attrition, and institutionalize a continuous learning loop that fundamentally reshapes how medicines are designed and validated.