Bristol Myers Squibb

Bristol Myers Squibb's Devens, Massachusetts facility—one of the company's most advanced biologics and cell therapy manufacturing sites—faced a fundamental challenge inherent to living-cell-based production: biological variability. Unlike small-molecule pharmaceuticals, biologics depend on living organisms whose behavior is difficult to standardize. This variability slowed New Product Introduction cycles and constrained throughput, creating a competitive bottleneck at a site expected to deliver complex therapies at scale. Without better tools to predict and control biological process performance, the site struggled to ramp new products quickly while maintaining quality and reducing its environmental footprint.

Devens pursued a deliberate integration of biopharma process science with AI and digital manufacturing strategies, ultimately developing more than 30 distinct use cases across the site. The core technology was predictive machine learning applied directly to biological process data—fermentation parameters, cell culture conditions, upstream and downstream variables—to anticipate deviations before they affected yield or quality. Rather than deploying a single platform, the team built a portfolio of ML models embedded into existing manufacturing workflows, enabling process engineers to act on real-time predictions. This science-first approach, grounded in domain expertise from BMS's own biopharma teams, distinguished it from generic digitalization efforts and allowed solutions to be validated against known biological mechanisms.

The program delivered simultaneous improvements across speed, volume, and sustainability—a rare combination in biologics manufacturing:

• 42% reduction in New Product Introduction (NPI) time, compressing the cycle from development to commercial-scale production
• 40%+ increase in manufacturing volume at the Devens site, expanding capacity without proportional capital investment
• 40%+ reduction in emissions, reflecting efficiency gains in energy and resource consumption

The breadth of impact—spanning operational speed, output, and environmental performance together—demonstrates that the AI use cases were integrated at a systems level rather than applied to isolated process steps. The site was subsequently recognized by the World Economic Forum's Global Lighthouse Network in January 2026.

• Domain expertise must lead AI deployment: BMS paired process scientists with data teams, ensuring models reflected actual biological mechanisms rather than statistical artifacts.
• Portfolio beats single-platform thinking: Building 30+ targeted use cases created compounding impact across NPI, throughput, and sustainability simultaneously.
• Biologics variability is manageable with predictive ML: Living-cell processes are not inherently incompatible with AI—they require models trained on the right process signals.
• Sustainability gains follow efficiency gains: Emissions reductions emerged as a byproduct of optimized resource use, not a separate initiative.
• Lighthouse recognition signals a replicable model: WEF validation provides a benchmark other pharma manufacturers can study and adapt.