NVIDIA's BioNeMo platform has secured formal co-innovation partnerships with Eli Lilly and Thermo Fisher, embedding AI foundation models directly into enterprise-scale drug discovery pipelines. The deals mark a structural shift: AI infrastructure in pharma is no longer experimental — it is operational and generating lock-in.
Novo Nordisk accelerated this pattern by licensing its cell therapy program to Cellular Intelligence, an AI-native spinout, while shutting its internal cell therapy unit.1 The company doubles down on GLP-1, its core commercial franchise. Cellular Intelligence advances a Parkinson's disease program using AI-driven development methods.1
The strategic logic is direct. Large pharma is offloading high-risk, capital-intensive programs to AI-native partners built to absorb early-stage uncertainty faster and cheaper than internal units can manage.
BioNeMo provides foundation models for protein structure prediction, molecular docking, and generative chemistry. Once these models are embedded in Lilly's or Thermo Fisher's pipelines, switching costs climb. Infrastructure lock-in follows adoption.
A wave of commercial foundation model platform launches in 2025 and 2026 has converted AI adoption from a competitive differentiator into a baseline requirement. Drug discovery programs without production AI pipelines are falling behind those that have them.
For biotech startups, enterprise-grade AI infrastructure is becoming accessible through partnership arrangements — lowering barriers to sophisticated discovery capabilities. The tradeoff is vendor concentration risk. Dependence on NVIDIA's stack ties discovery capacity to a single infrastructure provider.
Novo Nordisk's decision signals that even top-tier pharma has concluded internal AI-native cell therapy units aren't cost-effective when external partners can move faster.1 The model — spin out, license, partner — is spreading across the industry.
AI infrastructure platforms are becoming the organizing layer of pharmaceutical R&D. Fourteen backing claims tracked across this signal cluster confirm the trend is broad, not isolated. Companies that locked in early are setting the terms for how drug discovery operates at scale.