Roche announced plans to build a large-scale internal AI laboratory and struck a strategic deal with Nvidia to provision the computing infrastructure that will underpin the project. The collaboration follows recent supercomputing investments by other drugmakers and positions Roche to accelerate computational chemistry, target discovery and diagnostics development using high-performance GPUs. Roche’s initiative aims to centralize large-scale model training, simulation and inference for drug design and biomarker discovery. The company said the factory will speed target prioritization and diagnostic development while enabling in-house scale for AI-driven programs. Nvidia will supply hardware and software optimizations to support model-intensive workloads. The move signals increasing competition among pharmas to own end-to-end AI stacks and could shift where partners and startups choose to host computation or seek collaborations. Firms should expect rising demand for cloud-managed GPU capacity and tools that integrate wet-lab data with large-model training pipelines.