Roche is betting big on artificial intelligence. The Swiss pharmaceutical company just announced a major expansion of its AI computing power, adding 2,176 high-performance GPUs across facilities in the United States and Europe. This brings Roche’s total GPU count to more than 3,500 NVIDIA Blackwell chips – the largest announced setup of any pharmaceutical company.
The company calls this its “AI factory” – a supercomputing platform designed to speed up everything from drug discovery to manufacturing. Roche has been working with NVIDIA since 2023, but this expansion marks a significant escalation of that partnership.
“In healthcare, time is the most critical variable; every day saved means a life-changing medicine or diagnostic reaches a patient sooner,” said Wafaa Mamilli, Roche’s Chief Digital and Technology Officer.
How does it work?
Roche’s AI factory touches nearly every part of the company’s operations:
- Drug discovery: Scientists use NVIDIA’s BioNeMo platform to connect biological and chemistry experiments with AI models, testing hypotheses at scale
- Manufacturing: Digital twins of production lines help engineers optimize factory processes using NVIDIA Omniverse libraries
- Diagnostics: NVIDIA Parabricks software analyzes vast datasets to find patterns
- Digital pathology: AI scans large numbers of medical images to detect subtle disease patterns
- Digital health: NVIDIA NeMo Guardrails ensures conversational AI meets healthcare safety standards
The centerpiece is Roche’s “Lab-in-the-Loop” strategy, where AI models work alongside traditional laboratory experiments. This approach lets scientists build more sophisticated predictive models and potentially discover treatments that might otherwise be missed.
Why does it matter?
Drug development is notoriously slow and expensive. Getting a new medicine from lab to patient typically takes over a decade and costs billions of dollars. AI could change that equation by:
- Identifying promising drug candidates faster
- Predicting which compounds are most likely to succeed in clinical trials
- Optimizing manufacturing processes before building physical facilities
- Finding disease patterns in medical data that humans might miss
Aviv Regev, who heads research at Roche’s Genentech unit, says the computing power will help “build more sophisticated predictive frontier models and further shorten the path from biological insight to life-saving medicine.”
The scale matters too. With 3,500 GPUs, Roche can run complex AI models that smaller companies simply can’t afford to operate.
The context
Roche isn’t alone in this AI arms race. Major pharmaceutical companies are all investing heavily in artificial intelligence, but few have announced GPU deployments of this size. The company has been working on AI integration for over five years, giving it a head start over competitors just getting started.
The timing is significant. NVIDIA’s Blackwell GPUs are among the most powerful AI chips available today, and securing thousands of them represents a major capital investment. GPU supply has been tight across industries as companies rush to build AI capabilities.
Founded in 1896, Roche has grown into the world’s largest biotechnology company. The Swiss firm operates globally and owns Genentech in the United States. The company says sustainability has been part of its business for over 125 years and it aims to reach net zero emissions by 2045.