NVIDIA expands BioNeMo platform as pharma groups commit up to $1bn to AI-driven drug discovery

NVIDIA said its BioNeMo platform is enabling so-called lab-in-the-loop workflows that combine data generation, artificial intelligence training and model deployment across biology and drug discovery, as the industry grapples with research and development costs estimated at $300 billion a year.

The company said BioNeMo is designed to allow scientific data to be generated and processed continuously, with models trained, optimised and deployed in close coordination with physical laboratory work. The approach is intended to shorten development cycles and improve the productivity of research teams working on complex biological problems.

NVIDIA said the platform has been expanded with new Clara open models, including RNAPro, which focuses on RNA structure prediction, and ReaSyn v2, aimed at practical chemical synthesis. The company has also added BioNeMo Recipes, a set of workflows for model training and deployment, alongside new data libraries such as nvMolKit, a GPU-accelerated cheminformatics toolkit.

Pharmaceutical major Eli Lilly and Company announced a co-innovation lab with NVIDIA to apply BioNeMo, agentic AI and robotics across drug discovery and wider operations. The collaboration builds on Lilly’s existing NVIDIA DGX SuperPOD and its internal AI factory, which the company uses to support large-scale model training.

Diogo Rau, executive vice president and chief information and digital officer at Lilly, said the initiative would accelerate the company’s scientific capabilities. “We see this as a catalyst for the capabilities that will define the next era of drug discovery,” he said. NVIDIA and Lilly said the programme represents a total investment of up to $1 billion over five years, covering talent, infrastructure and compute.

Laboratory equipment group Thermo Fisher Scientific said it will integrate NVIDIA’s full-stack AI computing with its laboratory instruments to create scalable, automated data factories. The integration will use DGX Spark desktop supercomputers to support edge-to-cloud workflows, NVIDIA NeMo software to orchestrate multi-agent laboratory operations and BioNeMo tools for autonomous data analysis.

Gianluca Pettitti, executive vice president at Thermo Fisher Scientific, said the combination of AI and automation would reshape scientific practice. “Artificial intelligence coupled with laboratory automation will transform how scientific work is performed,” he said.

Beyond large pharmaceutical and tools companies, NVIDIA said a growing ecosystem of model builders and AI-native life sciences firms is adopting BioNeMo. These include Basecamp Research, Boltz PBC, Chai Discovery and Natera, which are using the platform to scale biological model development.

NVIDIA added that robotics and laboratory automation companies are also deploying its simulation and physical AI tools to connect digital agents directly to physical laboratory systems. The aim is to close the loop between AI-driven hypothesis generation and real-world experimentation, reducing manual intervention and increasing throughput.

The expansion of BioNeMo highlights how AI infrastructure providers are moving deeper into life sciences, positioning compute, models and automation as core enablers of the next phase of drug discovery and biological research.

The Recap

• NVIDIA expanded BioNeMo to integrate lab-in-the-loop workflows for drug discovery.
• Companies plan up to $1 billion investment over five years.
• Thermo Fisher and NVIDIA will build autonomous lab infrastructure.