NAIRR Science Program Reshapes Scientific Research, Powered by NVIDIA AI Infrastructure
NVIDIA Blog · Zoe Kessler · 2026-06-22
The NSF's NAIRR pilot program has used NVIDIA AI infrastructure to support over 700 U.S. research projects, with featured outcomes including a publicly released fluid simulation foundation model, a molecular AI toolkit for energy materials discovery, and an infectious disease monitoring pipeline that cut report generation from several hours to two minutes.
Appears in
Extraction
Topics: nairrai-for-sciencehpcscientific-computingfoundation-models
Claims
- The NAIRR pilot program has driven research across more than 700 U.S. projects over two years, with NVIDIA contributing dedicated DGX node access and technical support.
- Polymathic AI used NVIDIA GPUs to build the 'Well' large-scale dataset and 'Walrus' foundation model for fluid-like physical simulations, releasing model weights, data, and code publicly.
- University of Michigan's MIST molecular foundation models, trained on a 40-GPU DGX cluster and 200,000 additional GPU hours, match or exceed state-of-the-art performance across 400+ structure-property benchmarks spanning electrochemistry, quantum chemistry, and physiology.
- Boston University's BEACON LLM pipeline reduced infectious disease outbreak report generation time from several hours to approximately two minutes for field experts.
Key quotes
When you talk to infectious disease experts about what they used to do before we developed this pipeline, it used to take several hours for them to compose a report. Now, producing a report gets done in roughly two minutes.