NAIRR pilot accelerates scientific AI research with NVIDIA DGX

NVIDIA announced its contribution to the NAIRR pilot, providing dedicated NVIDIA DGX access and technical support to researchers.

• Main announcement: NVIDIA provided researchers in the NAIRR pilot with a cloud-based resource guaranteeing a minimum of four NVIDIA DGX nodes for at least a month plus technical onboarding and support; the NAIRR pilot has supported over 700 projects across the past two years. The University of Michigan team used a 40-GPU NVIDIA DGX cluster (from a NAIRR allocation) and an additional 200,000 NVIDIA GPU hours on ALCF’s Polaris for MIST model development.
• Background and project details: NAIRR-enabled projects highlighted include Polymathic AI’s Walrus foundation model (dataset, code and pretrained weights made public) for fluidlike simulations, University of Michigan’s MIST molecular models for energy storage fused with LLMs, and Boston University’s BEACON LLM pipeline for infectious disease detection (processing HealthMap, news, social media and other signals). The announcement summarizes implementation over the past two years and documents concrete compute allocations and public releases.