NVIDIA's support of the NSF's NAIRR pilot program, which has enabled over 700 research projects, has accelerated scienti

For the past two years, the U.S. National Science Foundation's National Artificial Intelligence Research Resource pilot program has supported over 700 innovative research projects spanning areas from protein prediction to infectious disease outbreak management. NVIDIA contributed to this initiative by providing cloud-based resources offering researchers dedicated access to a minimum of four NVIDIA DGX nodes for at least one month, along with technical support for onboarding and project assistance. With NVIDIA's infrastructure support and DGX reference architecture, researchers have accelerated their workflows and developed technologies that are advancing industries including healthcare, agriculture and energy.

Three prominent NAIRR projects exemplify this progress. Polymathic AI, a coalition of scientists from Flatiron Institute, Cambridge University and Lawrence Berkeley National Lab, used NVIDIA GPUs and NVLink technology to develop Walrus, a large-scale foundation model for fluidlike behavior. The publicly available Walrus model, along with its supporting dataset and code, represents the largest and most broadly applicable foundation model for fluid dynamics to date and builds on previous physics pretraining work.

Researchers at the University of Michigan, led by Professor Venkat Viswanathan, created MIST (Molecular Insight SMILES Transformers), a family of molecular foundation models designed for exploration of chemical space. Developed on a 40-GPU NVIDIA DGX cluster and utilizing 200,000 NVIDIA GPU hours on the Polaris cluster, MIST combines domain-specific molecular AI with general-purpose large language models to help scientists explore chemistry in natural language and identify promising materials for energy storage and conversion technologies. MIST has been fine-tuned on over 400 structure-property relationships and can match or exceed state-of-the-art performance across multiple scientific domains.

Boston University's Hariri Institute developed BEACON (Biothreats Emergence, Analysis and Communications Network), an outbreak monitoring program using NVIDIA accelerated compute to train a large language model on infectious disease documentation. The model analyzes global online posts and disease signals from sources including HealthMap, news, social media and expert communications to generate comprehensive outbreak reports. According to Ioannis Paschalidis, director of Boston University's Hariri Institute, the model has reduced report composition time from several hours to approximately two minutes, enabling internationally deployed doctors, government organizations and academic researchers to quickly identify and treat emerging infectious diseases.