NVIDIA, NSF invest $150M in open AI to turbocharge US science

NVIDIA and the U.S. National Science Foundation (NSF) have jointly committed over $150 million to develop open, multimodal AI models aimed at accelerating scientific discovery and maintaining U.S. leadership in AI-driven research. This partnership supports the Open Multimodal AI Infrastructure to Accelerate Science (OMAI) project, led by the Allen Institute for AI (Ai2). The NSF is contributing $75 million, while NVIDIA provides $77 million in advanced technology, including NVIDIA HGX B300 systems with Blackwell Ultra GPUs and the NVIDIA AI Enterprise software platform. These resources are designed to handle large-scale AI workloads, enabling faster model training and inference. OMAI will produce a fully open suite of large language models capable of processing diverse scientific data types such as text, images, graphs, and tables. These models will help researchers analyze data more rapidly, generate code and visualizations, and link new insights to existing knowledge, with applications ranging from material discovery to protein function prediction. All models,