The National Science Foundation (NSF) and NVIDIA are partnering to invest $152 million in the Open Multimodal AI Infrastructure to Accelerate Science (OMAI), a project aimed at advancing open-source AI for scientific research.

Led by the Allen Institute for AI (Ai2), the initiative will develop open-source, multimodal large language models trained on scientific data to support breakthroughs in fields like materials science, biology, and energy.

The AI tools created through OMAI will help streamline research processes, generate code and visualizations, and connect new insights with existing knowledge, with initial applications in discovering new materials and improving protein function prediction.

The project emphasizes workforce development by training AI talent beyond traditional tech hubs, involving universities such as the University of Washington, University of Hawaii at Hilo, University of New Hampshire, and University of New Mexico.

The first fully open AI model from this initiative is expected to launch in about 18 months, providing broader access to institutions outside Silicon Valley, including the participating universities.

White House OSTP Director Michael Kratsios highlighted that this partnership demonstrates the U.S. commitment to fostering open AI models that set global standards for research and industry.

Supporters like NSF director Brian Stone, NVIDIA CEO Jensen Huang, and Kratsios underscore the project's importance at both national and technological levels.

Ai2 CEO Ali Farhadi stressed that fully open AI ecosystems are crucial for maintaining U.S. leadership in science and technology, with initial applications aimed at material discovery, biomedical research, and enhancing language models.

The initiative aligns with the U.S. government's AI Action Plan, emphasizing open-source development, transparency, and workforce participation to secure global leadership and counter China's AI ambitions.

A key goal of OMAI is to democratize AI research, ensuring that breakthroughs are accessible beyond well-funded institutions, fostering innovation across diverse academic centers.

OMAI will promote transparency by providing access to training data, source code, model weights, and documentation, contrasting with current black-box AI models like ChatGPT.

NVIDIA's CEO Jensen Huang highlighted that AI models will drive the next industrial revolution and that this infrastructure will empower U.S. scientists to innovate.

Nvidia will supply advanced hardware, including Blackwell Ultra GPUs and HGX B300 systems, to support the computational demands of multimodal AI across disciplines.