U.S. and Nvidia Launch $152M Open AI Science Push
The NSF, Ai2, and Nvidia are teaming up to build fully open AI models aimed at accelerating discovery and securing America’s leadership in science
The United States is teaming up with Nvidia to develop a new generation of AI models aimed at accelerating scientific breakthroughs and strengthening the country’s leadership in AI research. The National Science Foundation (NSF) will invest $75 million in the Open Multimodal AI Infrastructure to Accelerate Science (OMAI) project, which is led by nonprofit research lab Ai2, while Nvidia is contributing $77 million. The initiative will produce a fully open suite of advanced AI models designed specifically for science applications.
“Bringing AI into scientific research has been a game changer,” Brad Stone, NSF’s chief of staff, who is performing the duties of the agency’s director, says in a statement. “NSF is proud to partner with Nvidia to equip America’s scientists with the tools to accelerate breakthroughs. These investments are not just about enabling innovation; they are about securing U.S. global leadership in science and technology and tackling challenges once thought impossible.”
This collaboration is a part of the White House’s AI Action Plan (PDF), which was introduced in July. It features three pillars of the nation’s technology approach: accelerating innovation, building American infrastructure, and leading international diplomacy and security. Encouraging open-source and open-weight AI falls under the first category, with the Trump administration writing:
“We need to ensure America has leading open models founded on American values. Open-source and open-weight models could become global standards in some areas of business and in academic research worldwide. For that reason, they also have geostrategic value. While the decision of whether and how to release an open or closed model is fundamentally up to the developer, the Federal government should create a supportive environment for open models.”
Under the terms of the partnership, Ai2 will be developing open-source, multimodal large language models that are trained on scientific data and literature. In turn, American researchers will use the LLMs to process and analyze research data faster, generate code and visualizations, and connect their insights to past discoveries.
Cirrascale Cloud Services, the company that the research institute started partnering with in July, will provide managed services for the hardware infrastructure needed to support this effort. Ai2 will also be using Supermicro’s platform.
“AI is the engine of modern science—and large, open models for America’s researchers will ignite the next industrial revolution,” Nvidia’s CEO Jensen Huang declares. “In collaboration with NSF and Ai2, we’re accelerating innovation with state-of-the-art infrastructure that empowers U.S. scientists to generate limitless intelligence, making it America’s most powerful and renewable resource.”
Open-source AI is nothing new for Ai2. The institute has established itself as a leader in the field, releasing LLMs such as MolmoAct, OLMo, Molmo, and Tulu3. It has called out companies for branding models as “open” while withholding data. And advancing science is one of its core missions under CEO Ali Farhadi. Ai2 has launched several efforts to support this cause, such as developing tools to help researchers answer complex questions across multiple papers, establishing a new benchmark for science AI models, and supporting AI-driven cancer research.
“Fully-open AI is not just a preference—it’s a necessity,” Farhadi is quoted as saying. “For the U.S. to continue leading the next era of scientific and technological discovery, we must create open, collaborative ecosystems where millions of researchers and developers can work together to improve and expand these systems. This infusion will supercharge the work we do at Ai2 and increase America’s ability to deliver breakthrough AI developments.”
While no specific timeline for delivery is stated, its initial use cases are. It’s believed that Ai2’s work goes towards “discovering new materials, improving protein function prediction for biomedical advancements, and addressing core weaknesses in today’s large language models.”