On April 29, 2025, the White House Office of Science and Technology Policy (OSTP) issued a Request for Information on the Development of a 2025 National Artificial Intelligence (AI) Research and Development (R&D) Strategic Plan. In this submission, experts at the Center for a New American Security (CNAS) outline recommendations for OSTP in response to this request.

Summary

Cutting-edge R&D is a pillar of America’s AI leadership. Private capital is essential to driving AI R&D progress, with U.S. firms investing an estimated $109.1 billion in 2024—more than twelve times the estimated $9.3 billion from Chinese firms that same year.

Although private capital is essential to maintaining America’s edge in AI R&D, it is insufficient. Certain categories of research will remain underfunded when guided solely by commercial incentives. In these areas, targeted federal investment plays a vital and complementary role. As the Trump administration develops its 2025 National AI R&D Strategic Plan, it should prioritize efforts that:

• Inform U.S. policy development;
• Strengthen U.S. policy implementation;
• Lack commercial incentives for prioritization or publication;
• Support foundational AI research.

Furthermore, federal agencies should expand, initiate, or sustain investment to strengthen U.S. leadership in AI R&D in the below areas:

Robustness and Reliability for National Security

Federal agencies should fund high-risk, high-reward projects to drive fundamental advances in AI robustness, reliability, interpretability, and explainability. Frontier AI systems offer enormous potential for analyzing data at speed and scale across operations. The lack of interpretability creates significant operational risks, and hinders broad-based adoption in the public sector.

AI Standards and Benchmarks

As AI systems advance and proliferate, establishing standards and benchmarks becomes critical for evaluating capabilities, ensuring interoperability, and building trust. Countries that lead in technical and policy frameworks will have stronger influence over international standards, conferring significant economic and political benefits.

The United States must lead in developing comprehensive evaluation frameworks for highly capable AI systems that keep pace with technological progress. Federal R&D can provide the institutional neutrality, continuity, and classification access needed for these efforts. Such foundational work requires long-term investment with uncertain commercial payoffs, making it well-suited for federal R&D priorities.

Support and Expand the National Artificial Intelligence Research Resource (NAIRR)

As the lead federal agency for computer and information science research, the U.S. National Science Foundation (NSF) should support the AI research community. The NSF can help by providing AI researchers with the infrastructure necessary to conduct cutting-edge AI R&D.

NSF should continue to support and expand the National Artificial Intelligence Research Resource (NAIRR) Pilot in the Strategic Plan. While the compute resources available through NAIRR represent only a fraction of what leading industry players can access, this initiative serves as a crucial first step in democratizing compute access for research institutions across America. By providing research infrastructure to universities, smaller companies, and independent researchers, federal investment can preserve the competitive innovation ecosystem that has long been America’s strategic advantage.

Anticipating Rapid AI Progress

The federal government should invest in research that anticipates continued rapid progress in AI capabilities. Recent years have demonstrated a consistent pattern: AI models have repeatedly performed tasks once predicted as impossible. Although this pace of progress is not guaranteed going forward, research investments that anticipate continued progress are particularly valuable because they deliver the highest impacts in precisely the scenarios where AI’s impacts prove most transformative.

AI-Biotechnology Nexus

Harnessing the convergence of AI and biotechnology is both a transformative opportunity and national security imperative. AI has demonstrated remarkable success in accelerating drug discovery, predicting protein structures, identifying novel drug targets, and optimizing clinical trial design—capabilities with both commercial and strategic significance. AI-enhanced biomanufacturing processes promise to revolutionize production of critical goods, from pharmaceuticals to sustainable fuels. To harness AI’s potential to accelerate biotechnology breakthroughs, the government must expand investment in AI-biotechnology research and infrastructure.