Federal AI R&D under the National AI Initiative

What the National AI Initiative does

Congress created the National Artificial Intelligence Initiative in Division E of the FY2021 National Defense Authorization Act to coordinate AI research, development, demonstration, standards, and education across the federal government and with industry and academia¹. The National AI Initiative Office (NAIIO) in the Office of Science and Technology Policy (OSTP) leads cross-agency coordination, public communication, and tracking of progress under this statutory framework². The government’s R&D priorities are articulated in the National AI R&D Strategic Plan (2023 update), issued by the National Science and Technology Council and the Networking and Information Technology R&D (NITRD) program³.

Strategy signal: priorities for federal AI R&D

The 2023 Strategic Plan sets the agenda for federal AI R&D with focus areas including long-term, high-risk research; advancing trustworthy and responsible AI; developing shared testbeds, data, and evaluation resources; enhancing safety and security; building a diverse AI workforce; and fostering public-private partnerships and international collaboration³. The plan emphasizes infrastructure for experimentation and reproducibility, explicitly calling for testbeds, shared datasets, benchmarks, and evaluation methodologies to underpin rigorous AI science³.

Executive Order 14110 reinforces these priorities by directing agencies to accelerate safe, secure AI development and use, including launching a pilot of the National AI Research Resource (NAIRR) to expand researcher access to compute, data, and tools⁴. NIST’s AI Risk Management Framework (AI RMF 1.0) provides voluntary guidance to help organizations design, develop, and deploy trustworthy AI systems, complementing the R&D strategy’s focus on safety, security, and accountability⁵.

Where the money and programs are: notable federal AI R&D investments

• National AI Research Institutes. In 2023, NSF and partners announced 11 new National AI Research Institutes with $220 million in funding, expanding a nationwide network that advances AI across fields and connects academia, industry, and government for use-inspired research⁶. These institutes are designed to catalyze multidisciplinary AI advances while building regional capacity and workforce pipelines⁶.

• DARPA AI Next Campaign. DARPA launched the $2 billion AI Next campaign to drive the “third wave” of AI with research into contextual reasoning, robust learning, and new architectures, building on decades of defense-funded AI innovation and seeding multiple program lines⁷.

• NIH Bridge2AI. NIH’s Bridge2AI program commits approximately $130 million to create high-quality, ethically sourced biomedical datasets and community practices that enable more reliable application of AI in health research, with an emphasis on standards, team science, and responsible practices⁸.

• DOE Office of Science: AI for Science. DOE’s Office of Science identifies AI as a cross-cutting pillar for accelerating discovery on leadership compute systems, including exascale platforms, and outlines research priorities and opportunities for integrating AI with scientific workflows, modeling, and experimental facilities⁹.

• NAIRR pilot. OSTP and NSF launched the NAIRR pilot to provide U.S. researchers with a coordinated pathway to compute, data, models, and user support aligned to national R&D needs, advancing the Strategic Plan’s infrastructure goals and EO 14110 directives¹⁰⁴. The NAIRR Task Force’s final report details a broader vision for a sustained, national-scale shared research resource to democratize access and strengthen U.S. AI innovation capacity¹¹.

Governance and risk management shaping R&D execution

OMB Memorandum M-24-10 directs agencies to strengthen AI governance, designate Chief AI Officers, inventory AI use cases, and manage risks consistent with NIST’s AI RMF, which directly affects how R&D programs plan experiments, evaluate models, and document risk controls¹²⁵. The voluntary nature of the NIST AI RMF is explicit, but OMB’s policy direction links agency practice to this framework when planning and executing AI activities, including research prototypes tested on federal systems⁵¹².

NIST’s AI RMF also foregrounds testing, evaluation, verification, and validation, which the Strategic Plan calls out as core to federal support for benchmarks and testbeds—aligning R&D infrastructure with risk management expectations set by EO 14110 and OMB³⁵⁴¹².

Infrastructure and acquisition context for federal AI R&D

For federally hosted AI research environments, FedRAMP authorizations govern cloud services used to store data and run experiments. Microsoft Azure Government holds a FedRAMP High authorization listed on the FedRAMP Marketplace, enabling agencies to host sensitive R&D workloads that require High baseline controls¹³. For defense R&D subject to DoD Cloud Computing SRG impact levels, Microsoft documents Azure Government offerings aligned to DoD IL4 and IL5, which are commonly relevant for controlled unclassified information and certain mission data¹⁴. While vendor tooling is not a substitute for policy compliance, Microsoft publishes guidance to operationalize workflows consistent with NIST’s AI RMF within Azure AI services, which can help research teams structure documentation, measurement, and risk controls aligned to federal expectations¹⁵⁵¹².

These acquisition and compliance anchors matter for executing the Strategic Plan’s call for shared testbeds and reproducible experiments, and for meeting EO 14110’s safety and security directives during model development and evaluation on federal infrastructure³⁴.

Areas of divergence and outstanding needs

• Scope gap: NAIRR pilot versus NAIRR vision. The NAIRR pilot provides an immediate pathway to resources; the Task Force’s final report outlines a larger, long-term national platform for AI research infrastructure. The difference in scope and duration between a pilot and the full resource indicates further policy, budget, and implementation decisions are still required to reach the Task Force’s envisioned capability¹⁰¹¹.

• Voluntary guidance versus policy mandates. NIST’s AI RMF is voluntary guidance, whereas OMB M-24-10 establishes binding governance and risk management expectations for agency AI use. R&D programs operating on federal systems need to reconcile the RMF’s voluntary posture with OMB’s directives when planning, documenting, and evaluating research activities⁵¹².

Implications and actions for federal missions

• Align proposals and portfolios to the Strategic Plan. Use the 2023 priorities to justify long-term, high-risk research, testbed development, and trustworthy AI methods, and to structure multi-agency partnerships³.

• Leverage the NAIRR pilot. Use the pilot’s access pathways for compute, data, and tools to broaden participation and accelerate experiments; incorporate NAIRR services into program plans and community calls for proposals where appropriate¹⁰.

• Engage NSF AI Institutes. Connect agency missions with institute research agendas to co-develop use-inspired work, share datasets and benchmarks, and build workforce pipelines in priority domains⁶.

• Apply NIST AI RMF in R&D. Integrate RMF functions and categories into research design (risk mapping, measurement, transparency artifacts) to meet OMB expectations for governance and risk management and to prepare for transition-to-practice⁵¹².

• Procure compliant infrastructure. For cloud-hosted experiments and testbeds, use FedRAMP-authorized services (e.g., Azure Government FedRAMP High) and, for defense workloads, ensure DoD SRG IL alignment, documenting control inheritance and residual risks in research plans¹³¹⁴.

• Use responsible tooling to operationalize risk processes. Employ platform tools that support documentation, dataset and model cards, and evaluation workflows mapped to NIST AI RMF, while recognizing that agency policy requirements govern acceptance and compliance¹⁵⁵¹².

Bottom line

The National AI Initiative provides a durable statutory and policy backbone for federal AI R&D, with clear strategic priorities and tangible investments across science, health, defense, and standards. Agencies should use the Strategic Plan, EO 14110, OMB M-24-10, and NIST’s AI RMF to structure programs; use NAIRR and institute networks to scale access; and anchor experimentation on compliant infrastructure to accelerate trustworthy, mission-focused AI.

¹: National Artificial Intelligence Initiative Act of 2020 (Division E of NDAA FY2021) — https://www.congress.gov/bill/116th-congress/house-bill/6395/text ³: National Artificial Intelligence Research and Development Strategic Plan: 2023 Update — https://www.whitehouse.gov/wp-content/uploads/2023/05/National-AI-RD-Strategic-Plan-2023.pdf ¹⁰: Launching the National AI Research Resource pilot — https://www.whitehouse.gov/ostp/news-updates/2024/01/24/launching-the-national-ai-research-resource-pilot/ ¹¹: National AI Research Resource Task Force Final Report — https://www.whitehouse.gov/wp-content/uploads/2023/01/NAIRR_Final_Report.pdf ⁷: DARPA Announces 2 Billion Campaign to Develop Next Wave of AI — https://www.darpa.mil/news-events/2018-09-07 ⁶: NSF and partners invest 220 million to expand the National AI Research Institutes with 11 new institutes — https://www.nsf.gov/news/news_summ.jsp?cntn_id=307767 ⁸: NIH launches Bridge2AI program to expand use of AI in biomedical research — https://www.nih.gov/news-events/news-releases/nih-launches-bridge2ai-program-expand-use-artificial-intelligence-biomedical-research ⁵: NIST AI Risk Management Framework 1.0 — https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf ⁴: Executive Order 14110 — Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence — https://www.federalregister.gov/documents/2023/11/01/2023-24283/safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence ²: About the National AI Initiative — National AI Initiative Office — https://www.ai.gov/national-ai-initiative/ ¹²: OMB M-24-10 — Advancing Governance, Innovation, and Risk Management for Agency Use of Artificial Intelligence — https://www.whitehouse.gov/omb/briefing-room/2024/03/10/m-24-10-advancing-governance-innovation-and-risk-management-for-agency-use-of-artificial-intelligence/ ¹³: FedRAMP Marketplace — Microsoft Azure Government — https://marketplace.fedramp.gov/products/product/?product=Microsoft%20Azure%20Government ¹⁴: Azure Government compliance offerings — https://learn.microsoft.com/azure/azure-government/documentation-government-compliance ¹⁵: Operationalize the NIST AI Risk Management Framework with Azure AI — https://learn.microsoft.com/azure/ai-services/responsible-ai/nist-ai-rmf ⁹: Artificial Intelligence for Science (DOE Office of Science report) — https://www.osti.gov/biblio/1603221