Key Takeaways
- NSF invested $700M+ in AI in FY2025 — National AI Research Institutes award up to $20M per institute over 5 years
- DOE Genesis Mission: $293M specifically for AI applied to energy systems — open to national labs, universities, and some companies
- NIH Bridge2AI: $130M for biomedical data infrastructure — universities and health research nonprofits are primary applicants
- SBIR/STTR is the primary path for small AI companies — NSF Phase I up to $275K, no preliminary data or equity required
- DARPA funding requires attending Proposers' Days and building PM relationships — cold submissions to open BAA have very low success rates
Quick Answer
Federal AI grant funding is at an all-time high in 2026.
NSF alone invested $700M+ in AI in FY2025. Top programs: NSF National AI Research Institutes (up to $20M per institute), DOE Genesis Mission ($293M total), DARPA AIE and AI Forward, and NIH Bridge2AI ($130M). Universities are the primary recipients, but nonprofits, small businesses via SBIR/STTR, and industry consortia are eligible in specific programs.
In This Article
Why AI Grant Funding Is Surging in 2026
Three converging forces are driving the federal AI funding surge: executive policy, congressional appropriations, and geopolitical urgency.
The White House's 2023 Executive Order on AI directed every major federal agency to incorporate AI into their research missions, creating a cascade of new funding programs. The CHIPS and Science Act pumped an additional $280 billion into science and technology, with AI as a named priority. And with China racing to dominate AI at the national level, congressional support for AI R&D has been bipartisan — one of the few areas of consistent agreement.
The numbers back this up. NSF's AI investment grew from under $200 million in 2020 to over $700 million in FY2025. DOD's AI spending exceeds $1.8 billion annually across programs. NIH launched its first dedicated AI institute program. DOE created an entirely new mission around AI for science.
For researchers, nonprofits, and technology companies, this means more competition — but also substantially more money available than at any point in history.
NSF AI Programs in Detail
The National Science Foundation is the dominant civilian funder of AI research in the United States. Its portfolio spans fundamental research, workforce development, and equity-focused capacity building.
National AI Research Institutes
The flagship NSF AI program. Institutes receive up to $20 million over five years, making these among the largest individual awards NSF offers. Each institute is organized around a specific AI application domain — current institutes cover AI in agriculture, manufacturing, cybersecurity, climate, education, healthcare, and more.
Awards go to university-led consortia. A lead university partners with co-PI institutions, industry affiliates, and sometimes government labs. The collaborative structure is intentional: NSF wants institutes to serve as national hubs, not single-lab projects.
Key requirements: Strong interdisciplinary team, clear use-inspired research agenda, explicit workforce development plan, and a plan for diversity and inclusion. Letters of collaboration from industry or government partners strengthen proposals considerably.
New institute competitions are announced via NSF's CISE directorate. Check nsf.gov/cise and Grants.gov for active solicitations. The solicitation cycle typically opens in Q1 with letters of intent due in Q2.
ExpandAI: Capacity Building at MSIs
ExpandAI is specifically designed to build AI research and education capacity at Minority-Serving Institutions (MSIs) — HBCUs, Hispanic-Serving Institutions, Tribal Colleges, and similar. Awards range from $1M to $5M depending on track.
Two tracks exist: Partnership Track (MSI partners with an existing National AI Research Institute) and Pilot Track (stand-alone capacity building). MSIs that have historically been locked out of large NSF AI awards should treat ExpandAI as their entry point. Historically Black Colleges and Universities in particular have had strong success here.
CyberCorps and AI-Adjacent STEM Education
NSF's CyberCorps: Scholarship for Service program funds students who pursue cybersecurity careers, and AI security is now a recognized track. Awards go to institutions, which then provide scholarships to students who commit to federal service after graduation.
Beyond CyberCorps, NSF's IUSE (Improving Undergraduate STEM Education) program increasingly funds AI literacy and data science curriculum development. These awards ($300K–$1.5M) are accessible to smaller colleges and teaching-focused institutions that lack the research infrastructure for Institute-scale grants.
DOE Genesis Mission: $293M for AI in Energy
The Department of Energy's Genesis Mission is a $293 million initiative aimed at integrating AI into energy science and national infrastructure. It represents a fundamental shift in how DOE approaches both its scientific and operational mandates.
Genesis covers five application areas:
- Grid modernization: AI-driven control systems for the national electric grid, including demand response and renewable integration
- Fusion and plasma physics: Machine learning for plasma control in fusion reactors, a major priority given the Commonwealth Fusion breakthrough
- Climate and earth system modeling: AI accelerators for climate simulations, including emulators that run 1,000x faster than traditional models
- Materials discovery: AI-assisted identification of new materials for batteries, solar cells, and nuclear applications
- Scientific computing: AI co-pilots for DOE national laboratories using exascale computing resources
The bulk of Genesis funding flows through DOE's national laboratories (Argonne, Oak Ridge, Lawrence Berkeley, etc.), but university partnerships and industry collaborations are funded via Funding Opportunity Announcements (FOAs) published on science.osti.gov and Grants.gov. ARPA-E, DOE's high-risk energy research arm, also runs AI-specific programs under the Genesis umbrella with awards typically in the $1M–$5M range.
DARPA AI Programs: High Risk, High Reward
DARPA (Defense Advanced Research Projects Agency) funds the most ambitious and unconventional AI research in the federal portfolio. DARPA doesn't do incremental work — it funds 4-year bets on transformative ideas that could reshape national security technology.
DARPA AI Exploration (AIE)
AIE is DARPA's rapid-funding mechanism for early-stage AI concepts. Awards are typically $1M–$2M over 18 months, designed to quickly validate whether a novel AI idea is worth larger investment. AIE solicitations are published with very short response windows (sometimes 30 days), so researchers need to monitor darpa.mil/work-with-us/opportunities regularly.
Successful AIE projects can feed into larger DARPA programs worth $10M–$50M+. Topics in recent AIE cycles have included AI for electronic warfare, neurosymbolic reasoning, AI robustness and assurance, and multi-agent coordination.
DARPA AI Forward
AI Forward is DARPA's strategic program for advancing AI capabilities relevant to long-term U.S. defense posture. Unlike AIE, AI Forward solicitations are multi-year programs with milestone-gated funding. The program office PM (program manager) plays a large role — DARPA PMs have significant discretion in selecting projects and mentoring performers. Attending DARPA BAA (Broad Agency Announcement) proposers days is essentially mandatory before submitting.
NIH Bridge2AI: $130M for AI-Ready Biomedical Data
NIH's Bridge to Artificial Intelligence (Bridge2AI) program is a $130 million, four-year initiative focused on generating and sharing AI-ready biomedical and behavioral data. The core problem NIH identified: most health data is too messy, siloed, and poorly annotated for AI to use effectively. Bridge2AI exists to fix that.
The program funds four Bridge2AI Data Generation Projects, each producing large standardized datasets in a specific health domain (heart disease, voice disorders, mental health, and rare diseases). It also funds the Bridge2AI Standards Module, which develops the FAIR data standards for making biomedical datasets usable by machine learning systems.
Eligibility is broad — universities, medical centers, and qualifying nonprofits with biomedical research infrastructure. NIH expects teams to include both clinical researchers and computational scientists. The key deliverable isn't just research findings but actual, publicly accessible datasets that the broader AI research community can use.
Future funding opportunities under the Bridge2AI umbrella are announced via grants.nih.gov. Look for parent announcements from the National Center for Complementary and Integrative Health (NCCIH) and the Office of Data Science Strategy (ODSS).
USDA AI Grants: Precision Agriculture and Food Systems
The USDA is a quieter but substantial AI funder, particularly for agricultural applications. The National Institute of Food and Agriculture (NIFA) runs several programs that explicitly fund AI research:
- NIFA Agricultural AI Initiative: Funds AI applications for precision farming, crop disease detection, yield prediction, and autonomous farm equipment. Awards typically $500K–$3M.
- AFRI (Agriculture and Food Research Initiative): The largest competitive grants program at USDA, with AI as a named priority across food safety, climate adaptation, and agricultural systems tracks. Awards up to $10M for large integrated projects.
- SBIR/STTR through USDA: Agritech startups building AI tools for the farm sector can access Phase I ($175K) and Phase II ($450K) SBIR awards specifically for agricultural technology innovation.
Who Is Eligible for Federal AI Grants
| Organization Type | Best Programs | Typical Award Range |
|---|---|---|
| R1/R2 Universities | NSF Institutes, NIH Bridge2AI, DOE Genesis, DARPA | $1M – $20M+ |
| MSIs / HBCUs / HSIs | NSF ExpandAI, IUSE, USDA NIFA | $500K – $5M |
| Research Nonprofits | NIH Bridge2AI, NSF Institutes (as sub-awardees), DARPA AIE | $500K – $5M |
| Small Businesses (<500 employees) | SBIR/STTR (NSF, NIH, DOE, DARPA, USDA) | $275K – $2M+ |
| Industry Consortia | NSF Institutes (industry partner), DOE Genesis, NIST | Cost-share arrangements |
How to Position an AI Grant Proposal
AI is a hot funding area, which means competition is fierce. Proposals that win share several characteristics:
Lead with responsible AI, not just performance. Every major federal AI funder now requires applicants to address fairness, accountability, transparency, and safety. This is not a checkbox — reviewers look for substantive plans. Proposals that treat responsible AI as a core design principle (not an afterthought) consistently outperform those that don't.
Frame around national competitiveness. Congressional interest in AI is driven by the China challenge. Proposals that explicitly tie their research to U.S. technological leadership, workforce development, or national security applications resonate with program officers and review panels alike.
Demonstrate broad impact beyond academia. NSF and NIH increasingly use "broader impacts" as a major evaluation criterion. This means clear plans for technology transfer, open-source tools, training the next generation of AI researchers, and reaching underserved communities or institutions.
Show you have the data. A common failure mode in AI proposals is proposing research that requires datasets the team doesn't actually have access to. Reviewers know this. Demonstrate data access agreements, existing datasets, or a credible plan for data acquisition before you need funding to get it.
Build a team that spans disciplines. AI grant reviewers are skeptical of single-PI proposals for large awards. Bring in domain experts (clinicians, farmers, engineers) alongside the ML researchers. Show that the team has successfully collaborated before — joint publications, shared labs, or existing partnerships are strong signals.
Private Foundations Also Funding AI
Federal programs dominate AI funding, but private philanthropies are filling important gaps — particularly for AI safety, social impact, and basic research that federal agencies treat as too risky.
Schmidt Futures funds AI researchers at the postdoctoral and early faculty stage through the Schmidt Science Fellows program and direct grants. Eric Schmidt's philanthropies have given hundreds of millions to AI-related research, particularly at the intersection of AI and climate, biology, and national security.
Simons Foundation funds basic math and computer science research with strong AI components. The Flatiron Institute (Simons-funded) is itself a major AI research hub. Simons Math + X grants are notable for bringing AI methods into other scientific disciplines.
Open Philanthropy is arguably the largest private funder of AI safety research, with grants to academic groups, nonprofits (Redwood Research, ARC, MIRI), and university centers working on alignment, interpretability, and governance.
MacArthur Foundation funds AI policy, governance, and social impact research — particularly work examining AI's effects on labor markets, civil rights, and democratic institutions.
◆ Primary Sources & Further Reading
Related Articles
Frequently Asked Questions
Can nonprofits apply for federal AI grants?
Yes, but eligibility varies by program. NSF National AI Research Institutes accept applications from universities and their nonprofit partners. NIH Bridge2AI welcomes nonprofits with biomedical research missions. DARPA AIE programs are open to nonprofits as well as companies and universities. The SBIR/STTR programs are limited to small for-profit businesses. The key is to read each program's eligibility statement carefully — most large NSF awards require a university as lead applicant, with nonprofits as sub-awardees.
How much can a university receive from NSF for AI research?
NSF National AI Research Institutes can award up to $20 million over five years. Smaller NSF AI grants through ExpandAI or CISE core programs range from $500,000 to $3 million. NSF also funds individual PI grants in AI-adjacent areas for $200,000 to $500,000 for 3-year projects. Awards are highly competitive — the Institute program typically funds fewer than 10% of proposals.
What is the DOE Genesis Mission and who can apply?
The DOE Genesis Mission is a $293 million initiative focused on applying AI to energy systems — including grid modernization, fusion research, climate modeling, and clean energy optimization. Eligible applicants include national laboratories (which receive most funding), universities, and in some solicitations, private companies. Applications go through DOE's Office of Science and are submitted via Grants.gov or PAMS. Watch for specific Funding Opportunity Announcements (FOAs) from the Office of Science and ARPA-E.
Are there AI grants specifically for small businesses?
Yes. The primary federal pathway for small businesses is SBIR and STTR. NSF, NIH, DOE, DARPA, and DOD all run SBIR/STTR programs with AI-specific topics. NSF Phase I awards are $275,000; NIH Phase I up to $300,000; Phase II awards can reach $1 million or more. DARPA AIE also accepts small business applicants. The SBA's SBIR Road Tour offers free workshops to help small AI companies navigate the process.
Action Checklist
- Check NSF's ExpandAI and NSF-IUCRC solicitations on research.gov — these have lower competition than the flagship National AI Research Institutes ($20M awards)
- Small businesses: register on sbir.gov and set up topic alerts for AI/ML subtopics from NSF, NIH, DOE, and DARPA — Phase I awards up to $275K, no equity required
- Attend DOE ARPA-E and Office of Science Proposers' Days — posted on arpa-e.energy.gov; pre-solicitation intelligence dramatically improves proposal alignment
- Build university partnerships early — most large NSF awards require a university as lead applicant, with nonprofits and companies as sub-awardees
- For DARPA: track Proposers' Days at sam.gov/opp and engage with Program Managers directly — cold submissions to the open BAA have very low success rates; PM relationships are essential
Last updated March 2026. Grant programs and award amounts are subject to change with agency appropriations. Verify current solicitations on Grants.gov and individual agency websites before applying.