There are 11+ federal agencies that fund startups, each with different award amounts ($75K to $600K Phase I), different timelines (90 days to 12 months), and different review cultures. Applying to the wrong agency wastes months and produces proposals that don't match what reviewers want to see. This guide helps you pick the right starting point based on your sector, stage, and how fast you need funding.
The short answer: most tech startups should start with NSF. Most health startups should start with NIH. Most defense startups should start with AFWERX. But the real answer depends on your technology, stage, and competitive position. The best strategy is usually applying to 3-5 agencies with tailored proposals. Cada Partners builds these multi-agency strategies for startups across every sector, and the agency decision is always the first question we answer.
Which 3 questions determine your best-fit agency?
Before comparing agencies, answer three questions:
1. What sector is your technology in? This narrows the field from 11 agencies to 2-4. AI and software companies have different agency options than biotech companies or defense hardware companies.
2. What stage is your technology at? Early-concept (TRL 1-3) technologies fit NSF and DARPA. Working prototypes (TRL 4-6) fit most SBIR programs. Commercial products ready for government use (TRL 7+) fit DIU better than SBIR.
3. How fast do you need funding? If you need an answer in 90 days, AFWERX and DIU are your options. If you can wait 6-9 months, NSF is the highest-value path. If you can wait 12 months, NIH provides the largest health-focused awards.
How do all SBIR agencies compare side by side?
| Agency | Phase I | Phase II | Review Type | Timeline | Best For |
|---|---|---|---|---|---|
| NSF | $305K | $1.25M | Peer review | 5-7 months | Deep tech, AI, materials, broad science |
| NIH | $314K | $2M | Study section (peer) | 6-9 months | Health, biotech, clinical, digital health |
| ARPA-H | $600K | $3.5M | PM-directed | 9-15 months | Breakthrough health (10x innovations) |
| AFWERX | $75-180K | $1.25M | AF evaluation | ~90 days | Defense/dual-use, any sector with AF application |
| Navy | $240K | $1.8M | Navy evaluation | 90-120 days | Maritime, undersea, naval aviation, cyber |
| Army | $250K | $2M | Army evaluation | 90-120 days | Ground systems, C4ISR, soldier systems |
| DARPA | $250K | $1.8M | PM-directed | 4-6 months (SBIR) | Breakthrough defense/dual-use technology |
| DOE | $200K | $1.1M | Merit review | 4-5 months | Clean energy, nuclear, grid, materials |
| NASA | $150K | $850K | Peer review | ~6 months | Space, aeronautics, earth science |
| USDA | $175K | $600K | Peer review | ~6 months | Agriculture, food safety, rural development |
| DHS | $175K | $1-1.5M | DHS evaluation | 3-6 months | Cybersecurity, border, disaster response |
| EPA | $100K | $400K | EPA review | ~6 months | Environmental technology, water, waste |
| DIU | $500K-5M (OTA) | Production follow-on | DIU + end-user | 3-6 months | Commercial products for defense (TRL 5+) |
NSF: the default starting point (and when it's not)
NSF is the best first SBIR for most technology startups. Here's why:
- Lowest barrier to entry. The Project Pitch is 3,000 characters -- about 500 words. You get feedback in 3 weeks. No other agency offers this.
- Broadest scope. NSF funds AI, materials, robotics, IoT, quantum, biotech, cleantech, edtech, and more. If your technology involves genuine R&D, NSF probably has a relevant topic area.
- Highest Phase I value. $305K is among the largest Phase I awards.
- Credibility signal. NSF SBIR is "America's Seed Fund." The brand carries weight with VCs and academic partners.
- Your technology is primarily a health/clinical application (go to NIH)
- You need funding in less than 4 months (go to AFWERX or DIU)
- Your innovation is product development, not R&D (NSF funds science problems, not engineering problems)
- You're a defense-only company with no commercial market (go to DoD SBIR)
NIH: health and biomedical (and the preliminary data catch)
NIH is the largest biomedical research funder in the world. For health startups, it's the anchor program in any grant roadmap.
The preliminary data reality. NIH reviewers heavily weight preliminary data. A startup with no pilot data, no publications, and no clinical results will score poorly regardless of the innovation. This is different from NSF (where preliminary data is helpful but not critical) and ARPA-H (where the concept matters more than the data).
Institute matching matters. NIH has 27 institutes and centers, each with different priorities. Submitting a digital health AI proposal to NIBIB (biomedical imaging) instead of NLM (library of medicine / informatics) can mean a different study section, different reviewers, and a different outcome. Match carefully.
The PI question. NIH doesn't require the PI to be a US citizen, but the company must be US-based. This is a significant advantage over some DoD programs with citizenship requirements.
DoD: the biggest SBIR funder (AFWERX, Navy, Army, DARPA)
The Department of Defense accounts for the largest share of SBIR funding across all agencies. But "DoD SBIR" isn't one program -- it's a collection of very different components:
AFWERX (Air Force). Fastest timeline (~90 days), lowest Phase I ($75K), but the most developed transition pathway (STRATFI can reach $15M). Open Topic is the most flexible DoD entry point.
Navy SBIR. Specific topics aligned with naval warfare needs. $240K-$280K Phase I. Topics released annually. More traditional proposal format than AFWERX.
Army SBIR. Similar structure to Navy. $250K Phase I, $1.7M Phase II. Topics focused on ground systems, soldier modernization, and C4ISR.
DARPA. SBIR and BAA pathways. SBIR is $250K Phase I with fixed topics. BAAs are open-ended with $500K-$10M+ awards. Highest innovation bar ("10x not 10%"). PM-directed review.
SOCOM. Special Operations Command. Smaller program, focused topics. Good for companies with SOF-relevant technology (ISR, communications, medical, mobility).
Which DoD component? Match by mission area. Air power = AFWERX. Maritime = Navy. Ground = Army. Special operations = SOCOM. Cross-cutting breakthrough = DARPA. Commercial product ready for deployment = DIU.
ARPA-H: breakthrough health (the 10x bar)
ARPA-H is the newest agency and the most ambitious. $1.5B FY2026 budget, average project awards of $26M, and SBIR Phase I up to $600K.
But the bar is fundamentally different from NIH. ARPA-H doesn't fund better versions of existing treatments. It funds technologies that cannot be achieved through conventional approaches.
When to choose ARPA-H over NIH: Your technology could transform health outcomes at population scale, not just improve them incrementally. You can articulate a clear Heilmeier narrative and you're comfortable with PM-directed review (no peer panel) and milestone-based OT contracts.
When to choose NIH over ARPA-H: Your technology needs iterative clinical validation, you have strong preliminary data that peer reviewers would value, or your innovation is significant but not "10x revolutionary."
DOE: clean energy and the LOI requirement
DOE SBIR funds clean energy, advanced manufacturing, nuclear science, materials science, and high-performance computing. $250K Phase I, $1.6M Phase II.
The LOI difference. DOE often requires a Letter of Intent before full proposal submission. This is a brief (1-page) description to help DOE assign reviewers. It's not competitive -- but missing the LOI deadline means you can't submit the full proposal.
Best for: Energy storage, solar/wind technology, grid modernization, nuclear innovation, hydrogen, carbon capture, advanced materials, and fusion-related technology.
USDA, EPA, DHS, and the smaller agencies
USDA SBIR. $200K Phase I, $650K Phase II. Narrower topic areas than large agencies. Best for agricultural technology, food safety, rural broadband, forestry, and natural resource management. Administered through NIFA.
EPA SBIR. $100K Phase I, $400K Phase II (the smallest SBIR program). Contracts, not grants. Best for environmental monitoring, water treatment, waste management, and remediation technology.
DHS SBIR. $250K Phase I, $1M Phase II. Focused on homeland security: cybersecurity, border technology, disaster response, chemical/biological detection. Fewer topics but less competition than DoD or NIH.
DIU: when you're past the R&D stage
DIU isn't SBIR -- it uses Other Transaction Authority to fund prototypes of commercial solutions. But it belongs in this decision framework because many startups that think they need SBIR actually need DIU.
Choose DIU over SBIR if: Your technology is commercially mature (TRL 5+), you have a working product, and you can articulate a specific defense use case. DIU awards in 60-90 days for $500K-$5M, with a direct path to sole-source production contracts.
Choose SBIR over DIU if: You're still in R&D, you need to prove feasibility, or you don't have a commercial product yet.
Decision tree by startup stage
| Your Stage | Best Primary Agency | Best Secondary | Why |
|---|---|---|---|
| Pre-seed, concept only | NSF (Project Pitch) | DARPA (if defense) | Lowest barrier, fastest feedback |
| Seed, early prototype | NSF, NIH (if health) | AFWERX (if defense) | Fund feasibility R&D |
| Post-seed, working prototype | NIH, DoD SBIR | ARPA-H (if 10x health), DOE (if energy) | Fund validation and testing |
| Revenue, commercial product | DIU (if defense), Phase II | STRATFI (Air Force) | Fund scale and deployment |
Decision tree by technology sector
| Your Sector | Start Here | Also Consider | Avoid |
|---|---|---|---|
| AI/ML (general) | NSF | DARPA I2O, NIH (health AI) | USDA, EPA |
| Biotech / health | NIH | ARPA-H, NSF (if fundamental) | AFWERX (unless military health) |
| Clean energy | DOE | NSF, EPA | NIH, DHS |
| Defense hardware | AFWERX | Navy, Army, DARPA, DIU | NSF (unless dual-use) |
| Cybersecurity | DHS | DARPA I2O, AFWERX | USDA, EPA |
| Space technology | NASA | SpaceWERX, DARPA STO | NIH, USDA |
| Agricultural technology | USDA | NSF, EPA, DOE (bioenergy) | DHS, Navy |
| Education technology | NSF (EHR) | ED (if it exists) | Most others |
What if your technology fits multiple agencies?
If your technology is relevant to 3+ agencies, that's a feature, not a bug. A multi-agency grant roadmap is the highest-ROI approach.
The key rules for stacking:
- Each proposal must be unique. Same technology, different framing for each agency's priorities.
- Sequence by timeline. Start with the fastest-feedback agency (NSF pitch or AFWERX), then layer on longer-timeline agencies (NIH, DOE).
- Don't duplicate scope. If you win two awards for overlapping work, you'll have compliance problems. Each proposal should cover a distinct phase or aspect of your R&D.
- Use wins as momentum. An NSF Phase I award strengthens your NIH proposal. A DoD Phase I strengthens your Phase II. Awards beget awards. This stacking approach is central to how Cada builds grant roadmaps for clients.
What are the agency-specific gotchas that trip up first-time applicants?
| Gotcha | Agencies Affected | What to Watch For |
|---|---|---|
| PI citizenship required | Some DoD components | Army and SOCOM may require US citizen PI. NSF and NIH do not. |
| Preliminary data critical | NIH | Proposals without pilot data score poorly. Plan 6-12 months of data collection before applying. |
| University partner required | STTR (all agencies) | STTR requires a research institution partner performing 30-40% of work. SBIR does not. |
| Customer letter required | AFWERX Phase II | Customer Memorandum from AF end-user is mandatory for Phase II evaluation. |
| LOI deadline | DOE | Missing the Letter of Intent deadline means no full proposal. Deadlines are strict. |
| Ownership rules | All SBIR | Company must be 51%+ owned by US citizens/permanent residents. VC-heavy cap tables can violate this. |
| SAM.gov registration | All federal | Takes 2-6 weeks. Start immediately if you haven't. Required for any submission. |
| SBIR reauthorization | All SBIR agencies | SBIR was reauthorized through 2031 in early 2026. All agencies are now issuing new solicitations. |
For a personalized analysis of which agencies fit your specific technology, team, and competitive position, our Strategy Review maps your startup to the best-fit programs and tells you where to start. See our SBIR guide for startups for the fundamentals, or dive into specific agency guides: NSF, AFWERX, DARPA, ARPA-H, or DIU.