The three SBIR phases fund fundamentally different activities. Phase I ($150K-$314K) proves feasibility. Phase II ($750K-$2.1M) builds the prototype. Phase III is a sole-source government contract with no funding ceiling and no further competition required. Most SBIR content focuses on Phases I and II. Phase III -- the least discussed phase -- is where companies have landed contracts worth hundreds of millions of dollars.
The three phases at a glance
| Phase I | Phase II | Phase III | |
|---|---|---|---|
| Purpose | Prove feasibility | Full R&D and prototype development | Commercialization (government or commercial) |
| Duration | 6-12 months | 24 months | No limit |
| Amount (SBA cap) | Up to $314,363 | Up to $2,095,748 | No ceiling |
| Funding source | SBIR set-aside | SBIR set-aside | Agency procurement budgets (not SBIR) |
| Competition | Open solicitation | Competitive (Phase I awardees + D2P2) | Sole-source -- no competition required |
| What you deliver | Feasibility study, proof of concept | Working prototype, test data, commercialization plan | Production, deployed product, or continued R&D |
| Data rights | 20-year protection | 20-year protection | Carries forward from Phase I/II |
| Success rate | ~15-20% | ~50-60% of applicants | N/A (not competitive) |
The phases build on each other, but they're not required sequentially. Direct-to-Phase-II lets you skip Phase I. Fast-Track combines Phase I and II. And Phase III can start during Phase II, not just after it.
Phase I: the feasibility test
Phase I answers one question: can this work?
You're not building a product. You're designing experiments to prove or disprove your core technical hypothesis. The output is a feasibility study with data, not a prototype.
Phase I amounts by agency
| Agency | Phase I Amount | Duration |
|---|---|---|
| NSF | $305,000 | 6-12 months |
| NIH | Up to $314,363 | 6 months-2 years |
| DoD (Army) | Up to $250,000 | 3-6 months |
| DoD (Navy) | Up to $250,000 | 6 months |
| DoD (AFWERX) | $75,000 | 3 months |
| DOE | Up to $250,000 | 6-12 months |
| NASA | $150,000 | 6 months |
| USDA | Up to $175,000 | 8 months |
AFWERX is the outlier -- $75K over 3 months, and it's customer discovery (not R&D). See our AFWERX guide for why that's intentional.
What kills Phase I proposals: 5-10% fail administrative review before reaching reviewers. Missing forms, ineligible PI, budget over the cap. Another large chunk fails on framing -- describing a product instead of a research question. See our how to win an SBIR grant guide for the cross-agency tactical playbook.
Work requirements: the small business must perform at least 2/3 of Phase I work. (Under STTR, the sister program that pairs you with a university, it's 40% by the business and at least 30% by the research institution.)
Phase II: the build phase
Phase II shifts the question from "can this work?" to "can this become a product?"
Your Phase I data is the foundation. Phase II proposals without strong Phase I results are dead on arrival. Reviewers want to see that feasibility was demonstrated, not just that work was done.
Phase II amounts by agency
| Agency | Phase II Amount | Duration |
|---|---|---|
| NSF | Up to $1,250,000 | 24 months |
| NIH | Up to $2,095,748 | 24-36 months |
| DoD (Army) | Up to $2,000,000 | 12-18 months |
| DoD (Navy) | Up to $1,700,000 | 24 months |
| DoD (AFWERX) | Up to $1,250,000 | 21 months |
| DOE | Up to $1,600,000 | 24 months |
| NASA | Up to $850,000 | 24 months |
What changes from Phase I to Phase II
- Commercialization plan becomes critical. It was important in Phase I; in Phase II it's arguably the make-or-break section. Agencies want named customers, revenue projections, and a go-to-market strategy.
- You need data, not just ideas. Your Phase I results are the evidence. Weak Phase I data = weak Phase II proposal.
- Team expectations increase. "TBD" hires in your staffing plan are a red flag. Reviewers want to see the team is ready to execute.
- Budget scrutiny is higher. Every line item needs clear justification tied to specific aims.
Transition rates
About 50-60% of Phase I awardees who apply for Phase II receive it. That's a strong success rate for a grant of this size. But many Phase I winners never apply -- they pivot, raise VC instead, or run out of runway during the gap.
Performance benchmarks matter: companies with 20+ Phase I awards must maintain a 25% Phase I-to-Phase II transition rate. Companies with 51+ awards must hit 50%. Fall below these thresholds and you're locked out of new Phase I proposals for a year.
Phase III: sole-source contracts with no funding ceiling
Phase III is the most misunderstood and most valuable part of the SBIR program.
It is not a grant. It is not competitive. It is a contracting mechanism that lets the government buy your product -- or fund continued R&D -- without running a new competition. SBIR firms have received Phase III contracts worth hundreds of millions of dollars.
How Phase III works
When a government agency wants to continue work that "derives from, extends, or completes" your prior Phase I or Phase II R&D, they can award you a Phase III contract. The legal authority comes from 15 U.S.C. 638(r)(4).
Key characteristics:
- No funding ceiling. Phase III awards have ranged from thousands to hundreds of millions.
- No competition required. The agency awards directly to you. This is not technically a sole-source -- it's an "award without further competition" by statute.
- Funded by non-SBIR sources. Agency procurement budgets, other R&D funds, or private sector investment. It doesn't count against SBIR set-asides.
- Any contract type. Fixed-price, cost-plus, time-and-materials -- whatever the contracting officer prefers.
- Any federal agency. Cross-agency Phase III is explicitly allowed. NIH could award Phase III based on NSF-funded Phase I/II work.
- Can include further R&D. Phase III isn't limited to production. Continued development, testing, and integration all qualify.
20-year data rights
Your SBIR data rights are protected for 20 years from date of award (expanded from the previous 5-year period by a DoD final rule in December 2024). During this period:
- The government gets a limited, nonexclusive license to USE your data
- The government CANNOT disclose your data outside government -- not to competitors, not to other contractors
- After 20 years, the government gets "Government Purpose Rights" (not unlimited rights)
This is what enables sole-source Phase III. The government can't describe your work to another firm without violating your data rights, so they can't compete the contract. Your IP is shielded from competitors and from government disclosure -- creating a legal moat that protects your technology for two decades.
Who uses Phase III most
DoD dominates Phase III. Army, Navy, and Air Force all have active Phase III contracting programs. AFWERX built entire bridge mechanisms (STRATFI at $3M-$15M, TACFI at $375K-$2M) specifically to accelerate the Phase II-to-III transition. In FY2024, AFWERX transitioned 470 companies to Phase III, representing $1.44B in contracts (per the AFWERX FY2024 annual report).
NASA and DOE use Phase III but less systematically than DoD. NIH Phase III is rarer but happens through follow-on contracts and cooperative agreements.
What founders get wrong about Phase III
"It happens automatically." No. You have to actively pursue it by building relationships with program managers and contracting officers during Phase II.
"I need to finish Phase II first." Phase III discussions should start during Phase II. The companies that transition fastest are the ones who identify their government buyer before Phase II ends.
"It can only come from my original agency." Cross-agency Phase III is explicitly allowed. Any federal agency can award it based on your prior SBIR work.
"Phase III is just production." It can include continued R&D, testing, integration, and technology maturation -- not just manufacturing or fielding.
Alternative pathways
Not everyone needs to start at Phase I.
Direct-to-Phase-II
Direct-to-Phase-II (D2P2) lets companies with existing feasibility data skip Phase I and apply directly for Phase II funding. Several agencies offer it. Available at NIH, DoD agencies, and the Department of Education.
You'll need to demonstrate Phase I-equivalent results: published research, prototype data, or prior VC-funded development that proves feasibility. Without this evidence, D2P2 proposals don't compete.
Fast-Track (combined Phase I + II)
Fast-Track combines Phase I and Phase II into a single proposal, eliminating the 6-12 month funding gap between phases. Submit once, get reviewed once. If funded, you move from Phase I to Phase II without recompeting.
- NSF Fast-Track: up to $1,555,000 total ($400K Phase I + $1,155K Phase II). Requires NSF research lineage.
- NIH Fast-Track: combined submission, single review, eliminates the funding gap between phases.
- DOE Fast-Track: combined Phase I/II proposal available.
Best for teams with strong preliminary data who want to minimize the 6-12 month gap between phases.
Bridge funding
The gap between Phase I and Phase II kills momentum. Several mechanisms exist to bridge it:
- DoD Phase I Option: up to $50K for 4 months (Army), proposed as part of Phase I
- NSF Phase IB: extends Phase I by 6 months with matched third-party investment
- AFWERX STRATFI/TACFI: $375K-$15M bridge between Phase II and Phase III
- State bridge grants: several states offer $50K bridge grants for companies between phases
What the 2026 reauthorization changes
SBIR was reauthorized through September 2031 as of March 2026. The program experienced a 5-month authorization lapse (October 2025 through March 2026) that froze new solicitations, but agencies are now restarting. Key changes:
Strategic Breakthrough Awards: up to $30M for Phase II alumni with 100% matching funds. Milestone-driven, focused on technology transition. The biggest structural change to the program in years.
Proposal submission caps: agencies will set limits on how many proposals a single company can submit, starting FY2027.
Enhanced security screening: mandatory due diligence on ownership, patents, and foreign affiliations for all awardees.
The core Phase I/II/III structure is unchanged. Check sbir.gov for the latest solicitation schedules as agencies restart.
How to think about the three phases
Phase I is cheap to try (3-4 hours for an NSF pitch, $305K if you win) and proves whether your technology is fundable.
Phase II is the build phase -- real money ($1M-$2M), real deliverables, and the commercialization plan that determines whether you get to Phase III.
Phase III is where the economic model flips. You're no longer applying for grants. You're a government vendor with sole-source contracting authority, 20-year data rights, and no funding ceiling. That's not a grant -- it's a business.
The smartest founders plan backwards from Phase III and use Phase I and II to get there.
For the full agency-by-agency breakdown, see our SBIR funding amounts reference. For tactical proposal advice, see how to win an SBIR grant.
Want help planning your SBIR pathway?
Whether you're considering a first Phase I, preparing a Phase II with your feasibility data, or exploring Phase III transition, our Strategy Review maps the right path for your technology and stage.