The Department of Defense handed startups $4.1 billion through SBIR and STTR programs in fiscal year 2025. Most founders have never heard of it. The ones who have usually butcher the application. This is how the process actually works — from someone building a defense tech company right now.
The Numbers Nobody Talks About
SBIR — Small Business Innovation Research — is the federal government's venture capital arm, except it takes zero equity. Let that sink in. The Pentagon will write you a check for $75,000 to $256,000 in Phase I just to prove your idea isn't insane. If it works, Phase II hands you $750,000 to $1.25 million to build it. Phase III has no ceiling — that's where the real contracts live.
In 2025, the DoD funded 4,217 Phase I awards and 2,108 Phase II awards across all branches. The success rate on Phase I proposals hovers around 15-25% depending on the agency. Compare that to Y Combinator's ~1.5% acceptance rate, and suddenly the Pentagon looks like the friendlier option.
Here's the part that makes VCs nervous: SBIR money is non-dilutive. You keep 100% of your company. The government gets usage rights, not ownership. Peter Thiel built Palantir partly on CIA money, but he had to give up equity for In-Q-Tel's investment. SBIR doesn't work that way.
How It Actually Works (vs. How People Think It Works)
What people think: You write a 50-page proposal, mail it to "The Pentagon," and wait six months for a form rejection.
What actually happens: You find an open topic on a specific agency's solicitation portal, write a 15-20 page technical proposal, submit it electronically, and hear back in 90-120 days. The whole process is surprisingly modern — especially at AFWERX.
The Agencies That Matter
| Agency | Focus | Phase I Award | Founder-Friendly? |
|---|---|---|---|
| AFWERX (Air Force) | Broad — AI, autonomy, space, cyber | $75K (Open Topic) | Most startup-friendly agency. Period. |
| DARPA | Bleeding edge R&D | $175K-$256K | High bar, high reward. They want moonshots. |
| DIU | Commercial tech → military | OTA contracts (different) | Fastest path to production contracts. |
| Army SBIR | Ground systems, logistics, C4ISR | $175K-$256K | More traditional. Specific topics only. |
| Navy SBIR | Maritime, undersea, EW | $175K-$256K | Good for hardware + software combos. |
AFWERX Open Topic: The Cheat Code
AFWERX runs an "Open Topic" that accepts proposals year-round. No specific problem statement — you pitch your solution to any Air Force or Space Force problem. Phase I is $75,000 for a feasibility study. That's enough to validate an MVP, run a pilot, and prove you're not hallucinating your product-market fit.
The process has three stages:
1. Pitch Day — Submit a brief (2-page) concept. AFWERX evaluators score it. This takes about two weeks.
2. Phase I Proposal — If selected, you write the full 15-20 page proposal. Technical approach, team qualifications, commercialization plan, budget.
3. Award & Execute — You get the money. You build the thing. You report results.
The 2-page pitch is the filter. Get through that, and your Phase I odds jump to roughly 40-50%.
Who Gets Funded and Why
I've read hundreds of SBIR abstracts on SBIR.gov. The winners share three traits:
1. They solve a specific warfighter problem. Not "we use AI to improve defense." That means nothing. "We reduce ISR data processing from 48 hours to 90 seconds using multi-agent correlation" — that gets funded. The more specific, the better. Name the platform. Name the unit. Name the pain.
2. They have a credible team. You don't need a PhD, but you need domain expertise. A solo founder with five years of intelligence community experience beats a team of three Stanford PhDs who've never talked to a military end user. The DoD cares about whether you understand their world.
3. They show commercial viability. This surprises people. The government wants dual-use technology. They want your company to survive after the SBIR money runs out. Show them a commercial market for your tech and your proposal immediately looks stronger.
The Proposal: What Actually Matters
Forget everything you know about grant writing. SBIR proposals read more like VC pitch decks with technical appendices. Here's the anatomy:
Technical Volume (60% of your score): This is the meat. Problem statement. Your approach. Why it works. What makes it novel. Technical risks and mitigation. Be specific — evaluators are subject matter experts, not bureaucrats.
Key Personnel (20% of your score): Who's doing the work? What's their background? If you're a solo founder, own it. Show relevant experience. If you're technical enough to build the prototype yourself, that's actually an advantage — less burn rate, more progress per dollar.
Commercialization Plan (20% of your score): Where does this go after SBIR? Who are your target customers — military and commercial? How big is the market? Do you have letters of support from potential users? (These matter. Get them.)
Common Mistakes That Kill Proposals
Being too vague. "Our AI platform will transform military intelligence" is a rejection sentence. What AI? What intelligence? Which military unit? Be ruthlessly specific.
Ignoring the customer. The best proposals include evidence of customer discovery. Did you talk to the end users? Do you have a letter of support from a squadron commander, a program manager, a COCOM J2? If not, get one.
Over-scoping Phase I. You have $75K-$256K and 6-12 months. Don't promise to build a production system. Promise to prove feasibility, run a limited demo, and validate the approach. Phase II is for building the real thing.
Treating it like academic research. SBIR is not a grant for your thesis. The DoD wants transition potential — tech that can move to a Program of Record. Frame everything around deployment, not publication.
The Hidden Benefit: Security Clearances
Here's something nobody mentions in SBIR guides. When you get a Phase II award, you can start the facility clearance process. Your company gets an FCL (Facility Clearance), and your key personnel can get individual security clearances. This is enormous.
Without a clearance, you can't work on classified programs. With one, you're suddenly eligible for contracts that 90% of startups can't touch. The SBIR program is one of the very few ways a startup can get cleared without already having a classified contract. It's a bootstrapping mechanism for the entire defense industrial base.
The Math That Makes This Obvious
Let's run the numbers on AFWERX Open Topic:
• Time to write 2-page pitch: ~20 hours
• Odds of making it past pitch: ~50%
• Time to write Phase I proposal: ~80 hours
• Odds of Phase I award: ~40-50% (post-pitch selection)
• Expected value: $75,000 × 25% = $18,750 per attempt
That's $18,750 in expected value for roughly 100 hours of work — $187/hour. Non-dilutive. And if you win, you have government validation, a path to Phase II ($750K+), and a clearance pathway.
Compare that to raising a seed round: 3-6 months of pitching, 5-10% of your company, and a board seat for someone who's never operated in your domain.
So What?
The SBIR program is the single most underutilized funding mechanism for technical founders. Palmer Luckey started Anduril with VC money because he could. Most of us can't. SBIR is how you build a defense company without selling your soul or your cap table.
The Pentagon needs what you're building. They've said so publicly — the National Defense Strategy explicitly calls for more innovation from non-traditional contractors. The money is allocated. The process is documented. The only missing piece is founders who actually apply.
Start with AFWERX Open Topic. Write the 2-page pitch. The worst outcome is you spend 20 hours learning how the defense acquisition system thinks about your technology. The best outcome is $75K in non-dilutive funding and a path to building something that actually matters for national security.
Your move.
