DARPA participates in SBIR like every other defense component, but its selection logic is unusual enough that applying with a normal proposal wastes everyone's time. DARPA exists to create and prevent technological surprise. It funds work that could fail outright and would matter enormously if it did not. A proposal promising a reliable twenty percent improvement on an existing capability is a good proposal almost anywhere else and a losing one here.

Program managers drive everything

DARPA is organized around program managers who serve limited terms, arrive with a thesis about what should be possible, and build programs to test it. Topics reflect that person's agenda. When you read a DARPA SBIR topic, you are reading one technologist's bet, and your proposal has to convince them that you can move their bet forward.

This has a practical consequence: DARPA interests turn over faster than other agencies'. An office's priorities can shift substantially as program managers rotate. Read current topics and current program descriptions rather than assuming continuity from what the agency funded a few cycles ago.

What high-risk actually means here

Reviewers want to see a specific technical claim that current understanding says is hard, a reason to believe it might nonetheless be achievable, and an experiment that would settle the question. The structure of a strong DARPA-style proposal is closer to a research plan than a product plan:

  • The capability that does not exist today, stated concretely.
  • Why the conventional approach cannot get there.
  • The insight or mechanism you think changes that.
  • The measurement in Phase I that would prove you wrong.

Naming what would falsify your claim reads as confidence at DARPA, not weakness. Hiding the hard part reads as inexperience. This differs from how review panels at more conservative agencies weigh risk, where a plan that might fail can cost you points.

Where DARPA is a poor fit

Do not bring DARPA an integration project, a manufacturing cost reduction, a well-understood product that needs ruggedizing, or a service business. Those are legitimate and fundable, just not here — the service components fund exactly that kind of work, and the agency comparison points to better homes for it. Also be honest about your team: DARPA topics often assume depth in a narrow specialty, and a strong university research partnership can supply what a small company lacks, which is also the logic behind the STTR variant described in SBIR versus STTR.

If you win

DARPA awards are contracts with deliverables and a technical point of contact who will engage seriously with your results. Expect a demanding relationship and expect the program's direction to evolve. Transition planning matters as much as at any defense component, because DARPA hands technology off rather than fielding it, and a successful project needs a receiving organization somewhere in the services or industry. Specific award sizes, phase structures and topic release patterns vary by cycle; confirm them in the current solicitation rather than trusting a summary.

Projects House builds experimental hardware and test rigs for teams chasing this kind of work. Describe the claim you want to prove through the contact form.