Founders discover SBIR, read that the program awards billions a year in non-dilutive funding, and then hit the real question: eleven federal agencies participate, each with its own solicitation, deadlines, award sizes, and idea of what a good proposal looks like. Applying to the wrong one costs a full cycle — often three to six months of work — and produces a declination that had nothing to do with your technology. The choice is mostly mechanical once you understand what each agency is buying. The mistake is assuming they all buy the same thing.
The single most important distinction: topics or no topics
Federal SBIR agencies split into two camps, and this matters more than any other difference.
Mission agencies publish topics. DoD, NASA, DOE, DHS, DOT and others release specific topic descriptions written by program managers with an actual problem to solve. Your proposal must respond to a topic; if your technology does not map to one, you wait for the next release. These agencies are procuring a solution.
Investigator-initiated agencies take your idea. NIH and NSF let you propose what you want within broad program areas. You are not answering their question; you are convincing them yours is worth funding.
That distinction drives everything downstream. A topic-driven application is a sales document aimed at one program manager's requirement. An investigator-initiated application is a persuasion document aimed at a peer review panel.
Agency profiles at a glance
| Agency | Structure | Best fit | Notes |
|---|---|---|---|
| NIH | Grant, investigator-initiated | Devices, diagnostics, therapeutics, health software | Largest civilian budget; three receipt dates a year; study section review |
| NSF | Grant, investigator-initiated | Deep tech, any field, commercial focus | Project Pitch screening required first |
| DoD | Contract (mostly) | Defense hardware, sensors, materials, autonomy | Topic-driven; each service branch runs its own |
| DOE | Grant | Energy, grid, storage, advanced manufacturing | Annual topics; letter of intent usually required |
| NASA | Contract | Space hardware, propulsion, materials, avionics | Single annual solicitation; tight subtopics |
| USDA | Grant | Agriculture, food science, rural development | Smaller applicant pool; annual cycle |
| DOT and DHS | Contract | Transportation safety, detection, screening | Small programs, very specific topics |
Grant agency or contract agency changes how you work
A grant is financial assistance: the agency wants the public benefit, you own the direction, the deliverable is progress toward stated aims. A contract is a purchase: the agency has requirements and acceptance criteria, and the government is a customer.
Contracts bring more administrative load — cost accounting, scheduled invoicing, government property rules. They also bring a program manager inside the agency who wants your technology to work, and a path to becoming a supplier. For a hardware company aiming at defense or aerospace, that relationship often beats the award itself. Our guide to selling a new product to the government covers what happens after the SBIR ends, and defense product development covers the requirements and testing culture you are signing up for.
Award sizes are not uniform
Statutory guidelines set reference amounts, but agencies deviate and some deviate a lot. Phase I awards commonly land between $50,000 and $310,000. DoD Phase I awards at some branches are deliberately small and short, while NIH Phase I awards can be considerably larger and run a full year. Phase II typically spans roughly $500,000 to $2 million.
Match the award to the work. If your feasibility question requires building and testing hardware, a small short Phase I leaves you unable to produce a result worth a Phase II — a common quiet failure. The phase mechanics are broken down in SBIR Phase I versus Phase II.
Read the topic release calendar before you plan anything
Each agency publishes when solicitations open and close, and the topic-driven agencies publish topics ahead of the open date — usually with a pre-release window in which you may contact the topic author directly. That window is the highest-value period in the whole process: you can ask what problem sits behind the topic, whether your approach is in scope, and what has already been tried. Once the solicitation opens, that contact is cut off.
Practical sequence:
- Subscribe to solicitation notifications for the two or three agencies plausibly relevant to you.
- At pre-release, read every topic in your technical area, not just keyword matches.
- Contact topic authors during the pre-release window with a short, specific question.
- Decide fit honestly — a partial match almost never wins.
- Start drafting only once you have a topic you can quote in your first sentence.
NSF's equivalent gate is the Project Pitch, a short screening submission that returns an invitation or a clear no. Treat it as a free fit check.
How to decide when two agencies fit
Overlap is common: a biosensor could go to NIH or DoD, a drone payload to DoD, DHS, or NASA. Weigh these:
- Who is the eventual customer? Selling to hospitals means NIH reviewers are clinicians who know your market. Selling to the Army means a DoD program manager may become your first buyer.
- Which review culture fits your evidence? NIH study sections reward rigorous study design and preliminary data; DoD topic authors reward a solution to their stated problem, on schedule.
- How much administrative capacity do you have? A solo founder carrying a DoD contract's reporting burden alongside the engineering will struggle.
- Timing. If DOE's annual solicitation just closed and NIH's next receipt date is eleven weeks out, that may settle it.
You may apply to more than one agency; many companies do. What you may not do is take duplicate funding for the same work, and concurrent submissions must be disclosed.
Do not overlook the smaller agencies
The competition math favors the unfashionable. USDA, DOT, NIST, EPA, and the Department of Education all run SBIR programs with far fewer applicants than NIH or NSF. A legitimate angle into one of those missions — a food handling device, a road safety sensor, a measurement instrument — improves your odds substantially for the same writing effort, and the same logic runs through the wider landscape of federal, state, and private funding sources for inventors.
Also check whether the STTR variant fits better: it requires a research institution partner performing a defined share of the work, an advantage when a university lab already holds the core science. See SBIR versus STTR and partnering with universities on product R&D. For health specifically, NIH grants for medical device startups covers that path.
Projects House helps founders scope the technical work behind an agency submission — feasibility milestones a program manager will accept, realistic cost and schedule, and a plan that turns a Phase I result into something manufacturable. Reach us through the contact form.