An "SBIR mill" is a company that has learned to win federal research awards repeatedly without ever turning the work into a product. The term is unofficial, but the idea behind it drives real decisions inside agencies, and a founder drifting in that direction will find later applications getting harder rather than easier.
What the pattern looks like
The profile is consistent: a strong technical team, a long list of Phase I awards, a few Phase II awards, and almost no revenue that did not come from a federal agency. Every project ends in a final report. None ends in a customer. Nothing fraudulent is happening — the research is real and the deliverables are delivered — but the program was built as a bridge to a market, and this company has made it the market.
Sustaining that takes genuine skill. Proposal writing quietly becomes the core competency, and engineers learn to size each idea to a solicitation rather than to a customer's problem. It is a workable business for a while. It is a fragile one.
Agencies track your commercialization history
Participating agencies collect data on what happened after past awards: follow-on funding, Phase III work, product sales, licensing, and outside investment raised. Prior awardees are generally asked to report this in a commercialization history section of a new proposal, and some agencies derive a score from it. The exact form and the weight it carries vary by agency, so read the current solicitation rather than assuming one agency's practice is the rule.
What is consistent is the direction of travel. A firm with many awards and no commercial outcomes is visible to reviewers and to program officers alike. It rarely produces a formal disqualification; it produces a quiet erosion of your odds, which is worse, because nobody tells you it happened. The other levers on that number are covered in our piece on what actually moves SBIR success rates.
Why the record follows you
A commercialization plan is read against your history. From a first-time applicant it is a promise, graded on whether the reasoning holds together. From a tenth-time applicant it is graded against nine earlier promises. If those did not come true, the new one carries less weight however well it is written — which is why the commercialization plan deserves as much attention as the technical volume, and why the jump from Phase I to Phase II gets harder for firms with a thin record.
The honest warning
Grant-funded research is easy to slip into. It pays salaries, it is intellectually satisfying, nobody dilutes you, and another solicitation is always coming. The test to run on yourself once a year is simple: name the customer who would pay for this work if the agency stopped funding it, and say roughly what they would pay. If you cannot answer, you are running a contract research shop. That is a legitimate business, but it is a different one, and it should be staffed and financed as such rather than run as a startup waiting for an exit that is not coming. The wider choice is laid out in grants versus investors.
Projects House helps deep-tech teams turn funded research into something a customer will buy. Tell us what you are building through our contact form.