The National Science Foundation runs its SBIR and STTR awards as a seed fund for startups commercializing deep technology, and it works differently from every other participating agency in one decisive way: you bring the idea. NSF does not publish a list of problems it wants solved and ask you to bid. It publishes broad technology areas, then judges whether your proposed innovation is technically risky, scientifically interesting and plausibly a business.
What NSF funds
NSF's remit spans essentially all of civilian science and engineering: advanced materials, biotechnology, semiconductors and photonics, robotics, distributed ledger and AI, energy and environment, medical devices and digital health, space technology, and more. The exact area names and groupings change between cycles, so read the current solicitation rather than an old blog post.
Two filters matter more than the area label. First, technical risk: NSF funds work that might fail for a scientific or engineering reason, not work that is merely unbuilt because nobody funded it. Incremental integration of existing parts is a weak fit. Second, commercial potential: NSF wants a company, not a paper, and expects evidence you have talked to customers. If your product is chiefly software, the fit question is worth reading up on in whether software wins SBIR awards.
The Project Pitch comes first
Before a full proposal, NSF asks for a short Project Pitch describing the technology, the innovation, the technical risk and the market. NSF reviews it internally and tells you whether you are invited to submit a full proposal — a fast answer, at essentially no cost, on whether you are in scope at all. Treat it as the cheapest fit test available anywhere in the SBIR system. Whether it is strictly required and how the submission windows are structured have changed over time; confirm both in the current solicitation.
Review and the program director relationship
Invited full proposals go to merit review against NSF's criteria for intellectual merit and broader impacts, adapted for a commercial context, with reviewers drawn from research and industry. A cognizant program director owns your proposal, and after an award that person is a real relationship rather than a mailbox — they can advise on scope, on supplements, and on the path from Phase I to Phase II. The habit of talking to a program officer early pays off at NSF more than almost anywhere.
Who should apply to NSF
NSF suits a small US company with a genuinely novel technical claim, a founder or partner who can defend the science, and a civilian market. It is a poor fit if your differentiation is business-model or design rather than technical, or if your customer is a defense program office — that work belongs at an agency that buys deliverables under contract, and the agency comparison covers where to go instead. Because NSF is investigator-initiated, the burden of framing the problem is entirely yours, which is exactly why the pitch step exists. For the broader picture of odds, see SBIR success rates.
Projects House provides the engineering substance behind an NSF proposal — feasibility analysis, prototypes and test data. Send your concept through the contact form.