Does NSF SBIR/STTR Fund Research, Not Development? Here's the 2026 Answer
Short answer: No — not in the absolute sense the claim implies. NSF SBIR/STTR funds commercialization-oriented R&D. A Phase I proposal wins when it centers on resolving high-risk technical uncertainty, not when it simply avoids the word "development." The real dividing line isn't research vs. development — it's high-risk R&D vs. straightforward engineering or incremental product build-out.
This is a common question for founders scoping their first NSF Phase I proposal, usually triggered by a widely shared Reddit thread claiming NSF proposals get rejected for being "too developmental." Below is what NSF's own 2026 guidance says, where that popular claim holds up, and where it oversimplifies.
Where the "research, not development" claim comes from
The claim traces back to a real document: George A. Hazelrigg's Research 101 for Engineers, hosted on NSF's America's Seed Fund site. After reviewing "many hundreds" of panels and thousands of proposals over 30 years at NSF, Hazelrigg concluded that proposals framed around new knowledge outperformed those framed around building an artifact. He defines research as the process of finding out something not already known, and argues that artifact-centered proposals tend to read as development and score lower.
That's a real and useful data point. But the same document includes a caveat easy to lose in a social-media summary: it explicitly states the views are the author's own and don't necessarily reflect NSF or federal policy. It's applicant guidance from deep institutional experience — not a binding statement of current program rules.
What does NSF actually fund in 2026?
NSF funds high-risk technical R&D with a credible path to commercialization — not research in a purely academic sense, and not routine engineering. The 2026 SBIR/STTR solicitation frames the program as relaunching to help startups and small businesses turn high-risk technologies into products and services with commercial impact, offering up to $2 million total across Phase I (up to $305,000, 6–18 months) and Phase II (up to $1.25 million).
NSF's own "What is R&D?" guidance sharpens the target further: the project should determine the scientific and technical feasibility of a new concept that could become a product, process, or service. That definition is broad enough to include design, development, and improvement of prototypes — so "development" itself isn't disqualifying. What NSF screens out is work that's really just execution: building something whose feasibility is already known.
Where exactly is the line between R&D and disqualifying "development"?
The line is between resolving genuine technical risk and doing predetermined engineering work. NSF's Project Pitch — required before any Phase I full proposal — asks applicants to show the work will prove technical feasibility or significantly reduce technical risk, and states plainly that proposed work must be R&D "rather than straightforward engineering or incremental product development tasks." Describing product features or customer benefits isn't sufficient; applicants have to name the specific high-risk technical innovation and the R&D needed to prove it out.
This shows up again in the full proposal instructions: the project summary must identify the technical hurdles the R&D addresses and why they're crucial to commercialization, and the project description needs a detailed R&D plan with milestones, risk mitigation, and quantitative success criteria — a generic task list can get a proposal returned without review.
So is "avoid development framing" the main reason proposals get rejected?
No — and treating it as the single dominant failure mode is the biggest way this popular claim oversimplifies things. NSF's merit review runs on three explicit criteria: Intellectual Merit, Broader Impacts, and Commercial Impact. Reviewers also weigh market opportunity, durable technical advantage, business model strength, and team readiness. NSF doesn't publish a ranked list of decline reasons — only the criteria proposals are judged against — so any claim about "the number one reason" should be treated with caution.
Consulting sources reinforce that rejections are usually multifactorial, not single-cause:
The NC Small Business and Technology Development Center flags poor agency fit, overpromising against the available budget/timeline, unclear measurable milestones, and insufficient innovation (a tweak rather than a genuine advance) as recurring pitfalls.
A University of Wisconsin Center for Technology Commercialization recap of a BBCetc workshop quotes consultant Megan Varnum on NSF wanting "revolutionary, not evolutionary" technology, paired with aggressive commercialization plans and real market understanding — plus credible letters of support tied to actual partners or customers.
E.B. Howard Consulting points to a more mundane but increasingly costly failure mode: teams working from stale templates instead of treating the solicitation, the America's Seed Fund instructions, Research.gov materials, and NSF's Critical Information page as one connected system.
What changed in the 2026 NSF SBIR/STTR cycle?
NSF consolidated its solicitation structure and tightened the pipeline rules teams need to plan around. NSF 26-510 (posted May 2026) merged the old separate Phase I, Phase II, and Fast-Track solicitations into one. Full proposal deadlines are now July 27, 2026; November 4, 2026; and March 4, 2027, with annual cycles after that. Phase I proposers still need an invited Project Pitch before a full proposal is reviewed.
Three practical changes matter most for founders scoping applications:
One active project, two pitches, two proposals. A company can have only one Phase I or Fast-Track project under review at a time, may submit at most two Project Pitches in any 12-month window, and at most two full proposals per fiscal year — making early screening and timing strategy more important than in looser prior cycles.
Letters of support are mandatory again. Standard Phase I and Phase II proposals need at least one Letter of Support (Fast-Track needs three), and letters must come from real stakeholders — potential customers, strategic partners, investors — not consultants or subcontractors. NSF wants these letters to validate market demand and reduce non-technical risk.
Post-award scrutiny is real. Successful proposals now face a due diligence process that can include clarification requests, research-security checks, and legal certifications, plus a separate financial capability review for Phase II that can take months and affect final budget.
The bottom line for founders writing a Phase I proposal
A winning NSF SBIR/STTR narrative has to hold up on three fronts simultaneously, not just one:
A technically substantive R&D plan with measurable milestones and quantitative success criteria that clearly targets unresolved technical risk, not assumed-solved engineering.
A broader impacts case showing tangible societal benefit.
A commercialization case with real market opportunity, technical defensibility, and a credible business model — backed by letters of support from actual market stakeholders.
The fastest way to sink an NSF SBIR/STTR application is to pitch routine product development when NSF is looking for high-risk, commercialization-relevant R&D that proves technical feasibility — but that's only one failure mode inside a broader scoring framework that also weighs market, impact, team, and execution. Get the research-vs-development framing right, and you've cleared the entry bar. The proposal still has to win on the other three criteria.
FAQ
Does NSF SBIR/STTR fund development work at all? Yes. NSF's own R&D definition explicitly includes design, development, and improvement of prototypes and new processes. Development isn't disqualifying — development without an unresolved technical-risk question behind it is.
Is the Hazelrigg "Research 101" paper official NSF policy? No. It's NSF-hosted applicant guidance from a former NSF program veteran, but the document itself states the views are the author's and don't necessarily reflect NSF or federal policy.
What are NSF's three merit review criteria? Intellectual Merit, Broader Impacts, and Commercial Impact.
Do NSF SBIR/STTR proposals need letters of support in 2026? Yes. At least one is required for standard Phase I and Phase II proposals, and at least three for Fast-Track. Letters must come from stakeholders like customers, partners, or investors — not consultants or subcontractors.
How many NSF Project Pitches or proposals can one company submit per year? Up to two Project Pitches in any 12-month window and up to two full proposals per fiscal year, with only one Phase I or Fast-Track project under consideration at a time.
What are the 2026 NSF SBIR/STTR full proposal deadlines? July 27, 2026; November 4, 2026; and March 4, 2027, under the consolidated NSF 26-510 solicitation, with annual cycles thereafter.