The Four SBIRs: A Pathway Framework for Choosing the Right Federal Innovation Program

If you're a deep-tech founder staring at NSF SBIR, AFWERX, DIU, ARPA-H, and DARPA, the question you're really asking is: which federal innovation program is right for my startup? Here's the short answer. Federal innovation programs serve four different missions -- Public Seed, Government Adoption, Dual-Use Scale, and National Spillover. Pick the program whose success metric you can actually hit. Topic-fit gets you in the door. Pathway-fit gets you funded.

That sentence is the whole guide. The rest explains how to classify yourself, why the wrong choice is so expensive, and how to map your company to the right pathway before you spend 60 hours on an application.

"Which federal innovation program has money for my field?" is the wrong first question

Most founders start here: "I'm building an AI diagnostics tool, who funds health AI?" That question returns a list. NIH SBIR funds it. ARPA-H funds it. AFWERX funds it if you can frame a defense use case. NSF funds it. Now you have five tabs open and no decision.

Topic-fit is necessary. It is not sufficient. Every one of those programs would accept your application. Only some of them reward what you can actually deliver.

Here is the failure mode that costs founders the most. You win a program whose success metric you will never hit.

A concrete version: a commercial-first company chases the DoD SBIR path because AFWERX Phase II awards dwarf an NSF Phase I. They win a $150K Phase I and build a demo. Then Phase II asks: who is your government customer? They don't have one. They never built the relationship, because they were optimizing for the grant, not the customer. The award was real. The pathway was wrong. The company is now 14 months down a road that ends at a wall.

We see this constantly. In a review of 232 companies in our pipeline, the single largest concrete pain pattern was wrong-program fit. Roughly 52 companies showed it -- about 26 disqualified at intake for chasing a program that didn't match their model, plus a cluster of companies whose roadmaps spanned two or more agencies because no single program fit cleanly. The problem appears at every stage of the funnel: before the first call, at disqualification, and inside delivered roadmaps. (Source: internal Cada pipeline analysis, 2026. Counts are approximate and aggregated.)

What is pathway-fit? Pathway-fit means the program's definition of success matches what your company can realistically achieve in the next 24 months. A program is a pathway-fit when its real customer, its timeline, and its success metric line up with your business -- not just when its funding topic matches your technology.

The four pathways, defined

Federal innovation programs are trying to do four genuinely different things. They use overlapping vocabulary -- everyone says "innovation," "commercialization," "transition" -- which is exactly why founders conflate them. Here is what each one is actually built to do.

Pathway A: Public Seed

Mission: de-risk early-stage R&D for a future commercial market.

The government is your funder, not your customer. It is buying down the technical risk of something it hopes the private market will eventually buy. Nobody at NSF expects to purchase your product.

Real customer: a future private or commercial buyer.

Programs that live here: NSF SBIR, NIH SBIR, DOE SBIR.

Timeline reality: long. The research-to-product arc runs in years, and there's a 9-18 month cash gap between Phase I and Phase II that has killed companies that "won" SBIR. NSF and NIH SBIR Phase I awards both top out near the SBA statutory cap of about $314K (before approved waivers), over 6-12 months.

Success metric that wins: commercialization. Follow-on private capital, commercial revenue, a product in market.

Wrong metric founders overuse: number of grants and phases won. Stacking awards looks productive and signals mill behavior to reviewers. A pile of Phase I awards is not a company.

Pathway B: Government Adoption

Mission: get a specific government office to buy and field your technology.

This is not about research. It's about procurement. The program exists to move your product from a prototype into the hands of an actual government user with a budget.

Real customer: a named government end user with a budget line -- a wing, a squadron, a program office, a command. Not "the Air Force." A specific office.

Programs that live here: DoD SBIR via AFWERX, Army xTech, Navy programs, DIU Commercial Solutions Openings, SOFWERX.

Timeline reality: medium, but gated by your customer relationship, not the grant calendar. The hard part isn't winning Phase I or II. It's the Phase III transition, which depends on a customer signal you have to develop yourself. DIU's "60-90 day" reputation is optimistic; real award timelines have run closer to 142-197 days (per published DIU award-cycle analyses).

Success metric that wins: a signed government customer. A Phase III contract, a program of record, a sole-source justification.

Wrong metric founders overuse: Phase I/II award count and letters of support. A letter of support costs a colonel ten minutes and means nothing at transition. The signal that matters is a government customer memorandum from a named official with a non-SBIR budget line.

Pathway C: Dual-Use Scale

Mission: scale a company that sells to both commercial and government buyers, using matched private capital.

This pathway assumes you already have traction and a real dual-use product. It's a bridge, not a seed. Programs here multiply private capital rather than replace it.

Real customer: both a commercial market and a government buyer, plus a private investor putting up the match.

Programs that live here: AFWERX STRATFI/TACFI, In-Q-Tel, DIU scaling vehicles.

Timeline reality: medium-long, and gated by your ability to commit matching capital. STRATFI can stack government and private money into a bridge in the $15M range, but only if you've lined up the match. No match, no deal.

Success metric that wins: capital efficiency and scaled deployment. How many private and government dollars did each SBIR dollar pull in, and did the product actually scale across both markets?

Wrong metric founders overuse: treating it like a grant. STRATFI is not free money you spend down. It rewards capital efficiency. If your only plan is to absorb the government tranche, you've misread the program.

Pathway D: National Spillover

Mission: build a breakthrough capability that shifts an entire field. The outcome is transformation, not a single product.

These programs fund things that might not work, on purpose. They want a moonshot, and they're staffed by program managers with the authority to fund you fast and kill you faster.

Real customer: the program manager's vision, and the national or field-level capability they're chartered to create.

Programs that live here: DARPA, the transformational end of ARPA-H, large NSF research efforts.

Timeline reality: long, high-variance, and subject to kill authority. A program manager can end the work mid-stream if a milestone slips. Term-limited PMs want results inside their tenure.

Success metric that wins: a transformational capability, a spinout ecosystem, a field that moved because you existed. DARPA PMs think in the Heilmeier Catechism: what are you trying to do, why is it hard, why will you succeed where others failed.

Wrong metric founders overuse: near-term revenue and incremental milestones. If you pitch a product with a tidy roadmap to a PM who wants a breakthrough, you lose. They're not buying a product. They're buying a leap.

Who reads your application changes what you write

The four pathways aren't just funded differently. They're judged by different people, and that changes everything about how you write.

Pathway A runs on peer review. NIH study sections are panels of 15-20 scientists who score you 1-9 on criteria like significance and approach. NSF SBIR uses external reviewers plus a program director. You're writing for a committee of experts, so the application has to survive the most skeptical reviewer in the room. Rigor wins.

Pathway B is closer to a sales process than a review. AFWERX and DIU care about whether a government user wants this and whether you can deliver. The "reviewer" is partly an evaluator and partly a future customer. You're writing to prove demand and feasibility, not scientific novelty.

Pathway C is evaluated by people asking one question: will private capital actually show up? STRATFI/TACFI reviewers want to see committed matching funds and a credible scaling story. A brilliant technology with no investor on the hook fails here.

Pathway D is a single program manager with strong opinions and the authority to act on them. You're writing for one expert who wants a breakthrough and will apply the Heilmeier Catechism. There's no committee to average out one skeptic. You have to convince the one person who matters.

If you write a peer-review-style application (dense, rigorous, citation-heavy) for a Government Adoption program, you'll bore the evaluator who wanted to see customer demand. If you write a sales-style application for an NSF panel, the scientists will mark you down for hand-waving the technical risk. Matching your writing to the reader is downstream of matching your company to the pathway.

The four pathways at a glance

How to classify yourself: a five-question diagnostic

You don't need a consultant to find your primary pathway. You need five honest answers.

1. Who is the first paying customer you can name today? A commercial buyer points to A. A specific government office points to B. Both, with real traction in each, points to C. Nobody yet because it's a capability, not a product, points to D.

2. What does winning look like in 24 months? Commercial revenue or a priced round is A. A government contract is B. Matched private capital plus dual-use revenue is C. A working breakthrough is D.

3. Can you survive the 9-18 month funding gap between phases? If only with bridge capital you haven't lined up, flag a runway risk before you commit to any phased program. This question kills more SBIR companies than rejection does.

4. Is your technology genuinely dual-use, or are you reverse-engineering a defense angle to chase money? Genuinely dual-use keeps C and B open. Reverse-engineered is a mismatch risk -- reviewers and program officers see through a bolted-on defense story fast.

5. Do you need money, or do you need something money can't directly buy? If you need cash, that's A. If you need a customer, that's B. If you need scale capital, that's C. If you need a capability or breakthrough you can't build alone, that's D.

If your answers point to one pathway, you have your primary. If they point to two, you're not stuck -- you're a sequencing problem, which the next section covers. The rule: pick the pathway whose success metric you can actually hit in the next 24 months, and treat the others as later moves.

Frankly, the messy cases are common. Plenty of strong companies straddle A and B, or B and C. That's fine if you sequence them and keep the metrics separate. It's dangerous when you blur them -- when you run a Government Adoption application while privately measuring yourself on commercial revenue, you end up serving neither.

The most common mismatches (and how to fix them)

These examples are fictional and illustrative. The patterns behind them are real.

Mismatch 1: Commercial-first SaaS chasing DoD SBIR because the checks are bigger. A team building workflow software for hospitals notices AFWERX Phase II awards dwarf NSF Phase I. They frame a thin defense logistics angle, win Phase I, and build a demo nobody in the government asked for. At Phase II transition there's no customer, because there was never a customer relationship. Fix: this is a Pathway A company. NSF or NIH SBIR, or commercial capital, matches their real customer.

Mismatch 2: A genuine defense-adoption play treating an NSF grant as the goal. A team with a real shot at a Navy fielding decision wins an NSF SBIR because it was the easiest application. They spend a year optimizing for an academic-leaning reviewer, build the version that wins the grant, and never develop the program office relationship that would have produced a customer. Fix: this is Pathway B. The work should have started with the customer signal, not the grant.

Mismatch 3: A founder pitching DARPA a product with a milestone chart. A strong engineering team applies to a DARPA program with a clean Gantt chart and revenue projections. The program manager wanted a breakthrough and a defensible answer to "why will you succeed where others failed?" The polished product plan reads as incremental. Fix: either reframe as Pathway D with genuine breakthrough framing, or step back to Pathway A and let the technology mature first.

The thread through all three: the founder optimized for the application instead of the pathway's actual definition of success.

There's a fourth mismatch worth naming because it's the quietest. A company on the right pathway, measuring the wrong metric inside it. Picture a Pathway B company that wins three Phase I awards across three services and treats the award count as traction. Investors aren't impressed by three Phase I grants and no government customer. The pathway was right. The internal scorecard was wrong, and it took a year of board meetings to notice. Every pathway has a wrong-metric-to-overuse for exactly this reason -- the seductive number that feels like progress and isn't.

The cash gap is part of the decision, not a detail

One more thing that should shape your pathway choice up front: the funding gap.

Phased programs don't pay continuously. Between a Phase I award and a Phase II award, there's commonly a 9-18 month gap where no new federal money arrives. A Phase I award of roughly $275K-$314K covers maybe six months of runway for a small team burning $50K-$100K a month. Do that math before you apply, not after.

This matters for pathway choice because the pathways carry different gap risk. Pathway A and B both run through phased SBIR and expose you to the gap. Pathway C exists partly to bridge it. Pathway D programs are often structured as longer single efforts. If you have under 12 months of runway and no bridge plan, that's a reason to either line up bridge capital first or rethink which pathway you start on.

Sequencing across pathways: the portfolio move

Here's what strong deep-tech companies actually do. They don't pick one pathway and stay there. They sequence.

A common arc is A to B to C. Start with a Public Seed grant to de-risk the core technology. Use the credibility and the working prototype to develop a Government Adoption customer. Once you have dual-use traction, use a Dual-Use Scale vehicle to pull in matched capital and grow across both markets.

That's a multi-year portfolio, not a single bet. The companies that struggle are the ones treating each application as an isolated lottery ticket.

Two rules make sequencing work.

First, separate the metrics. When you're running a Pathway A application, measure it on commercialization potential. When you move to Pathway B, measure on customer development. Don't grade a B application on A's scorecard.

Second, don't run two pathways at full intensity at the same time unless you have the team to staff both narratives honestly. A blurred dual-pathway pitch is weaker than a sharp single-pathway one.

This is the heart of the portfolio approach. Most grant advice treats programs one application at a time. The real edge is in mapping a sequence across pathways, so each win sets up the next, instead of betting everything on one program and one reviewer's mood.

Frequently asked questions

Which federal innovation program is right for my startup? It depends on your pathway, not your technology field. Classify yourself first: Public Seed (NSF, NIH) if your customer is a future commercial market; Government Adoption (AFWERX, DIU) if your customer is a government office; Dual-Use Scale (STRATFI) if you sell to both with matched capital; National Spillover (DARPA) if you're building a breakthrough.

What's the difference between NSF SBIR, DoD SBIR, and DIU? NSF SBIR is Public Seed: it de-risks R&D for a commercial market and never buys your product. DoD SBIR (AFWERX, Army, Navy) is Government Adoption: the goal is a government customer who fields your technology. DIU is also Government Adoption but uses faster commercial-style contracting (CSOs and Other Transaction Authority) instead of the SBIR process.

Is DARPA an SBIR program? No. DARPA runs its own programs aimed at breakthrough capabilities, evaluated by individual program managers rather than peer-review panels, though DARPA components do issue some SBIR topics. Treat core DARPA work as National Spillover: it funds field-shifting research, not product development, and the program manager can start or stop the work directly.

Can I switch pathways after winning a grant? Yes, and many companies do as they mature. The risk is switching by accident -- winning a Pathway A grant and then behaving as if you're on Pathway B without developing the customer. Switch deliberately, and reset your success metric when you do.

Should I apply to a program if I don't have a government customer yet? For Public Seed programs (NSF, NIH), no customer is fine -- the government is your funder. For Government Adoption programs (AFWERX, DIU), applying with no customer and no plan to develop one is the most common expensive mistake. Develop the customer signal first, or start on Pathway A.

Find the right federal innovation program before you write a word

The most expensive grant application is the one you win on the wrong pathway. You spend the hours, you get the check, and then the program's definition of success diverges from your company's -- and you find out 14 months too late.

So before you ask which federal innovation program is right for your startup, answer the pathway question first. If you're not sure which pathway fits, that's exactly the thing to settle before investing 40 to 80 hours in an application.

We do a free pathway-fit assessment that gives you a straight answer: your primary pathway, whether to apply at all to a given program, and which programs to sequence over the next two years. No pitch, no obligation. If your strongest move is a portfolio across pathways rather than a single program, our Grant Roadmap Builder maps that sequence -- because winning the first grant is easier when it's the right one.

Questions about where your company fits? Happy to walk through any of this.