Most founders compare SBIR programs by dollar amount. NIH pays more than AFWERX, so NIH wins, right? But the SBIR grant vs contract distinction matters more than the award size.
That is because the instrument, not the check, changes how the money behaves. Whether your award is a grant or a contract, and whether it settles fixed-price or cost-reimbursement, decides when you get paid, who audits your books, and how you keep your IP.
Here is the short version.
SBIR is a program, not a single legal instrument. Some agencies award grants (financial assistance for your research). Some award contracts (a purchase of R&D deliverables). NIH and NSF grants and DoD contracts all carry the same "SBIR" label, but they budget, pay, audit, and handle IP differently.
If you are crossing from a civilian grant to a DoD contract, that difference is where founders get surprised. This guide maps it.
Is an SBIR a grant or a contract?
Both, depending on the agency. SBIR runs across 11 federal agencies, and each one picks the legal instrument it uses to fund you.
A grant is financial assistance. The government funds your research because it serves a public purpose, and it is governed by the Uniform Guidance (2 CFR 200). A contract is procurement. The government is buying R&D as a deliverable it will use, and it is governed by the Federal Acquisition Regulation (FAR), plus the DFARS supplement at the Department of Defense.
That legal difference sounds academic. It is not. It sets the rules for everything downstream.
Here is the split for the agencies founders ask about most:
| Agency | Instrument |
|---|---|
| NSF | Grant |
| DOE | Grant |
| USDA | Grant |
| NIH | Both (grant and contract, by topic) |
| DoD (Air Force/AFWERX, Navy, Army, DARPA) | Contract |
| NASA | Contract |
| DHS | Contract |
| DOT | Contract |
Source: SBIR.gov "Contracts vs Grants" tutorial and participating-agency listings. A few agencies mix instruments by program, so confirm the instrument in your specific solicitation rather than assuming from the agency name.
The pattern most first-time applicants hit: your first award is an NIH or NSF grant, your second is a DoD contract, and nobody tells you the rules just changed.
SBIR grant vs contract is only the first axis
The SBIR grant vs contract split is one axis. How the money settles is a second, separate axis, and this is the one that trips up founders who think they already understand grants.
Cost-reimbursement means the government pays your actual allowable costs as you incur them, and you return anything you don't spend. Firm-fixed-price (FFP) means you agree to a set price for defined deliverables. If you deliver for less, you keep the difference. If it costs more, you eat the overrun.
Grants settle as cost-reimbursement. Contracts can go either way. Put the two axes together and you get three real-world combinations:
| Combination | Example | What it means |
|---|---|---|
| Grant + cost-reimbursement | NIH, NSF, DOE Phase I | Draw funds as you spend, return the unspent, no profit from underspending |
| Contract + firm-fixed-price | AFWERX Open Topic Phase I ($75K / 90 days) | Paid on deliverables, keep the underspend, bear the overrun |
| Contract + cost-reimbursement | Many DoD Phase II awards | Invoice actual costs monthly, DCAA-audited, no underspend to keep |
So "I've done an SBIR before" can mean three operationally different things. A founder who ran an NIH cost-reimbursement grant and then wins an AFWERX firm-fixed-price contract has not done the same thing twice.
How the award instrument changes your cash flow
This is the difference founders feel first, because it decides whether you are floating payroll out of your own bank account.
Grant (drawdown). With an NIH or NSF grant, you draw funds as you incur costs, through the HHS Payment Management System (NIH) or Research.gov (NSF). The float is short, because you are pulling cash close to when you spend it.
The catch: unspent funds get de-obligated. You cannot pocket the difference by running lean.
Firm-fixed-price contract (milestone pay). With an AFWERX Open Topic contract, payment follows a deliverable schedule written into the contract, not your incurred costs. You might get paid at kickoff and at final report acceptance, or against defined milestones.
That means you float costs until a deliverable is accepted. If a milestone slips or a deliverable gets kicked back, your payment waits.
The trade-off cuts both ways. Deliver the $75K scope for $68K and the $7K is yours. Spend $82K and you cover the overage.
Cost-reimbursement contract (monthly invoice). With a cost-type DoD contract, common at Phase II, you invoice incurred allowable costs monthly and get paid in roughly 30 days under the Prompt Payment Act. Steadier than milestone pay, but you float 30 to 60 days, and you cannot keep an underspend because you are billing actual costs.
The practical takeaway: a $75K firm-fixed-price contract can put more strain on your cash than a $150K grant, because the grant lets you draw as you go and the contract makes you wait for a deliverable. Size is not the variable. Instrument is.
How the award instrument changes your audit exposure
Founders assume "federal money means audits." True, but the type of audit depends entirely on the instrument, and the two regimes are not close.
Grants: the Single Audit. Grant money falls under the Single Audit (2 CFR 200, Subpart F). Here is the part that surprises people: a Single Audit is only required if your organization expends $1,000,000 or more in federal awards in a single fiscal year. That threshold rose from $750,000 effective October 1, 2024. One SBIR Phase I grant, whether it is $150K or $314K, will not trigger it on its own. Stack several awards in one year and you can cross the line.
Contracts: DCAA. Contract money falls under the Defense Contract Audit Agency (DCAA), and the Single Audit does not apply to FAR procurement contracts at all. Different regime, different rules, different threshold logic.
The intensity depends on the contract type. A firm-fixed-price contract shifts cost risk to you, so the government does less digging into your actual incurred costs. But DoD still expects a compliant setup from the day of award, not from the day of an audit. Based on how Cada structures AFWERX cost volumes, that means:
- An accounting system that tracks costs by contract number and cost category
- Daily timesheets, signed and dated contemporaneously, not reconstructed at invoice time
- Direct costs kept segregated from indirect (overhead, G&A)
- Financial records retained for at least 3 years after final payment
A cost-reimbursement contract goes further. You generally need a DCAA-approved accounting system before award, and you face incurred-cost audits on your actual spending.
The most common Phase I audit problems Cada sees on the contract side are reconstructed timesheets, unsupported indirect rates, and labor billed without documentation. None of that is a concern on a small grant. On a contract, it is the difference between clean and flagged.
How the award instrument changes your IP and data rights
Good news first: you keep your patents either way.
Patent title is the same under grants and contracts. Under the Bayh-Dole Act (35 U.S.C. 200-212), the small business retains title to inventions made under the award. The government gets a nonexclusive license to practice the invention, and it holds march-in rights, but you own the IP. This does not change between a grant and a contract.
Data rights are where the instrument matters. Your SBIR data (the technical data and software you develop under the award) is protected. The government gets a limited license and cannot use your data for competitive procurement or hand it to a competitor during a protection period.
That protection period is now 20 years from the date of award. For DoD contracts, DFARS 252.227-7018 codifies the 20-year period, effective for contracts on or after January 17, 2025. It replaced the old period of roughly 5 years. After 20 years, the government moves to Government Purpose Rights rather than unlimited rights.
The instrument-specific risk is marking. On a DoD contract, DFARS clauses require you to mark your SBIR data with the correct legend. Deliver unmarked data and you can lose the protection you were entitled to. Grants are lighter-touch on marking administration, so a founder who never had to think about legends on an NIH grant can walk into a contract and hand over unmarked deliverables.
The exact clause and period should be confirmed in your specific award, because agencies can negotiate different terms. But the headline holds: same patent ownership, tighter data-rights discipline on the contract side.
The "grants have no profit" myth
One belief worth killing: that grants forbid profit and only contracts pay a fee.
SBIR is one of the only federal grant programs that lets a for-profit company book a fee. NIH and NSF both allow a fee of up to 7% of total project costs on SBIR grants. For NIH, the fee has to be in your budget at the time of application, and it is not treated as a cost for audit-threshold purposes. DoD firm-fixed-price contracts also allow a fee, typically 5 to 7%, built into the fixed price.
So the real difference between the grant and the contract is not fee versus no fee. Both pay you a margin. The difference is how the money flows to you and how your books get checked. That is the reframe: stop asking "which one lets me make money" and start asking "which cash-flow and audit regime am I signing up for."
A founder crossing from NIH grant to DoD contract: what actually changes
Here is a fictional example to make it concrete. The company and details are illustrative, not a real client.
Imagine a robotics startup, "Ridgeline Robotics," that won an NIH SBIR Phase I grant for an autonomous inspection system. The grant funded the research, the team drew funds through the Payment Management System as they spent, and their biggest compliance chore was a progress report. No timesheets scrutinized, no Single Audit (they expended well under $1M that year), no data legends to manage.
A year later, Ridgeline wins an AFWERX Open Topic Phase I contract, $75K over 90 days, to adapt the same system for flightline inspection. Same team, same core technology, completely different rulebook.
What changed on day one:
- Cash flow. No more drawdown. Payment now follows a deliverable schedule, so Ridgeline floats costs until AFWERX accepts a deliverable. They budgeted for the gap instead of assuming they could draw weekly.
- Audit. DCAA rules now apply. Everyone charging the contract keeps daily, contemporaneous timesheets from the start, and costs get segregated by contract number. The spreadsheet that passed for accounting on the grant got upgraded before the first invoice.
- IP. Patent title stayed with Ridgeline (Bayh-Dole, same as the grant). But now every SBIR data deliverable needs the correct DFARS marking, or they risk losing the 20-year protection they are entitled to.
- Budget mechanics. Some SBIR rules carried over unchanged: the 33% subcontractor cap and the option to book a fee apply to grants and contracts alike, so those were not new. What was contract-specific was the mechanics of submission. Ridgeline had to enter the whole cost volume by hand into the DSIP portal web form, because there is no file upload for the cost volume.
None of that showed up on the NIH grant. All of it showed up on the DoD contract. The dollar amount went down. The complexity went up.
What to check before you accept an SBIR award
Run this before you sign, whether it is your first SBIR or your fourth:
- Identify the instrument. Is this a grant or a contract? Check the solicitation, not the agency reputation.
- Confirm how costs settle. Cost-reimbursement or firm-fixed-price? This decides whether underspending helps you or just gets returned.
- Set up timekeeping if it is a contract. Daily, contemporaneous timesheets and cost segregation, from award, not from audit.
- Budget the fee. Up to 7% on NIH and NSF grants, 5 to 7% on DoD FFP. For NIH, it must be in the application budget.
- Plan for the float. Milestone-paid contracts mean you cover payroll until a deliverable is accepted. Model the gap.
- Confirm the data-rights clause and marking. Especially on contracts. Get the legend right or lose the protection.
Before you write a budget for the wrong instrument
If you are moving from an NIH or NSF grant to a DoD contract, the SBIR grant vs contract distinction is not paperwork trivia. It changes when you get paid, who audits you, and how you protect your data. Get the instrument wrong and you can build a cost volume that is non-compliant before a reviewer reads a word of your technical approach.
Cada has written 100+ proposals across 30+ agencies, structuring cost volumes for both instruments, from cost-reimbursement civilian grants to firm-fixed-price DoD contracts. If you are not sure which instrument your target agency uses, or how to structure a budget for it, book a free budget structure review. You get a straight read on the instrument and how to build the cost volume for it. No pitch, no obligation.
Award instruments, audit thresholds, and data-rights clauses change between cycles and can be negotiated in your specific award. Always verify the details against your solicitation and award document before you budget or sign.