SBIR and STTR Grants: Non-Dilutive Capital for Deep Tech

The US federal government runs what is frequently described as the largest seed fund in the world, and a striking number of eligible founders never apply to it.

The Small Business Innovation Research and Small Business Technology Transfer programmes require eleven federal agencies to set aside a percentage of their external research budgets for small businesses. The money is a grant or contract, not an investment. No equity, no board seat, no liquidation preference, no pressure to exit within a fund's life.

The cost is a demanding proposal process, long timelines, and government contracting compliance. For a deep tech, biotech, defense or climate company, the trade is usually excellent.

This guide covers the structure, what the 2026 reauthorisation changed, eligibility, how to actually win, and how to run the award once you have it.

1. The 2026 Reauthorisation — Read This First

The programmes' prior authorisation expired on 30 September 2025, producing a roughly six-month lapse during which agencies could not issue new awards. That lapse was real and disruptive.

The Small Business Innovation and Economic Security Act of 2026 was signed on 13 April 2026, reauthorising both programmes through 30 September 2031.

Material changes to be aware of:

• A new large-award Phase II vehicle aimed at strategic breakthrough technologies, with a ceiling reported at $30 million — far above the traditional Phase II range.
• Increased foreign risk screening, reflecting ongoing concern about foreign ownership, control and influence. Companies with foreign investors, foreign parent entities or foreign-national key personnel should expect closer scrutiny and should prepare for it deliberately.
• Per-company proposal caps from fiscal year 2027, with agencies setting their own limits on submissions per company, per solicitation or per topic. This changes strategy for companies that have historically submitted at volume.

Because the rules have just moved, verify current agency-specific requirements at SBIR.gov rather than relying on any secondary source, including this one.

2. The Phase Structure

Phase I — feasibility

Establishes technical merit and feasibility. Typically six to twelve months, with award sizes varying by agency from tens of thousands up to a few hundred thousand dollars. Some agencies now run direct-to-Phase-II routes for companies that can demonstrate feasibility has already been established.

Phase II — development

The substantial money. Typically around two years, with awards commonly in the high six figures to low seven figures, and considerably more at agencies with larger budgets. Only Phase I awardees can generally proceed, absent a direct-to-Phase-II mechanism. Several agencies offer supplements and matching programmes that increase the total.

The new strategic breakthrough vehicle introduced in 2026 sits above this range and is worth understanding if your technology is genuinely at that scale.

Phase III — commercialisation

No SBIR funds. Phase III refers to work that derives from the earlier phases, funded from elsewhere — including sole-source federal contracts, which is the part founders most often miss. A Phase III sole-source authority means an agency can contract with you directly, without competition, for work deriving from your SBIR. For defense and government-adjacent companies this is frequently worth more than the grant funding itself.

3. SBIR Versus STTR

The difference is the research partnership.

• SBIR: the small business performs the majority of the work. Subcontracting is permitted within limits that vary by phase.
• STTR: requires formal collaboration with a research institution — a university, federally funded research centre or non-profit research organisation — with minimum work percentages allocated to each party.

STTR suits technology emerging from a university lab where the academic team remains involved. SBIR suits a company with in-house capability. The intellectual property allocation between company and institution must be agreed before submission, and getting it wrong creates problems that surface later in investor diligence.

4. The Agencies, and Why the Choice Matters

Eleven agencies participate, and they behave very differently — in award size, cadence, review culture and what they want.

• Department of Defense — the largest by volume. Topic-driven: they publish specific problems and you propose solutions. Strong path to procurement through Phase III. Multiple solicitation cycles a year.
• NIH (HHS) — the largest for life sciences, with among the highest award ceilings. Investigator-initiated: you propose what you want to work on. Peer-reviewed on scientific merit, with standard receipt dates and a resubmission culture.
• NSF — broad technology scope, strong emphasis on commercialisation potential and customer discovery. Uses a project pitch process before full proposal, which is an efficient way to test fit early.
• Department of Energy — energy, climate, materials, advanced computing. Topic-driven, with a letter of intent stage.
• NASA — space, aeronautics, sensing, materials.
• Others — USDA, DHS, DOT, EPA, Department of Education, and Commerce agencies including NIST and NOAA.

The single most common failure is proposing to the wrong agency, or to a topic that does not genuinely match the work. Talk to the topic author or programme manager before writing. At most agencies this contact is permitted and encouraged during defined windows, and it is the highest-return hour in the whole process.

5. Eligibility

• US-based small business, organised for profit, performing the work substantially in the United States
• 500 or fewer employees, including affiliates — the affiliate rules catch companies with common investors and are more expansive than founders expect
• Ownership requirements — broadly, majority US ownership by individuals or other qualifying small businesses. Certain agencies permit majority ownership by multiple venture capital, hedge or private equity funds under specific provisions, but this is agency-elective and conditional.
• Principal investigator employment — the PI must be primarily employed by the company for SBIR; STTR is more flexible, which is precisely why university spinouts use it
• Registrations — SAM.gov, a Unique Entity ID, SBIR.gov company registry, and agency-specific systems. Start these early; they routinely take weeks and have blocked more submissions than weak science ever has.

If you have foreign investors, a foreign parent, or foreign-national key personnel, address this deliberately given the tightened screening. It is a solvable problem but not one to discover a week before a deadline.

6. Writing a Proposal That Gets Funded

• Answer the topic, literally. Reviewers score against stated criteria. A brilliant proposal that does not address the topic loses to a competent one that does.
• Talk to the programme manager first. Within the permitted window, ask what problem they are actually trying to solve and what a strong response looks like.
• Lead with the technical innovation. What specifically is new, and why has nobody done it? Reviewers are technical and are unimpressed by market language.
• Define measurable milestones. Specific technical objectives with quantified success criteria. Vague deliverables score poorly and are easy to identify.
• Take commercialisation seriously. Agencies increasingly weight it heavily. Name customers, describe the path to revenue, show evidence of demand. NSF in particular expects genuine customer discovery.
• Follow the formatting rules exactly. Page limits, fonts, margins and section ordering. Non-compliant proposals are rejected without review, every cycle, in volume.
• Expect to resubmit. First-time success rates are low, reviewer comments are detailed and genuinely useful, and a revised resubmission is a normal and effective path. Treat rejection as free consulting.

7. Running the Award Once You Have It

Winning is roughly half the work, and the half founders prepare for. The other half is administering a federal award without creating problems for yourself, and it starts before the money arrives.

Set up the accounting before you spend anything. Federal awards require costs to be tracked by project, with a defensible basis for allocating indirect costs and time. Retrofitting this from a general ledger that mixed everything together is painful and is exactly what an audit finds. Talk to an accountant who has done government work — this is a specific competence, not general bookkeeping.

Understand allowable costs. Federal cost principles determine what you can charge, and some ordinary startup expenses are not allowable — certain marketing, entertainment, lobbying and some financing costs among them. Charging an unallowable cost is not fraud, but it is a finding, and findings compound across an audit.

Track time honestly. Personnel is usually the largest line, and the supporting evidence is timekeeping records. Reconstructed timesheets are the single most common audit problem in small business federal awards. Have people record time contemporaneously against the project.

Watch the cash flow. Many awards reimburse after cost is incurred, so you are funding the work and waiting. Build that into runway planning rather than assuming the award is cash on day one. Some agencies offer advance payment arrangements; ask.

Report on time and report problems early. Technical and financial reports have deadlines, and programme managers deal with technical difficulty constantly — what they respond badly to is discovering a problem at the final report. A milestone that will not be met is a conversation, not a catastrophe, if it happens early enough.

Report inventions, and elect title. Under the Bayh-Dole framework you keep title to subject inventions, but only if you report them and elect title within the required deadlines. Missing those deadlines can cost you ownership of your own invention, which is an entirely avoidable and genuinely serious outcome.

Mark your data correctly. SBIR data rights protect your technical data for a defined protection period, and the protection depends on the deliverables being marked with the correct legend. Unmarked data can lose protection. This is mechanical, it takes minutes, and companies skip it.

Plan Phase II during Phase I. The gap between phases is where companies run out of money. Start the Phase II proposal well before the Phase I work ends, and keep the programme manager informed of progress throughout — they are the person who will be reading it.

8. What It Costs You

The money is non-dilutive, but it is not free.

• Time. A serious proposal takes weeks of senior technical effort. Award decisions can take six months or more.
• Compliance. Government contracting brings accounting requirements, allowable-cost rules, reporting, and potentially audit. Your accounting setup may need adjustment.
• Direction risk. The topic shapes what you build. A grant that pulls you toward a government use case away from your commercial market is a real cost, and it is how some companies quietly become contract research organisations.
• Cash flow. Many awards reimburse costs rather than paying in advance, so you fund the work first.
• Data rights. SBIR data rights protect your technical data from disclosure and unlimited government use for a defined protection period. The rules have been revised in recent years — understand precisely what applies to your award and mark your deliverables correctly.

Frequently Asked Questions

Does taking SBIR money hurt our ability to raise venture capital?

No — the opposite, generally. Investors read a competitive federal award as independent technical validation, and every dollar of grant funding is a dollar of runway that did not cost equity. The exception is a company whose entire revenue is grants: investors distinguish sharply between grant-funded product development and a business model that is grant dependency.

Can venture-backed companies apply?

Yes, subject to ownership rules. Majority ownership by multiple qualifying funds is permitted at agencies that have elected to allow it, under specific conditions. Ownership by a single fund, foreign entities or large corporations is more likely to cause a problem. Check the current rules for the specific agency before assuming eligibility.

Do we keep the intellectual property?

Generally yes. Under the Bayh-Dole framework the small business retains title to subject inventions, subject to government rights including a licence for government purposes and march-in rights that are rarely exercised. Compliance with invention reporting and election-of-title deadlines is mandatory — missing them can cost you title.

How much does grant writing help cost?

Consultants charge either a flat fee or, less commonly, a success fee. Good ones improve odds meaningfully, particularly on your first application. Note that a percentage-of-award success fee may raise allowability questions under federal cost principles — confirm the arrangement is compliant before signing.

What other non-dilutive funding exists?

Considerably more than most founders realise: state and regional innovation programmes, agency prize competitions, ARPA-style programmes, and R&D tax credits, which can offset payroll tax for qualifying small businesses and are frequently unclaimed.

Can we apply to several agencies at once?

Yes, and many companies do — but note the per-company proposal caps arriving from fiscal year 2027, and be aware that you cannot receive duplicate funding for the same work. Agencies check. The strategy that works is proposing genuinely different applications of your technology to agencies whose missions differ, not the same proposal reformatted. If two awards would fund overlapping work, disclose it and let the agencies resolve it rather than staying quiet.

How long does it take to get paid after an award?

Longer than founders plan for. Between notification and first payment there is contracting, registration verification and invoicing setup, which commonly adds one to three months, and most awards then reimburse in arrears. A company that treats an award notice as immediate cash relief can find itself funding several months of work from its own balance sheet. Build that gap into the runway model explicitly.

What happens if the research does not work?

Generally nothing adverse, provided you performed the work in good faith and reported honestly. These are research awards and negative results are a legitimate outcome — agencies fund high-risk work knowing some of it fails. What creates problems is misrepresenting results, or continuing to bill against a project you have effectively stopped. Tell the programme manager, document the findings, and close the award properly.

Does an SBIR award help us win defense contracts?

Materially, yes, and this is the most under-appreciated part of the programme. The Phase III sole-source authority allows an agency to contract with you directly for work deriving from your SBIR, without competition — which in a procurement environment built around competition is a substantial advantage. Companies that treat SBIR purely as research funding and never pursue Phase III leave the most valuable part unused. Our guide to defense tech funding covers the procurement path.

The Bottom Line

For a deep tech, biotech, defense or climate company, SBIR and STTR are the cheapest capital available. The programme is authorised through September 2031, the award ceilings have risen, and the screening has tightened.

Pick the right agency, speak to the programme manager before you write, answer the topic literally, and plan to resubmit. Set the accounting up properly before you spend, and start the registrations now, because they will take longer than you think.

Global Capital Network connects deep tech founders with investors who understand grant-funded development. See upcoming events or get in touch.

Programme rules changed materially in 2026 and vary by agency. Verify current requirements at SBIR.gov and with the relevant agency before relying on any of this.