Government Grants for Tech Startups: The Complete Sector-by-Sector Map

Over $4 billion in federal grant funding is available to tech startups every year through the SBIR program alone. Add DARPA, ARPA-H, ARPA-E, BARDA, and other non-SBIR programs, and the total grows significantly. The problem isn't the money -- it's knowing which programs fund your type of technology.

Most "government grants for startups" content lists the same 5-6 programs and calls it a day. This guide maps the full landscape sector by sector -- every relevant agency, what they fund, how much they award, and where to start if you've never applied.

The sector map: where your technology fits

Every tech sector has a different grant landscape. Some (biotech, defense) have multiple overlapping programs worth billions. Others (fintech) have almost nothing. This table shows the primary funding path for each sector, then we break each one down.

Now the sector-by-sector breakdown.

AI and machine learning

AI startups have multiple federal entry points because AI is a cross-cutting priority for nearly every agency.

NSF SBIR is the default starting point. NSF has a dedicated AI topic area covering deep learning, NLP, computer vision, robotics, and AI infrastructure. Phase I awards $305,000. Phase II awards up to $1.25M with supplemental funding potential exceeding $500K. NSF uses a rolling Project Pitch model -- submit a 4-page pitch anytime, get a response in about 3 weeks, and only write a full proposal if invited. No preliminary data required.

DARPA is where the big money lives. The Information Innovation Office (I2O) publishes a standing BAA covering trustworthy AI, explainable AI, resilient software, and cybersecurity. Award amounts vary from $500K to multi-millions per performer. DARPA is harder to win -- they want genuinely novel approaches to hard problems, not incremental improvements -- but the funding and prestige are unmatched for defense-relevant AI.

Other paths worth knowing:

• NIST SBIR: AI safety, measurement science, and standards-related AI. Smaller awards ($100K Phase I, $400K Phase II) but a much smaller applicant pool means lower competition.
• IARPA BAAs: Intelligence-community AI applications. $500K-$5M per performer. May require security clearances.
• NIH SBIR: Clinical AI, health data analytics, diagnostic algorithms. Apply under the relevant NIH institute (NCI for cancer AI, NIMH for mental health AI, etc.).

Where to start: If you're building commercial AI with broad applications, start with NSF's Project Pitch. If your AI solves a defense or intelligence problem, start with DARPA's I2O BAA. If it's health-focused, go to NIH.

Biotech and healthtech

This is the richest sector for federal grants. NIH alone distributes roughly $1.2 billion in SBIR funding annually, and ARPA-H and BARDA add billions more through non-SBIR mechanisms.

NIH SBIR/STTR is the backbone. Each of NIH's 27 Institutes and Centers (NCI, NIAID, NHLBI, NIMH, etc.) runs its own SBIR topics. Phase I awards up to $314,363 (the SBA cap). Phase II awards up to $2,095,748. The Commercialization Readiness Pilot extends that to $4.19M over 3 years.

Key differences from other agencies: PI citizenship is NOT required. Preliminary data is critical for competitiveness (unlike NSF). Individual institutes may set lower budget caps. Success rates vary by institute (13-24% for Phase I). Match your technology to the right institute -- a diagnostic for cancer goes to NCI, a mental health platform goes to NIMH, an infectious disease tool goes to NIAID.

ARPA-H is NIH's more aggressive sibling. Launched in 2022, ARPA-H funds health moonshots through program manager-directed investments. SBIR Phase I goes up to $600,000 (nearly double NIH's cap). Phase II up to $3.5M. Non-SBIR programs can reach $50M+ per performer. The bar is high -- ARPA-H wants 10x improvements, not incremental advances. If your technology could fundamentally change how a disease is detected, treated, or prevented, this is where to look.

BARDA DRIVe funds medical countermeasures and health security technology. Their EZ-BAA (rolling, up to $750K) is one of the fastest paths to federal funding for startups in pandemic preparedness, antimicrobial resistance, diagnostics, and biodefense. Accelerator programs (I-CREATE, VANGUARD) start at $50K-$200K with milestone-based progression.

Where to start: For most healthtech startups, NIH SBIR at the institute matching your therapeutic area. If your technology is truly breakthrough (not incremental), also explore ARPA-H. If you're in medical countermeasures or pandemic preparedness, BARDA DRIVe's EZ-BAA is worth a look.

Cleantech and energy

DOE SBIR/STTR is the primary path. DOE publishes 60+ technical topics across energy production, grid technology, nuclear, fossil energy, and environmental management. Phase I awards up to $200,000. Phase II up to $1.1M. DOE typically releases solicitations in two annual cycles. They award roughly 400 Phase I and 200 Phase II grants annually from a $315M+ budget.

ARPA-E is DOE's moonshot arm. Awards average $2-3M over 3 years, with SPARKS awards starting at $500K for earlier-stage concepts. If you're building truly transformational energy technology -- new battery chemistries, grid-scale storage, novel solar architectures, fusion-adjacent concepts -- ARPA-E is the most ambitious funder in the space. But it's also the most competitive (2-5% acceptance rate).

EPA SBIR is the smallest SBIR program by award size ($100K Phase I, $400K Phase II) but covers environmental technology that DOE doesn't -- clean water, air quality, PFAS remediation, circular economy. Lower competition and a good fit for startups with environmental remediation applications.

A note on DOE's Loan Programs Office: The LPO offers loan guarantees for commercial-scale clean energy deployment ($62B in available authority). This is for later-stage companies deploying proven technology, not early-stage R&D. Not suitable for seed or Series A startups, but worth knowing about for your growth trajectory.

Where to start: DOE SBIR for most cleantech R&D. ARPA-E if your technology is genuinely breakthrough. EPA if your focus is environmental remediation.

Defense and cybersecurity

Defense has the most entry points of any sector. The Department of Defense spends roughly $2.3 billion on SBIR annually across multiple branches, plus billions more through DARPA BAAs, OTAs, and DIU.

AFWERX (Air Force) is the fastest and most accessible DoD entry. Their Open Topic SBIR accepts any defense-relevant technology -- no specific topic match required. Phase I is fast but small ($75K, 3 months). Phase II jumps to $1.25M over 21 months. STRATFI and TACFI post-Phase II programs can extend funding to $3-15M. AFWERX is where most first-time defense tech startups should start.

Navy SBIR covers maritime, undersea, naval aviation, and cyber topics. Phase I up to $256K, Phase II up to $1.7M. Specific topics published through defensesbirsttr.mil.

Army SBIR covers ground systems, communications, soldier technology, and network modernization. Same award range as Navy. The Army also runs xTechSearch -- a separate competition pathway that combines prizes with SBIR Phase II awards.

DARPA publishes office-wide BAAs through its Information Innovation Office (cyber, AI), Strategic Technology Office (large-scale systems), Defense Sciences Office (fundamental science), and Microsystems Technology Office (hardware). Award amounts vary by proposal. DARPA targets breakthrough capabilities -- if you're building something genuinely new for defense, this is where the biggest awards and strongest validation come from.

DHS SBIR covers border security tech, critical infrastructure protection, cybersecurity, and chemical/biological defense. Phase I up to $175K, Phase II $1-1.5M. Smaller program than DoD but less competitive.

Other paths:

• DIU (Defense Innovation Unit): Not a grant -- a commercial procurement pathway. DIU buys commercial technology for military use through OTAs. $100K-$50M+, with 60-90 day target from problem to prototype award.
• OTAs (Other Transaction Agreements): Government contracts that bypass traditional procurement rules. Faster and more startup-friendly than standard federal contracts.

Where to start: AFWERX Open Topic for your first defense application. Navy or Army SBIR if your tech maps to a specific topic. DARPA if you're building something genuinely novel. DIU if you have a working commercial product with defense applications. See our AFWERX SBIR guide for a deep dive on the Air Force pathway.

Agriculture and food tech

USDA NIFA SBIR is the primary (and essentially only) dedicated program. Phase I awards $125K-$175K over 8-12 months. Phase II up to $600K over 24 months. Topics include plant and animal production, food safety, bioprocessing, precision agriculture, and bioenergy.

The award amounts are notably smaller than other agencies. A USDA Phase II ($600K) is less than most agencies' Phase I. If your ag tech has broader applications -- say, AI-driven sensing that works for both agriculture and environmental monitoring -- consider applying to NSF or DOE instead of (or alongside) USDA.

FFAR (Foundation for Food & Agriculture Research) occasionally funds for-profit companies, but requires 1:1 matching funds and historically favors academic institutions. Pre-seed startups that can't meet the matching requirement should skip this.

Where to start: USDA NIFA if your technology is specifically agricultural. NSF SBIR if the underlying technology (sensors, AI, biotech) has applications beyond farming.

Space and aerospace

NASA SBIR covers autonomous systems, in-space manufacturing, lunar surface tech, Earth observation, propulsion, and materials for extreme environments. Phase I awards $150,000, Phase II $850,000. NASA is restructuring its SBIR from a single annual solicitation to a BAA model with multiple submission windows throughout the year -- more opportunities to apply.

NASA awards are smaller than DoD, but NASA is a strong brand signal for space startups. The agency's SBIR alumni include several companies that went on to win major NASA and commercial contracts.

SpaceWERX (Space Force) runs SBIR through the AFWERX model. Phase I $75K, Phase II $1.25M. Topics include rapid-launch spacecraft, on-orbit servicing, space domain awareness, and satellite communications. STRATFI post-Phase II funding ($3-15M) is available for successful performers.

Where to start: NASA SBIR for space science and exploration technology. SpaceWERX for defense-relevant space capabilities. Both accept dual-use technology.

Advanced manufacturing

DOE SBIR covers industrial decarbonization, critical materials processing, and energy-efficient manufacturing through its Advanced Manufacturing Office topics. Phase I $200K, Phase II $1.1M.

NIST SBIR covers manufacturing measurement, additive manufacturing standards, and smart manufacturing sensors. Smaller awards ($100K Phase I, $400K Phase II) but the smallest SBIR applicant pool -- lower competition for a narrower scope.

DoD ManTech funds manufacturing innovation through the Manufacturing USA network (America Makes for additive manufacturing, LIFT for lightweight metals, etc.). Awards range from $500K to $10M+ but typically require partnership with a Manufacturing Innovation Institute. Small businesses can participate as partners or subcontractors.

Where to start: DOE SBIR if your manufacturing technology has an energy angle. NIST if it involves measurement or standards. DOE or DoD ManTech for larger awards, though these typically require partnerships.

EdTech

NSF SBIR is the stronger path for edtech startups. The Learning and Cognition Technologies topic covers AI in education, workforce development, adaptive learning, and STEM education platforms. Phase I $305K, Phase II $1.25M. Rolling submissions, broader scope than the alternative.

ED/IES SBIR (Department of Education, Institute of Education Sciences) is smaller and more restrictive. Phase I $250K, Phase II $1M. Only about 10 Phase I and 5 Phase II awards per year. Must focus on PreK-12 or postsecondary education technology and include an evaluation component. Phase II requires a prior Phase I award.

Where to start: NSF SBIR unless your technology is specifically focused on K-12 or postsecondary education outcomes (in which case, consider both).

Fintech

There are no dedicated federal grant programs for financial technology. Treasury, OCC, FDIC, SEC, CFPB, and SBA do not participate in SBIR and don't run grant programs for startups.

This is a real gap in the federal funding landscape. The workaround: position your underlying technology rather than your application domain.

• If your fintech uses novel AI/ML, apply to NSF SBIR under Artificial Intelligence
• If it involves cybersecurity innovation, try NSF or DHS SBIR under cybersecurity topics
• If it addresses financial inclusion for underserved communities, NSF's Social and Behavioral Sciences topic may be a fit

The reframing isn't a stretch for most fintech companies -- the technical innovation often IS the AI model, the security architecture, or the data infrastructure, not the financial services wrapper. But be honest about the fit. If your innovation is primarily a business model or user experience, SBIR probably isn't the right path.

Transportation and mobility

DOT SBIR covers all modes -- highway safety (FHWA), aviation (FAA), transit (FTA), maritime (MARAD), and pipeline safety (PHMSA). Phase I up to $200K, Phase II up to $1.5M (one of the higher Phase II amounts). DOT is a smaller SBIR agency but has less competition than DoD or NIH.

FAA topics within DOT SBIR are particularly relevant for drone, eVTOL, and airspace management startups. Advanced air mobility is a high-priority area.

Alternative paths for mobility startups:

• DOE SBIR for electrification, EV charging, and battery technology -- often a better funding fit than DOT for the underlying technology
• ARPA-E for transformational transportation energy (e.g., NEXTCAR program for connected and automated vehicles)
• AFWERX/Army SBIR for autonomous vehicles with defense applications

Where to start: DOT SBIR if your technology is specifically transportation infrastructure or safety. DOE if it's electrification or energy storage. AFWERX/Army if it's autonomous with defense relevance.

How to choose your first program

If you've read through the sector breakdowns and still aren't sure where to start, here's the decision framework:

Is your technology defense-relevant? Start with AFWERX Open Topic. Lowest barrier, fastest process, and the DoD ecosystem has the clearest Phase III commercialization path.

Is it health or biotech? Start with NIH SBIR at the institute matching your disease area. If your innovation is genuinely breakthrough (10x, not 10%), also explore ARPA-H.

Is it broadly scientific or engineering? Start with NSF SBIR. The widest scope, rolling submissions, and the most founder-friendly process of any agency.

Does it involve energy, climate, or manufacturing? Start with DOE SBIR under the relevant office.

Does it cross multiple sectors? Apply to multiple agencies. There's no rule against simultaneous SBIR applications at different agencies for the same technology, as long as the proposed work doesn't overlap. A medical AI company could apply to both NIH (health application) and NSF (AI technology) with different proposals emphasizing different aspects.

Is it pure software without a hardware or science component? You can still qualify, but the bar is higher. SBIR funds R&D, not product development. Your application needs to demonstrate genuine technical uncertainty and innovation beyond building features.

What counts as R&D for software companies?

This is the most common question from SaaS and platform founders. SBIR doesn't fund building features on known technology stacks. It funds solving technical problems where the solution isn't known in advance. The test: is there genuine technical uncertainty in what you're building?

Likely qualifies as SBIR-fundable R&D:

• A novel ML model that doesn't exist yet (new architecture, new training approach, new domain application)
• A cybersecurity system using a fundamentally new detection method
• An NLP system for a language or domain where existing models fail
• A platform solving a computational problem at a scale or speed that current approaches can't handle

Probably doesn't qualify:

• Building a dashboard on top of existing APIs
• Applying standard ML libraries (scikit-learn, off-the-shelf transformers) to a new vertical without algorithmic innovation
• Implementing features using established patterns, even if complex
• Repackaging existing open-source technology with a better UX

The gray area is real. If you're unsure, NSF's Project Pitch is the lowest-risk way to test -- it's a 4-page document, reviewed in 3 weeks, and costs nothing but a few hours.

2026 status update: the reauthorization restart

All SBIR programs paused when authorization lapsed in September 2025. Congress passed reauthorization legislation in March 2026, extending the programs through 2031. Here's where things stand:

• Large agencies (DoD, NIH, NSF, DOE, NASA): Expected to resume new solicitations in spring 2026
• Smaller agencies (USDA, EPA, DHS, DOT, ED): Expected spring to summer 2026
• Non-SBIR programs (DARPA BAAs, ARPA-H, ARPA-E, BARDA DRIVe): Were NOT affected by the lapse and are accepting applications now
• New: Strategic Breakthrough Awards: Up to $30M per award for Phase II alumni with matching funds. First solicitations expected Q4 FY2026

If you've been waiting to apply, the window is reopening. This is also a good time to prepare -- agencies will likely release a burst of solicitations as they clear their backlog.

The bottom line

Federal grant funding exists for nearly every tech sector, but the depth varies enormously. Biotech and defense startups have billions in dedicated programs and multiple entry points. Cleantech and AI have strong support across several agencies. Fintech startups have almost nothing.

The founders who capture the most grant funding don't just pick the agency with the biggest check. They match their technology to the agency whose mission it serves, start with the most accessible program (often NSF or AFWERX), and build a track record that opens doors to larger awards.

Want to know which programs fit your specific technology?

This guide gives you the map. But which 3-5 programs are actually worth your time depends on your specific technology, stage, and competitive positioning -- factors that a sector overview can't evaluate.

We've written 500+ proposals across 30+ federal agencies and know which programs are worth the effort for each type of technology. Our Strategy Review identifies the programs where your startup has the strongest shot -- ranked by fit, award size, and timeline.