DOE's Genesis Mission SBIR/STTR: $10M for ~40 Phase I Awards, $147M in Phase II, and a September 10 Deadline for Small Businesses in AI-Driven Science

When the White House announced the Genesis Mission — its $5 billion, 15-agency push to make AI the engine of American science — the abstract question for a small company was: where does the money actually come out? On July 22, 2026, the Department of Energy answered part of it, opening a Genesis Mission SBIR/STTR funding opportunity aimed squarely at American small businesses. It is not a huge number by federal standards — $10 million across roughly 40 Phase I awards — but it is a clean, early, well-defined entry point into the mission, with a full-proposal deadline of September 10, 2026. And it arrives alongside a much larger $147 million in FY25 Phase II funding for companies already in the pipeline.

For deep-tech founders, the interesting story is not just the dollars. It is that DOE is using this opportunity to debut a redesigned SBIR/STTR process it describes as "faster, simpler, and more accessible." Here is what is on offer, what the four topics reveal, and how to approach a window this short.

The two opportunities, side by side

Phase I — the front door. $10 million total, approximately 40 awards, deadline September 10, 2026, applications through the DOE SBIR/STTR Application Hub at sbir-sttr.connectwerx.org. Phase I is proof-of-concept money: enough to demonstrate technical feasibility and earn the right to compete for the far larger Phase II. DOE has also signaled an additional Phase I opportunity opening in fall 2026, so a company that misses September 10 is not shut out of the mission for the year — but the first movers get the first shot.

Phase II — the scale money. Approximately $147 million, open to prior-year Phase I, Phase II, and Follow-on Phase II award recipients. This is the tell that matters most for anyone new to the program: the real money in SBIR is in Phase II, and Phase II is only reachable through Phase I. A $10 million Phase I pool is small precisely because it is a filter — the on-ramp to a $147 million pool. Founders should read Phase I not as the prize but as the qualifying round.

The four topic areas — and what they signal

DOE organized the Genesis Mission SBIR opportunity around four topic areas, each of which maps directly to one of the mission's named scientific challenges:

1. Scaling the Biotechnology Revolution. Moving biotechnology from lab-scale demonstration toward manufacturable, scalable production. For a small company, the winning angle is usually a specific bottleneck — a bioprocess, a measurement, a control system — that stands between a promising biological product and being made at real volume, with AI accelerating the design-build-test loop.

2. Realizing Quantum Systems for Discovery: AI for Quantum Computing and Networking. Note the framing carefully — this is not "build a quantum computer." It is AI in service of quantum systems: using machine learning for error correction, calibration, control, or networking. That is a far more tractable target for a small team than full-stack quantum hardware, and the topic wording tells you DOE knows it.

3. Designing Materials with Predictable Functionality. AI-driven materials discovery — predicting a material's properties from its structure and inverse-designing materials to hit a functional target. This is one of the most mature AI-for-science domains, which cuts both ways: the science is ready, but so is the competition, so specificity about which materials problem and which validated pipeline wins.

4. Achieving AI-Driven Autonomous Laboratories. Self-driving labs — closed-loop systems where AI designs experiments, robotics runs them, and results feed back to refine the next round. This is arguably the mission's signature ambition, because an autonomous lab is a force multiplier across every other challenge area.

Read as a set, the four topics tell you what DOE thinks small businesses can uniquely contribute: not the frontier models (national labs and hyperscalers handle those) but the applied, domain-specific systems that turn AI capability into scientific throughput. A proposal that positions the company as the specialist who makes AI usable inside one of these domains is speaking the program's language.

The "faster, simpler" redesign — and what to actually expect

DOE Chief Commercialization Officer Anthony Pugliese framed the redesign as an effort to make the programs "faster, simpler, and more accessible" while "reducing unnecessary burden" for small-business innovators. That is worth taking seriously but not taking on faith.

For founders, "faster and simpler" most credibly translates to a streamlined application and a compressed timeline — which is a double-edged benefit. A lighter application lowers the cost of trying, especially for first-time applicants who have historically been scared off by SBIR's paperwork. But a compressed timeline — the September 10 deadline is roughly seven weeks from the July 22 opening — means less time to assemble the technical narrative, line up any required partnerships, and get registered in the necessary federal systems. The redesign helps most the companies that are already prepared to move; it does the least for companies that treat registration and teaming as things to start after they read the notice.

The single most avoidable way to lose a fast SBIR window is administrative. SAM.gov registration, SBC registration, and the DOE application-hub account are not instantaneous, and a technically excellent company that discovers a registration gap in the final week is simply out. If you intend to apply, verify every registration today, not in the last fortnight.

STTR vs. SBIR — a quick but consequential choice

Because this is an SBIR and STTR opportunity, some topics may be open under both. The practical difference: STTR requires formal partnership with a research institution (a university or national lab), with a minimum share of the work performed there; SBIR does not require it but permits subcontracting. For an AI-for-science topic — quantum, autonomous labs, advanced materials — many small companies have exactly the kind of university or national-lab relationship that makes STTR natural, and STTR can be a way to borrow credibility and instrumentation a small company cannot own. Decide early which track fits, because it shapes your teaming and your budget.

Where it sits, and how to move

This opportunity is one raindrop in a much larger storm. It shares its DNA with NSF's AI Datasets program (NSF 26-512, up to $100 million — see our breakdown) and both trace back to the Genesis Mission. It also lands in the same season as NSF's separate SBIR/STTR relaunch, making this one of the richest stretches for deep-tech small-business funding in recent memory. Companies working at the intersection of AI and the physical sciences have an unusual number of doors open at once.

A concrete plan for the September 10 deadline:

• Confirm all registrations now. SAM.gov, SBC, and the ConnectWerx application hub. This is the failure mode that kills good applications; eliminate it in week one.
• Pick the topic where you are already ahead. Seven weeks is not enough to start a new research thread. Choose the topic area where you have existing results, and lead with them.
• Decide SBIR vs. STTR early. If a research-institution partnership strengthens your case and you have the relationship, STTR may fit — but the partnership must be real and documented, not aspirational.
• Frame Phase I toward Phase II. Reviewers fund Phase I projects that have a credible path to the $147 million Phase II pool and to commercialization beyond it. Show that path explicitly.
• Aim at a bottleneck, not a field. The strongest small-business proposals name a specific problem inside "scaling biotech" or "AI for quantum" and solve it, rather than gesturing at the whole topic area.

DOE's $10 million is modest, but its position is not: it is the earliest, cleanest small-business entry point into a $5 billion national mission, and it opens onto a $147 million follow-on pool. For a deep-tech company whose work already lives at the intersection of AI and the physical sciences, the question is not whether the money is worth chasing. It is whether your registrations are current — because the deadline is September 10, 2026, and it will arrive faster than the paperwork does.