DOE Just Opened Genesis Mission SBIR/STTR: ~40 Phase I Awards Across Four AI-for-Science Topics, Plus $147M in FY25 Phase II — A Small-Business Playbook

The Department of Energy has a habit of announcing its biggest ambitions at the level of national laboratories, exascale computers, and billion-dollar programs — and then quietly leaving small businesses to figure out where they fit. The Genesis Mission, DOE's flagship effort to turn artificial intelligence into an engine for scientific discovery, was announced at exactly that altitude: a multi-billion-dollar, whole-of-department push spanning fourteen national science and technology challenges and hundreds of projects across the lab complex. For most of the year, the obvious question from the startup side of the deep-tech world was simple: is there a door for us?

On July 22, 2026, DOE answered it. The department opened the first Phase I SBIR/STTR release explicitly tied to the Genesis Mission — an initial tranche of roughly 40 awards across four AI-for-science topic areas — and, on the same day, opened approximately $147 million in FY25 Phase II funding for businesses already in the pipeline. This is the small-business on-ramp to the Genesis Mission, and it is worth understanding precisely, because the initial July release is deliberately narrow and precedes a broader Phase I funding opportunity DOE says is coming "later this fall."

If you want the top-line policy framing of the overall program, we covered it in our deep dive on the $5 billion Genesis Mission and its fourteen national challenges. This piece is about the part of it a company with a few engineers and a good idea can actually apply to.

The four topic areas — and what they signal

The July Phase I release is organized around four Genesis Mission-focused topic areas. Each is a compression of a much larger DOE priority into something a small team can prototype in a Phase I feasibility effort:

1. Scaling the Biotechnology Revolution. DOE's bioeconomy ambitions — engineered organisms, biomanufacturing, carbon-negative feedstocks — increasingly depend on AI to design and predict biological systems rather than discover them by trial and error. Phase I proposals here will be judged on whether AI meaningfully compresses the design-build-test-learn cycle, not on incremental lab automation.

2. Realizing Quantum Systems for Discovery: AI for Quantum Computing and Networking. Note the framing carefully — this is AI for quantum, not quantum for AI. DOE wants machine learning applied to the hard problems of making quantum systems usable: error mitigation, calibration, control, and networking. That is a topic where a small team with deep expertise in one narrow control problem can be competitive against much larger organizations.

3. Designing Materials with Predictable Functionality. This is the materials-discovery thrust: using AI to predict a material's function before it is synthesized, inverting the traditional make-it-then-measure-it workflow. It sits directly on top of DOE's decades of investment in materials characterization data at the national labs — and the strongest proposals will show a credible path to using that data.

4. Achieving AI-Driven Autonomous Laboratories. The "self-driving lab" — closed-loop systems where AI proposes experiments, robotics execute them, and results feed back into the model without a human in the loop. This is the connective tissue of the other three: autonomous labs are how biotech, quantum, and materials all accelerate. Expect intense interest and correspondingly sharp review.

The common thread is that DOE is not funding AI research in the abstract. Every one of these topics asks a small business to point AI at a specific, expensive, slow scientific bottleneck and show that it moves. That is a meaningful filter. A proposal that reads like a generic large-language-model application will not survive; one that names the bottleneck, quantifies the current cost, and shows a believable Phase I experiment will.

Eligibility and the shape of the award

The Phase I release is, as always for SBIR/STTR, limited to for-profit U.S. small businesses that meet Small Business Administration eligibility requirements — generally 500 or fewer employees, majority U.S.-owned, and performing the work domestically. The STTR variant additionally requires a formal partnership with a research institution (typically a university or a national laboratory), with the small business performing at least 40% of the work and the research partner at least 30%. For Genesis Mission topics, an STTR structure that pairs a nimble company with a national-lab dataset or instrument can be a genuine strength rather than a compliance box to check.

DOE did not publish per-award dollar figures or a firm application deadline in the initial July release — the department framed it as an initial release that "precedes a broader Phase I funding opportunity expected later this fall." In practice, DOE Phase I SBIR/STTR awards have historically landed in the low-to-mid six figures for roughly a year of feasibility work, with Phase II awards an order of magnitude larger. The strategic implication of the "initial release" language is important: if your technology maps cleanly onto one of the four topics, the July window is your earliest shot; if it is adjacent, the broader fall solicitation is where you want to be aiming, and the smart move is to use the intervening weeks to line up a research partner and sharpen the feasibility experiment.

Applicants start through DOE's ConnectWerx portal (sbir-sttr.connectwerx.org), the same system routing the department's other SBIR/STTR activity this cycle.

The other $147 million: Phase II for the existing pipeline

The July 22 announcement paired the new Phase I release with roughly $147 million in FY25 Phase II funding, aimed at a different audience entirely. Phase II is not open to newcomers — it is available to prior DOE SBIR/STTR Phase I, Phase II, and Follow-on Phase II awardees, who receive direct outreach with application requirements and timelines. This is the commercialization tranche: the money that carries a validated Phase I feasibility result toward a product.

For companies already in the DOE pipeline, the message is to watch for that direct outreach and treat it as time-sensitive. For companies not yet in it, the $147 million is a useful signal of scale — DOE is funding the back half of the SBIR pipeline at real volume this year, which means a Phase I award now is an entry into a program with money behind its later stages, not a one-and-done grant that dead-ends at feasibility.

How this fits the 2026 SBIR restart

The Genesis Mission SBIR release does not exist in isolation. It lands in the middle of a broad federal SBIR/STTR restart following the program's reauthorization earlier in 2026. The NSF's $250 million SBIR/STTR relaunch hit its first Phase I deadline on July 27, and DARPA's Release 4 SBIR topics opened on July 22 and close August 19. For a deep-tech founder, late July 2026 is one of the densest SBIR windows in recent memory — three major agencies, all live, all inside a four-week band.

That density is both opportunity and hazard. The opportunity is obvious: a company with genuinely dual-relevant AI-for-science technology could reasonably prepare proposals for more than one agency. The hazard is dilution — spreading a small team across three agencies' worth of proposals almost guarantees three mediocre ones. The disciplined play is to pick the agency whose mission your technology most naturally serves and write one excellent proposal. For AI applied to biology, quantum control, materials design, or lab automation, DOE Genesis Mission is that agency, and the topic language above is your rubric.

What a competitive Genesis Mission Phase I proposal looks like

Three things separate the fundable proposals from the rest in a topic area this specific.

Name the bottleneck, then quantify it. Reviewers in these topics are scientists who know the field's pain points intimately. "We will use AI to accelerate materials discovery" is a slogan. "Current inverse-design workflows require N synthesis-characterization cycles at $X per cycle; our approach targets a 5x reduction validated against the following DOE materials dataset" is a proposal. The specificity is the differentiator.

Show the data path. Every one of these topics leans on data that DOE and its national labs have spent years and billions generating. A proposal that demonstrates how it will access, use, or connect to that data — through an STTR partnership, a CRADA, or a documented public dataset — is far stronger than one that assumes it will generate everything from scratch inside a one-year Phase I budget.

Design a Phase I experiment that actually resolves risk. Phase I is a feasibility study, not a mini product build. The best proposals identify the single riskiest assumption in the technology and structure the entire Phase I around testing it, so that a positive result de-risks the Phase II investment. Reviewers reward proposals that know exactly what they are trying to learn.

Bottom line

DOE opened the small-business door to the Genesis Mission on July 22, 2026, with an initial ~40-award Phase I release across biotechnology, AI-for-quantum, predictable-materials design, and autonomous laboratories, plus $147 million in Phase II funding for the existing pipeline. The initial release is narrow and precedes a broader fall solicitation — which means the window to position is now. If your technology points AI at a specific, expensive scientific bottleneck in one of those four areas, line up a research partner, sharpen the feasibility experiment, and get into the pipeline before the fall opportunity turns it into a crowd. To scope your company against DOE's active solicitations and the wider SBIR restart, start with Granted's DOE and SBIR grant discovery and build your positioning from the topic language, not the press release.