The National Science Foundation funds startups through its SBIR/STTR program, branded “America’s Seed Fund,” and it’s one of the most AI-friendly federal funding sources available. This guide covers how it works in 2026, cited to NSF. (dgm builds the AI; proposal writing is a separate specialty — see the end.)

What NSF’s America’s Seed Fund is

America’s Seed Fund is NSF’s implementation of the federal SBIR and STTR programs. It provides non-dilutive funding — no equity taken, no repayment — to early-stage US companies developing high-risk, high-impact technology across nearly every field, with a strong and explicit place for AI.

Award amount

An NSF Phase I award is up to about $305,000 for a project of roughly 6 to 18 months, inclusive of indirect costs and fees. Successful Phase I projects can pursue larger Phase II funding to continue development. As with all SBIR figures, confirm the current number against the live solicitation.

The AI topic areas

What makes NSF especially relevant for AI startups is its dedicated artificial intelligence topic, which spans eight subtopics: cognitive-science-based AI, computer vision, conversational AI, language-based AI, novel AI hardware, sustainable/low-resource AI, trustworthy AI, and an “other novel” category. If your work advances AI itself, there’s a clear home for it rather than forcing it into a general technology bucket.

Eligibility

To apply, you must be a US for-profit small business, more than 50% owned by US citizens or permanent residents, with the work performed in the US and the principal investigator employed by the company at least part-time. Notably, NSF does not fund companies majority-owned by multiple venture-capital, private-equity, or hedge funds — a rule that catches some venture-backed startups off guard, so check it early.

The Project Pitch process

NSF’s process has a helpful on-ramp: before a full proposal, you submit a short Project Pitch describing the innovation, the technical risk, and the commercial opportunity. NSF typically responds within a month or two on whether to invite a full proposal. This means you can test fit with relatively little effort before committing to a full application.

Full-proposal deadlines are set per solicitation and move over time, so check the current NSF SBIR/STTR solicitation page for the live dates rather than relying on a fixed calendar.

Is NSF right for your AI project?

NSF funds genuine innovation and R&D — pushing AI forward — not the adoption of existing tools. If you’re building something new in AI and can articulate the technical risk, it’s an excellent fit. If you’re rolling out off-the-shelf AI internally, the R&D tax credit (if you customize) or other levers will serve you better.

How dgm helps

dgm implements osFoundry and builds AI for US businesses. We help startups develop the underlying technology and architecture — the substance an NSF project is built on — while the Project Pitch and full proposal are best handled with an experienced SBIR/STTR advisor or NSF program staff.