AI Grants Weekly: SBIR Is Back, DOE Drops $320M, and NIH Rewrites the Rules

After five months of limbo, America's seed fund for small business innovation is finally coming back. The Ernst-Markey deal to restart SBIR/STTR isn't just a reauthorization — it comes with new rules and a restructured program that every startup founder and university tech-transfer office needs to study before the solicitations reopen. If you spent the winter wondering whether to pivot your commercialization timeline, the answer is: dust off those proposals.

Meanwhile, the Department of Energy made the week's boldest move with its Genesis Mission — 26 AI grand challenges backed by $320 million. The explicit goal is to double American scientific output using AI, and DOE isn't stopping there: a separate $500 million grant window across seven research programs just opened. For labs already working at the intersection of machine learning and physical sciences, this is a once-in-a-decade alignment of money and mission.

NIH, by contrast, is tightening. The agency capped grant applications at six per year — and AI is the stated reason, a direct response to what internal data shows is an uncomfortable paradox: machine-assisted proposals win more often but innovate less. Add in plummeting success rates under the new multiyear funding policy, and the picture for biomedical researchers is genuinely shifting. The one bright spot: thousands of previously frozen grants were reinstated following the court settlement.

NSF is undergoing its own transformation. The foundation is cutting its grant solicitations in half while simultaneously overhauling the peer review process to give program officers more discretion. The new Tech Labs program offering $10–50 million per year signals that NSF wants fewer, bigger bets — a structural shift away from the broad, small-grant portfolio that defined the agency for decades. On the infrastructure side, $100 million just opened for quantum and nano lab sites.

The private side of the ledger is almost surreal. We dug into the numbers and found $195 billion in AI capital deployed in a single month — a figure that makes federal research budgets look like rounding errors. Yet the philanthropic world is stepping into the gap: Stephen Schwarzman's $48 billion foundation commitment and MacArthur's $500 million dedicated to humane AI together represent a new scale of private funding for responsible AI research. And Google is giving away $30 million for AI-powered science through a program worth studying closely.

Two deadlines to act on now: the EPA's $250,000 environmental education grants close March 3, and Mozilla's $50K democracy-and-AI cohort closes March 16. Neither requires a massive institutional apparatus to apply.

Watch next week for the first post-reauthorization SBIR solicitations and DOE's detailed FOA releases under Genesis. The money is moving — the question is whether your proposal is ready when the window opens.