DOE Will Put $100 Million Behind 100 AI-Fluent PhDs — and Your Institution Gets Exactly One Letter of Intent, Due November 9

Academic PIs and research-office leaders at doctoral-granting institutions have until 5 p.m. Eastern on November 9 to file the mandatory letter of intent for the Department of Energy's new $100 million Genesis Mission Graduate Fellowship Pilot, posted to Grants.gov on October 8 as DE-FOA-0003673. One letter per institution — not one per department, not one per college.

That single line of eligibility language is why this announcement is an institutional governance problem before it is a scientific one.

One Letter of Intent Per Institution, and the Clock Started October 8

Opportunity 364033 on Grants.gov lists DE-FOA-0003673 under the DOE Office of Science, Assistance Listing 81.049, with a close date of January 12, 2027 at 11:59 p.m. Eastern. The posted summary shows $100,000,000 in total program funding, an award ceiling and floor of $20,000,000, and five expected awards. Cost sharing: none required.

But the calendar that actually governs you is earlier. The letter of intent is due November 9, 2026 at 5 p.m. Eastern, must be submitted by an Authorized Organizational Representative, and DOE will respond by November 23. An informational webinar runs November 4 at 4 p.m. Eastern. That is a three-week runway from the day the solicitation hit the street to the day your institution must have already decided whose proposal it is.

Because only one LOI is permitted per institution as lead applicant, research offices have converted this into an internal limited-submission competition — and their internal deadlines are brutal. The University of Colorado Boulder's posted guidance set an internal letter-of-intent deadline of 11:59 p.m. Mountain on October 25, two full weeks ahead of DOE's. If your institution follows a similar cadence, the real deadline for a PI is roughly ten days from today, and it is not a proposal — it is an internal pitch for the right to write the proposal.

PIs who treat November 9 as their deadline will find the slot already assigned.

What DOE Is Buying: Roughly $1 Million Per Doctoral Fellow

The Office of Science framing is specific about scale. DOE says it will fund an inaugural group of pilot projects supporting approximately 100 Ph.D. fellows, structured as an accelerated four-year doctoral pathway. Pair that with five expected awards and the arithmetic is uncomfortable in a useful way: about twenty fellows per institutional pilot, at roughly $1 million of federal money per fellow over four years.

For context, an NSF Graduate Research Fellowship under NSF 26-526 pays a $37,000 annual stipend plus a $16,000 cost-of-education allowance to the institution — $159,000 of federal support across three years of tenure. DE-FOA-0003673 is operating at something closer to an order of magnitude above that per head.

That is not generosity for its own sake, and reviewers will not treat it that way. The solicitation requires budget requests to cover "highly competitive stipends, full tuition, and health insurance" — which at a private doctoral institution can consume $120,000 to $150,000 per fellow-year before anyone touches research costs. It also requires two high-impact research experiences per fellow: one at a DOE National Laboratory or Office of Science user facility, and one with an industry partner. Those placements carry relocation, host-side mentorship, security and badging overhead, and in many cases compute allocation. Add program staff, curriculum development, recruiting, and assessment, and the $1 million-per-fellow figure stops looking like a windfall and starts looking like a constraint you must defend line by line.

Note one discrepancy worth resolving at the webinar. Grants.gov shows a flat $20M ceiling and floor with five awards; several university limited-submission notices summarizing the full FOA PDF describe an award range of roughly $12M–$25M, four to eight anticipated awards, and a five-year project period against a four-year student pathway. Those are not the same budget envelope. The announcement also states that only $20 million is available in Fiscal Year 2027, with outyear funding contingent on congressional appropriations — so the $100 million is a plan, not an obligation. Read the actual FOA document and Part 2 attachment before you size a budget off the Grants.gov summary card.

"Super Intelligence" Is DOE's Own Term, and It Is Load-Bearing

The solicitation does not say "AI for science." It says dual competency in Super Intelligence (SI) and another scientific discipline. Under Secretary for Science Darío Gil put it directly: "To secure America's position at the forefront of global scientific innovation, we must cultivate a workforce that is fluent in both foundational sciences and advanced Super Intelligence."

Terminology is not cosmetic in a first-round competition. It tells you which reviewers DOE recruited and which vocabulary signals insider fluency. A proposal that reads as a machine-learning methods certificate bolted onto an existing Ph.D. program is answering a different solicitation than the one DOE wrote. The word "dual-competency" is doing specific work: DOE wants fellows who are genuinely credentialed in a domain — fusion, nuclear physics, structural biology — and genuinely capable in frontier model work, not domain scientists who took two courses.

The participating program offices confirm where the domains live: Advanced Scientific Computing Research, Biological and Environmental Research, Basic Energy Sciences, Fusion Energy Sciences, High Energy Physics, Isotope R&D and Production, and Nuclear Physics. Seven offices means seven internal constituencies for roughly five awards. If your pilot maps cleanly onto one office's workforce pipeline rather than gesturing at all of them, you have a sponsor inside the building.

This is the same institutional logic that governed the research side of the program. When DOE and the White House committed $5 billion across 278 Genesis Mission projects, the winners were overwhelmingly teams whose science was physically adjacent to a federal instrument — we broke down that selection pattern in our analysis of the Genesis Mission's Phase 2 structure and the $6–15 million award band. The fellowship pilot carries that pattern into workforce development: the lab placement requirement is not a nice-to-have enrichment activity. It is the mechanism by which DOE converts its facility portfolio into a talent pipeline. Pilots with pre-negotiated, named lab hosts will outcompete pilots that promise to find hosts later.

The Citizenship Line Splits Your Applicant Pool in Half

All fellows must be U.S. citizens or lawful permanent residents. There is no corresponding restriction on PIs, mentors, or collaborators.

For most AI-adjacent doctoral programs in physics, computing, and engineering, that requirement removes a large share — at many departments, the majority — of the incoming cohort from eligibility. Any pilot that claims it will recruit twenty exceptional domestic fellows per cohort needs to show its recruiting pipeline, not assert it. Reviewers in a first-round competition have no track record to fall back on, so the recruitment plan carries disproportionate weight. Partnerships with minority-serving institutions, undergraduate feeder programs, and national-lab internship cohorts are the credible evidence here.

Federal graduate-funding programs have been converging on this design — cohort-scale money with a hard citizenship gate and a tight enrollment-timing rule. USDA's National Needs Fellowship did something structurally similar, funding whole graduate cohorts while restricting eligibility to students who had not yet started, as we covered in our breakdown of NIFA-NNF-011737's $14 million and its new graduate-student rule. If you have run an NNF competition, much of that institutional machinery transfers.

Why This Money Arrives as the NSF Pipeline Contracts

The timing is not incidental. The NSF Graduate Research Fellowship Program — for decades the default prestige credential for early-career American scientists — has become the least predictable line item in American graduate funding. NSF cut the 2025 cohort to 1,000 awards from 2,036 the prior year, then reversed course and issued a record roughly 2,600 fellowships in 2026. Awards are now projected to fall about 42 percent from that FY26 peak. We traced that whipsaw and what the disciplinary steering inside it meant for life-sciences applicants in our piece on the GRFP's 50 percent reduction.

DE-FOA-0003673 does not replace the GRFP, and it is not a substitute from a student's perspective: GRFP money is portable and student-held, while Genesis fellowship money is institutional and tied to a specific training design. But for a department chair who cannot forecast how many NSF-funded students will walk in next fall, a program that can underwrite twenty fellows for four years on a known schedule is the most consequential graduate-funding opportunity of this cycle. It also means the competition will not be limited to the obvious AI powerhouses. Institutions with strong single-domain programs and genuine lab relationships have a real path, because DOE needs breadth across seven program offices, not five versions of the same computer-science proposal.

What to Do in the Next Ten Days

Three concrete moves, in order.

Find your institution's internal deadline today. Search your research office's limited-submission page for DE-FOA-0003673. If an internal LOI date has passed, ask whether a slot was claimed — in some cases nobody filed and the slot is open.

Register for the November 4 webinar and bring the budget question. Ask directly about the award-range discrepancy, the FY2027 $20 million constraint versus the $100 million plan, and whether the five-year project period against a four-year pathway implies a cohort ramp.

Lock your DOE lab host and industry partner before you write prose. Named, committed hosts with a signed letter are the differentiator in a first-round competition. Email the program contact at GM-Uni-Engagement@doe.gov with scoping questions; grants and contracts questions go to sc.grantsandcontracts@doe.gov.

If you are mapping which Office of Science programs align to your department's domain strengths — or hunting the companion research solicitations that make a training pilot credible — search active DOE Office of Science graduate training and workforce solicitations on Granted and filter by deadline to see what else closes before January 12.

First rounds reward institutions that move while everyone else is still reading the PDF. The letter of intent is due November 9.