NOAA's Climate Program Office Opens Its FY2027 'Super Intelligence' Competition: 15 Awards, a November 17 LOI, and the First Big Test of Whether OAR Grants Survive

On October 27, 2026, NOAA's Climate Program Office (CPO) posted NOAA-OAR-CPO-2027-33392, the Climate Program Office FY 2027 – Super Intelligence/Machine Learning Innovation to Advance NOAA's Earth System Science and Support Hazard Resilience. The Grants.gov listing was updated the next day. It is a joint competition run by three CPO programs: Modeling, Analysis, Predictions and Projections (MAPP), Adaptation Sciences (AdSci) and the National Integrated Heat Health Information System (NIHHIS).

The headline terms:

If every award ran three years at the full $250,000, the competition would total about $11.25 million. That is our ceiling math, not a NOAA figure; the NOFO lists no program total and says FY2028 funds may be used. A realistic portfolio is probably smaller: a mix of two- and three-year awards, some below the cap.

The competition is small by Genesis Mission standards. It matters more than its size suggests, for two reasons. It is one of very few competitive solicitations from NOAA's research office since early 2025. And it shows how the "Super Intelligence" relabeling is now reshaping research priorities at agencies well outside the AI mainstream.

Why this NOFO is a signal

The administration's FY2027 budget request proposed abolishing NOAA's Office of Oceanic and Atmospheric Research (OAR), the home of CPO and of nearly every competitive research grant NOAA makes. The FY2026 request proposed the same thing, and Congress instead provided $634 million for OAR. (See our FY2027 science budget analysis and coverage of the bipartisan pushback at NOAA's budget hearing.)

Even with appropriations in hand, CPO's extramural pipeline has been nearly silent. A September 28, 2026 Resources for the Future review found that a May 2026 Regional Integrated Sciences and Assessments (RISA) announcement was "the first NOFO issued since the beginning of the second Trump administration." Our own tracking found no posted successor to OAR's 2024 broad agency announcement when NOAA's other line offices put out their FY2027 BAAs.

So an FY2027 CPO competition, posted in the first month of the fiscal year and run jointly by three programs, is the clearest sign yet that OAR's grant-making machinery is still running. The way it is framed shows what kind of work can still get through.

The "Super Intelligence" frame

The NOFO uses "Super Intelligence/Machine Learning (SI/ML)" throughout where a 2024 document would have said "AI/ML." Among its citations is the White House's September 2026 presidential action "Inaugurating the Era of Super Intelligence," along with the January 2025 order on removing barriers to American AI leadership and OMB memo M-25-21.

We have seen the same rename at NSF's Super Intelligence-Native Instruments program and in DOE's Early Career priorities. Two practical consequences follow:

1. Search alerts keyed only to "artificial intelligence" will miss these. This NOFO's title never says "AI"; the only clue is the office name. Add "super intelligence" and "SI/ML" to every saved search.
2. The climate science is intact, but the framing has moved. The three priority areas are decadal prediction, hazard decision support and extreme heat, all core CPO science. The justification now rests on hazards, resilience, preparedness, protecting "people and property," and economic sectors such as water, energy and agriculture. Write your proposal in that language.

The three priority areas

Proposals may address one or more areas, and NOAA says selections will be balanced across them.

Priority Area 1 — Multi-annual to decadal (M2D) prediction (MAPP)

The NOFO says the U.S. has experimental prototypes but "no U.S. system that currently delivers such information routinely and authoritatively" at 2–10 year lead times. Proposals should address one or both of:

• (a) Predictability discovery and diagnosis: SI/ML methods that improve signal-to-noise for known M2D predictability sources or find new ones. The research "must be grounded in physical interpretability" and tested for robustness across independent datasets. NOAA specifically invites work separating the forced trend from internal variability, and names NOAA SPEAR, CMIP6, NOAA 20CR and ERA5 as data sources.
• (b) Computational efficiency: SI/ML emulators of NOAA models (or their components) that preserve model performance and process fidelity, including very large emulator ensembles, hybrid prediction systems, and forecast-of-opportunity triggers.

Area 1 proposals must benchmark against conventional approaches, quantify uncertainty, and lay out a plausible path to improving NOAA's M2D predictions. MAPP-directed proposals that need NOAA high-performance computing must include an HPC request and a one-page transition plan.

Priority Area 2 — Hazard resilience decision support (AdSci)

This is the social-science lane, and it is unusual for an AI solicitation. NOAA observes that Americans already use SI/ML tools to find hazard information, and says "it is often unclear how people are using or interpreting NOAA information provided by SI/ML tools." Research should:

• (a) Evaluate current and potential SI/ML uses for hazard decision-making at the community or organization level, including successful use cases, pitfalls and risks, and integrating qualitative and infrastructure data; and/or
• (b) Identify socio-technical factors that enable adoption: institutional barriers, trust, governance, data literacy.

NOAA flags small- to mid-sized communities and organizations with limited staff as "of particular interest." That gives emergency management researchers, planning schools and boundary organizations with municipal partners a real opening.

Priority Area 3 — Extreme heat (NIHHIS, MAPP, AdSci)

Extreme heat is "a leading cause of weather-related fatalities each year," and this is the most data-integration-heavy area. Proposals should:

• (a) Identify and model heat thresholds and interactions across health, infrastructure and agriculture, combining Earth system data with things like power outages, cooling-center data, labor data and urban heat campaigns. Urban heat island work is explicitly welcome.
• (b) Find extreme-heat forecast-of-opportunity windows, with physical interpretation of the predictability sources and a check of whether NOAA models capture them.

Area 3 proposals should engage NIHHIS partner agencies.

Requirements across all areas

Every proposal should interpret SI/ML results in terms of established processes, release any SI/ML-ready NOAA data it creates, consider contributing a Python notebook to NOAA's Learning Journeys Library, address AI workforce development through students and early-career staff, and include letters of support from the communities and partners that will use the results.

Eligibility and money rules

Eligible: institutions of higher education, other nonprofits, commercial organizations, local and federally recognized tribal governments, and federal agencies with statutory authority to receive transferred funds. Federal applicants cannot rely on the Economy Act; they must show other authority. Foreign and international institutions are not eligible. There is no cost share.

Two points trip up first-time applicants:

• Commercial firms are eligible, but "fees and profits are not permitted." A climate-analytics startup can apply, but only at cost. Contract work delivering products to NOAA is also excluded.
• Current or past CPO grantees may apply for new work that builds on, but does not replicate, prior awards. Current grantees may not use this NOFO for supplements.

Indirect costs follow your negotiated rate, or the 15% de minimis rate if you have none.

How it will be scored

Each criterion is scored 0–5, and proposals with a weighted score below 3.0 are not considered. Panels do not reach consensus; each reviewer scores independently. Relevance scoring explicitly includes the PI's past record of making data accessible, as well as the data-sharing plan, so a thin data history costs you in a quarter of the score.

The selecting official may fund out of rank order for geographic balance, institution type, partner type, research priority, duplication, past performance, and partnerships with targeted groups.

Two SI-specific rules are new. Applicants are "encouraged to indicate in the project narrative the extent to which, if any, generative SI technology was used" to write the proposal. And reviewers are "prohibited from uploading any content from grant applications" into generative SI tools.

Format

• 25 pages total if under three PIs, 30 pages if three or more
• Title page (1–2 pages), abstract (1 page), project narrative (15 pages; 20 with three or more PIs), data/information sharing plan (2 pages), budget table and narrative, all within the limit
• Outside the limit: bibliography, CVs, current and pending federal support, NICRA, federal forms (SF-424 family, CD-511), letters of support
• 12-point font, single spaced, one-inch margins; noncompliant applications are returned without review

Co-applicant institutions submit applications identical to the lead's except for their own budgets and forms. Subawardees do not submit separately.

Timeline and the eRA Commons trap

The LOI is optional. You may submit a full proposal without one, and a discouraging response does not bar you. But the feedback is the only free relevance check you will get. Skipping it in a 15-award competition split three ways is a mistake.

The bigger risk is eRA Commons. NOAA moved its grants to the eRA system, the same one NIH uses, and organizational registration has to be in place before Grants.gov will accept the submission. Many universities are already registered through NIH. Small nonprofits, tribal governments and companies often are not. The six-week window collides with the holidays, so start in November, not January.

Strategy

1. Pick one priority area as your anchor and contact that program manager before the LOI. Contacts are listed by area: Dan Barrie (MAPP) for Area 1; Bhaskar Subramanian (AdSci) for Area 2; and for Area 3, Barbara Boustead (NIHHIS), Dan Barrie or Andrea Miralles-Barboza (AdSci). General questions go to competition manager Kathleen Palermo. A short email testing your idea against the program's current gaps will make the LOI sharper.
2. Build interpretability in from page one. "Physical interpretability," "process fidelity," "benchmarking" and "robustness" show up again and again in the NOFO. Black-box skill gains will score poorly on the 50% technical criterion. Describe how you will explain what the model learned in domain terms.
3. Name your NOAA counterpart. Every area asks you to engage NOAA labs, the NOAA Center for AI, the National Weather Service or other line offices. A support letter from a GFDL, PSL or NWS scientist who will actually work with your team is a strong relevance signal. NOAA notes that substantial NOAA involvement may convert the award to a cooperative agreement, which is not a downside.
4. Area 2 is the least crowded lane. Most ML teams will chase emulators and heat prediction. Social scientists who study how emergency managers, utilities or small-town planners actually use AI tools for hazard decisions are answering a question NOAA calls "just emerging." Partner with a small or mid-sized community and say so.
5. Budget at or below $250,000 per year and justify every line. "Most awards" at that level means a $350,000-per-year request needs an unusually strong case, and cost is 10% of the score. A three-year project at about $230,000 a year with clear deliverables is easier to fund than a maximal one.
6. Plan for funding risk. The NOFO is explicit that awards depend on CPO's allocation and that FY2028 money may be used. With a continuing resolution through December 11 and OAR proposed for elimination, keep the start date flexible and don't commit cost-shared staff lines on the assumption of a September 1 start.
7. Treat data sharing as a scored deliverable. Commit to publishing SI/ML-ready datasets, and if you can, a Learning Journeys notebook. Cite your past public datasets in the narrative, because reviewers are told to weigh your track record.

The bottom line

NOAA-OAR-CPO-2027-33392 is a modest competition: up to 15 awards, about $250,000 a year each, with no cost share. Its importance is what it shows. NOAA's research office is still running extramural competitions in FY2027, and the way to get climate science funded right now is to frame it as SI/ML-driven prediction, hazard resilience and heat preparedness.

The next deadline is November 17. A two-page LOI costs little and gets you a relevance read from CPO by mid-December. Start eRA Commons registration at the same time.