NSF's Quietest Earthquake: What the Two-Reviewer Merit Review Rewrite, 1,752 Cancellations, and $8.75B Floor Mean for Your Next Proposal

The most consequential change to federal research funding in 2026 did not arrive with a press release. NSF policy notice NSF 26-200, effective December 15, 2025, rewrote the merit review process that has governed the agency's grantmaking for more than two decades — and did so in the kind of dry, technocratic language that disguises just how much it shifts the odds for individual proposals. Combined with the 1,752 grant terminations NSF issued in spring 2025 and a FY2026 appropriation that, while well above the White House's proposed floor, still trails FY2024 by 3.4%, the practical effect is a funding environment where the human judgment of a single program officer matters more than at any point in the modern history of the agency.

If you are a researcher who has been writing NSF proposals on muscle memory, that muscle memory is now wrong.

What NSF 26-200 actually changed

The new policy makes four discrete adjustments. None is independently dramatic. Together they re-route how decisions get made.

The agency's stated rationale is operational: NSF lost a meaningful fraction of its program officer workforce during 2025 — by some directorate-level reports, up to one-third — and the resulting proposal backlog became untenable under the prior process. Fewer reviews, shorter summaries, and fewer panels reduce per-proposal staff hours. The trade-off, which the notice does not name explicitly, is that the signal carried by external review has thinned, while the weight carried by program officer judgment has thickened.

For applicants, this is not a neutral substitution. External reviewers tend to evaluate proposals on technical merit and field-specific norms. Program officers evaluate proposals on technical merit and portfolio fit, agency strategic direction, prior PI track record, congressional optics, and the particular mix of priorities they have been told to weight in the current cycle. Both are legitimate. But they are not the same.

What the 1,752 terminations told us about the new portfolio logic

Between April and May 2025, NSF terminated 1,752 active grants totaling roughly $1.4 billion. The directorate breakdown was stark:

• STEM Education (EDU): 839 grants, $888 million
• Social, Behavioral, and Economic Sciences (SBE): 320 grants, $91 million
• Physical sciences, engineering, computer science, geosciences: proportionally fewer terminations

The terminations were concentrated in two research veins: studies framed around diversity, equity, and inclusion, and work on misinformation, disinformation, and information ecosystems. Many of the canceled grants had been peer-reviewed and approved under prior rules. The decision to terminate did not come from peer reviewers. It came from the agency.

That sequence — peer-approved, agency-terminated — is the data point that matters most for understanding how NSF 26-200 will function in practice. The new merit review rules formalize what the spring 2025 terminations demonstrated empirically: agency-level priorities can and do override peer review verdicts. A two-reviewer minimum with shortened summaries and optional panels gives program officers more room to act on those agency-level signals without being constrained by lengthy panel consensus statements pointing the other way.

This is not a claim that NSF is now ignoring science. It is a claim that portfolio fit has become a first-order variable, and applicants who write proposals as if portfolio fit is implicit will be outcompeted by applicants who write proposals where portfolio fit is explicit.

The FY2026 budget context

Congress approved $8.75 billion for NSF in the FY2026 Commerce, Justice, Science spending bill, signed January 23, 2026. The number deserves more scrutiny than it has received.

The White House's initial FY2026 budget proposed roughly $3.9 billion for NSF — a 55% cut from FY2024 levels. The final $8.75 billion appropriation represents Congress decisively rejecting that proposal, and it is the strongest legislative signal in years that federal investment in basic research has bipartisan staying power. But $8.75 billion is still 3.4% below FY2024 in nominal terms, and meaningfully more in real terms once inflation since 2024 is accounted for. Demand for NSF funding has not declined; supply has.

Funding rates that hovered around 25-30% in healthy years for many NSF programs are likely to compress further. In a tight-budget environment with abbreviated reviews and elevated program officer discretion, the marginal proposal — the one that would have been funded in a 28% success rate world but not in a 22% success rate world — is now decided by judgment calls that are harder to predict from the outside.

How proposal strategy has to shift

Three adjustments are now load-bearing.

1. Write to the program officer, not the reviewer. This sounds cynical but it is mechanically correct. With two reviews instead of three, with shorter summaries, and with elevated PO discretion, the program officer is reading every proposal with substantially less external scaffolding than before. The cover letter, the project summary, and the broader impacts section now need to make the case for portfolio fit directly to the agency rather than relying on external reviewers to do it. If a program officer cannot articulate in two sentences why your project matters to this program's specific strategic direction in 2026, the proposal is in trouble regardless of technical merit.

2. Pre-submission engagement is now non-optional. Every NSF program officer accepts one-page concept summaries before formal submission. Historically, researchers who had strong publication records often skipped this step. In the new regime, pre-submission engagement is the single highest-leverage activity available — it is the only point at which you can directly read the program officer's interpretation of agency priorities before you commit to a narrative.

3. Recalibrate your portfolio mix. Researchers whose work sits primarily in NSF EDU or SBE directorates need to think hard about exposure concentration. That does not mean abandoning those directorates — the $8.75B appropriation is real, the programs are active, and good proposals will still be funded. It means that researchers with portfolios concentrated in the directorates that absorbed 67% of the spring 2025 terminations should be actively diversifying. Cross-directorate collaborations, joint funding with NIH or DOE, foundation philanthropy, and state-level workforce funding all serve as natural hedges. (For one immediate option, see our analysis of the new TechAccess AI-Ready America hubs, which are explicitly designed to absorb cross-disciplinary AI workforce work.)

The reviewer pool problem

There is one structural consequence of the two-reviewer minimum that has received almost no discussion: reviewer recruitment. Under the old three-reviewer norm, NSF program officers had room to assemble panels with redundancy — if one reviewer dropped out or turned in a thin review, two strong reviews could still anchor a decision. Under the new minimum, a single weak or absent review can leave a proposal effectively under-evaluated.

For applicants, this argues for one small tactical move: when NSF's submission system asks for suggested reviewers and reviewer exclusions, treat that field as load-bearing. Suggesting two or three substantive reviewers who are not direct collaborators (and therefore not conflicted) gives the program officer easy access to qualified evaluators. In a constrained-staff environment, that small reduction in PO workload has compounding effects.

The longer arc

The history of federal merit review reform — from the 1997 NSF Criteria Working Group through the 2013 Public Access Plan to the 2020 IT modernization round — suggests that changes of this magnitude rarely revert. NSF 26-200 is now the baseline. Researchers planning multi-year submission cycles should plan against it lasting through the rest of the decade, not against an eventual rollback.

That is not necessarily bad news. A merit review process with sharper agency priorities and more visible program officer judgment is more navigable than one that pretends those forces do not exist, provided you adapt your strategy to it. The applicants who do the work of pre-submission engagement, portfolio diversification, and explicit portfolio-fit framing in 2026 will be the ones who hold or grow their NSF funding through the back half of the decade.

For real-time tracking of NSF solicitations, NSF funder updates, and the broader federal funding landscape, Granted's research grant pages stay current with every major change. (Granted News)