NSF Will Issue About 6,100 New Grants This Year — the Fewest Since the Early 1980s. A $1 Billion Central Account Is the Reason.

The number that matters is 6,100.

That is roughly how many new grants the National Science Foundation is on track to issue in fiscal year 2026, which ends September 30. As of August 19, the agency had made 5,684 new awards. The final figure will land somewhere near 6,100 — about 46 percent below the 2021–2024 average, and the lowest annual total since the early 1980s, when NSF's budget was a fraction of today's in real terms and the U.S. research enterprise was a fraction of its current size.

This is not the grant-terminations story. It is not the indirect-cost story. It is not the reorganization story. Those are all real, and we have written about each. This is a separate mechanism, and it is the one most likely to determine whether your proposal gets funded between now and the end of FY2027.

The mechanism: money that exists but cannot be reached

NSF's FY2026 appropriation is approximately $8.8 billion. Congress appropriated it. It has not been rescinded, impounded through a formal special message, or terminated.

Roughly $1 billion of it is being held in a central account at the agency level, outside the budgets of the directorates and divisions that actually issue awards. Program officers in BIO, CISE, ENG, GEO, MPS, SBE, EDU, and TIP cannot commit funds they do not control. When a program officer tells you the panel loved your proposal but there is no money in the program, that statement can be simultaneously true and consistent with an $8.8 billion appropriation.

According to reporting in Nature — and corroborated by NSF staff accounts — much of that centrally held money is allocated to a White House initiative called Grand Research Challenges, an Office of Science and Technology Policy effort oriented toward advanced materials and artificial intelligence. Neal Lane, who directed NSF from 1993 to 1998, described withholding this share of the agency's budget in a single year as unprecedented.

The critical wrinkle: NSF has two fiscal years to obligate funds appropriated by Congress. Money held centrally in FY2026 can be spent in FY2027. That is what makes this legally durable and strategically confusing. From the agency's perspective, nothing has been lost. From a principal investigator's perspective, an entire submission cycle has evaporated.

Pending purgatory

Internal NSF data show roughly 400 grants waiting at the Office of Award Management (OAM) for processing. These are not proposals awaiting review. They have been reviewed. Panels convened, ad hoc reviews came in, program officers wrote their analyses and made recommendations. The proposals are in the final administrative step before an award letter issues.

They are not declined. They are not funded. Researchers call it pending purgatory, and it is operationally worse than a decline for three specific reasons:

1. You cannot resubmit. A pending recommendation is not a decision. Most PIs will not submit a competing version of the same idea while one is under consideration, and NSF policy discourages duplicate submissions.
2. You cannot budget. Departments cannot commit startup matching, bridge funding, or graduate lines against an award that may or may not materialize.
3. You cannot recruit. Doctoral admissions and postdoc offers run on hard calendars. An offer made in February against a grant that funds in November is an offer the candidate takes elsewhere.

A decline, however painful, is information. Pending purgatory is the absence of information, and it propagates outward through every downstream commitment a lab makes.

The congressional oversight track

On July 22, 2026, OSTP Director Michael Kratsios appeared before the House Science, Space, and Technology Committee and, in the assessment of committee Democrats, declined to answer questions about OSTP's redirection of NSF resources toward OSTP-selected priorities including Grand Research Challenges.

Representatives Zoe Lofgren and George Whitesides subsequently demanded a comprehensive briefing on the Grand Research Challenges initiative and on OSTP's control of NSF operations and appropriated funding, setting an August 14, 2026 deadline. The core legal question they raise is whether an executive office component directing the allocation of another agency's appropriated funds — without a formal impoundment message to Congress under the Impoundment Control Act — constitutes an end run around the appropriations process.

For grant seekers, the oversight fight matters for one reason: it is the only mechanism that could release the central account before FY2027. Litigation over terminations restores individual awards. Appropriations oversight is what unfreezes a pool.

Why this compounds with the August 17 solicitation reset

Eleven days before the Nature report, on August 17, 2026, NSF published 12 new notices of funding opportunity carrying more than $1.5 billion for foundational research, consolidating and replacing a large set of prior core solicitations. Several of those NOFOs list "Proposals Accepted Anytime" in place of fixed deadline dates. We covered that reset in detail in NSF Just Deleted the Deadline.

Read the two developments together and the picture sharpens considerably.

A no-deadline submission regime moves discretion from the calendar to the program officer. Under a target-date or deadline model, proposals arrive in batches, panels convene on a schedule, and the funding decision is bounded by a cycle. Under "accepted anytime," a program officer decides when to convene review and when to forward a recommendation — and if the program has no obligation authority because the money sits in a central account, the rational administrative behavior is to slow the intake rather than generate a queue of recommendations that cannot be funded.

That is not a conspiracy theory. It is what a constrained system does. And it means the 400-proposal OAM backlog is likely a floor, not a ceiling.

What to actually do

Ask the direct question, in writing, before you submit. For any NSF program under a no-deadline NOFO, email the cognizant program officer and ask two things: whether the program is currently making new awards in FY2026, and what the program's expectation is for FY2027 obligation authority. Program officers will not always answer the second question. Many will answer the first. That answer is worth more than any amount of proposal polish.

Treat "recommended" as unfunded until the award letter arrives. Do not extend offers, sign subaward commitments, or purchase equipment against a recommendation. If your institution's research office allows pre-award spending at risk, understand that the risk in FY2026 is materially higher than the historical base rate that policy was written against.

Time submissions toward the first quarter of FY2027. Because NSF has two-year obligation authority, the centrally held FY2026 money remains available through September 30, 2027. If it is released — by oversight pressure, by OSTP finalizing Grand Research Challenges awards, or by ordinary year-end obligation behavior — the release will show up as an unusual burst of awards. Proposals that are already reviewed and sitting at the program will absorb that burst first. Being in the queue when it opens beats being early to a closed window.

Diversify the agency mix aggressively. This is the least satisfying advice and the most important. A lab whose funding model assumes a steady NSF core award every three to four years is exposed to a 46 percent contraction with no hedge. DOE Office of Science, NIH institutes outside the most contested areas, DoD basic research offices (AFOSR, ARO, ONR) under their standing BAAs, NASA ROSES elements, and private foundations all operate on different appropriation and allocation logic. Some of those are contracting too. They are not contracting for the same reason at the same time, which is the entire point of diversification.

If your work plausibly fits advanced materials or artificial intelligence, position for it now. The $1 billion is not being destroyed. It is being aimed. Grand Research Challenges is oriented toward exactly those fields, and the FY2028 R&D budget priorities memorandum, "Ushering in a New Golden Age of American Innovation," names foundational research in those domains as the top objective. A materials group that can credibly frame its program against a national challenge statement is better positioned than one that cannot — not because the framing is honest signaling, but because that is where the obligation authority actually sits.

The structural read

The federal research funding system has historically been slow but predictable. Success rates moved a few points a year. Program budgets grew or shrank at the margins. A PI could build a decade-long research program around that predictability.

What changed in 2026 is not the size of the pot. NSF's appropriation is roughly what it was. What changed is who controls the allocation and on what timeline — and the answer is increasingly a central account whose disbursement schedule is not published, not appealable, and not tied to any panel outcome.

Plan accordingly. The proposal you write is still the part you control. The calendar you write it against is not.