The Last of the Twelve: NSF Physics Is $180 Million, 300 Awards, and a Deadline That Now Lives on a Program Page

The twelfth and final piece of NSF's August 17 foundational research reset is also the largest single check in the Directorate for Mathematical and Physical Sciences. NSF 26-523, MPS Physics Research Programs, carries $180 million and an estimated 300 awards — more money than mathematics, materials or astronomy, and more awards than any MPS solicitation except mathematical sciences.

Do the division before you do anything else. $180 million across 300 awards is roughly $600,000 per award in total cost, which over a standard three-year project is about $200,000 a year, and over five years about $120,000 a year. That is one graduate student, part of a postdoc, and travel — in a field where a single cryostat, a beamline run or a detector test stand can consume a year's budget. The headline number is big. The unit of funding is not.

This is the last solicitation in the set we have been working through since NSF replaced its core research portfolio wholesale: the $1.5 billion, twelve-NOFO reset that deleted submission deadlines across the sciences. Physics is where the consequences of that design are clearest, because physics is the directorate with the widest spread between a single-investigator theory grant and a multi-institution instrumentation build — and NSF has now put both in the same document.

Fifteen programs, one solicitation, two dead predecessors

NSF 26-523 replaces NSF 23-615 (Division of Physics: Investigator-Initiated Research Projects) and NSF 23-607 (the theoretical physics hubs solicitation). What used to be two competitions with different review cultures is now one.

Twelve core programs live inside it:

• Atomic, Molecular and Optical Physics — Experiment (AMO-Experiment)
• Atomic, Molecular and Optical Physics — Theory (AMO-Theory)
• Elementary Particle Physics — Experiment (EPP-Experiment)
• Elementary Particle Physics, Astrophysics and Cosmology — Theory (EPP/Cosmology-Theory)
• Gravitational Physics
• Integrative Activities in Physics (IAP)
• Nuclear Physics — Experiment (NP-Experiment)
• Nuclear Physics — Theory (NP-Theory)
• Particle Astrophysics — Experiment (PA-Experiment)
• Physics of Living Systems (PoLS)
• Plasma Physics
• Quantum Information Science (QIS)

Three cross-cutting programs sit alongside them:

• Advanced Physics Instrumentation (API) — "the acquisition, development and/or implementation of the advanced instrumentation needed for NSF MPS physics research programs."
• Focused Research Hubs in Theoretical Physics (FRHTP) — team awards to "advance cutting-edge foundational research in theoretical physics, enhance the impact of the field and foster the development of the next generation."
• Physics at the Information Frontier (PIF) — science-driven cyberinfrastructure at "the convergence of big data and high-performance computing."

The consolidation of FRHTP is the structural change most likely to be underestimated. Theoretical physics hubs previously competed against each other in their own solicitation with their own budget line. They now sit inside a $180 million envelope shared with 300 single-investigator awards. Nobody has published the internal split. Every dollar that goes to a hub is a dollar that does not go to a PoLS grant, and the tradeoff is now made inside one program office rather than across two competitions.

The deadline is gone. The dates are not.

NSF 26-523 lists "Proposals Accepted Anytime" where a due date used to be. That is not the same as saying timing does not matter, and physicists in particular should resist reading it that way.

Individual programs retain target dates, published on program webpages rather than in the solicitation:

• AMO-Experiment targets the fourth Wednesday in November — November 25, 2026 for fiscal year 2027.
• Gravitational Physics targets the same fourth Wednesday in November — November 25, 2026.
• Quantum Information Science targets December 8, 2026.

Other programs carry their own dates. This is the operational trap in the new architecture: the authoritative date for your submission is no longer in the PDF you downloaded. It is on a webpage that NSF can revise without reissuing anything. A lab that prints the solicitation in September, works to it through the fall, and never re-checks the program page is working from a document that is deliberately silent on the one fact it needs.

Build the habit now. Before you commit to a submission month, open the program page for the exact program you are targeting, screenshot the target date, and re-check it thirty days out. Astronomy's sibling solicitation, NSF 26-522, was explicit that late proposals may be reviewed in the following fiscal year. Assume physics program officers will manage panels the same way, because panels are the constraint — a proposal that misses a panel waits for the next one.

No proposal cap — and that is not a small thing

Read across MPS and the inconsistency is striking. MPS Materials imposes a one-proposal-per-fiscal-year limit. MPS Chemistry allows two per senior person per year. MPS Physics imposes no PI or co-PI proposal limits at all.

For a physicist with a foot in more than one program — and there are many, since AMO overlaps QIS, gravitational physics overlaps particle astrophysics, and PoLS overlaps half of biophysics — the absence of a cap is real freedom. It also comes with a warning shot: NSF "strongly encourages single proposal submission for possible co-review rather than submission of multiple related proposals to several programs."

Translate that. If you submit three variations of one idea to three programs, you have not tripled your odds; you have created three program officers who each discover the other two, and you have signaled that you could not decide what the project is. The intended path is one proposal plus a conversation with program officers about co-review, so that two programs share the cost of one award. That is the mechanism that quietly funds interdisciplinary work at NSF, and it only works if you ask for it in advance.

The place where the cap-free rule genuinely helps is a distinct second project with distinct personnel — an instrumentation proposal to API alongside a science proposal to your home program. Those are not variations. They are two projects, and 26-523 lets you file both.

Three compliance changes that will fail proposals mechanically

Voluntary committed cost sharing is prohibited. If your sponsored programs office uses a boilerplate budget template with a cost-share line — and many do, as a holdover from solicitations that rewarded it — that line now makes your proposal non-compliant. Check the template before the first internal deadline, not the week of submission.

Indirect cost rates are now a stated tiebreaker. Under Executive Order 14332, NSF states that all else being equal after merit review, it may give preference to institutions with lower indirect cost rates. This language now appears across the 2026 NSF solicitation set, not just physics. You cannot renegotiate a NICRA in a quarter. What you can control is how much of the total request is visible as science: a leaner scope with a smaller total ask carries a smaller absolute indirect burden, which is what a program officer comparing two fundable proposals actually sees. Our full breakdown of the order is in Executive Order 14332 Is Rewriting the Rules of Federal Grantmaking.

Awards may include termination for convenience provisions. This is the same clause moving through the October 1 Uniform Guidance rewrite, and it changes how a five-year physics project should be structured. Front-load a deliverable into year one. Build the apparatus in a sequence where each stage produces a publishable result rather than only a milestone. A project whose entire scientific output arrives in year four is a project whose entire scientific output is contingent on four years of appropriations behaving.

The new portfolio posture: comparative review and phase-out

Buried in the long-duration language is a shift in how NSF intends to manage its physics portfolio. The solicitation invites extended-duration projects — "proposals for such activities should indicate expected project duration and milestones" — and then reserves the right to conduct comparative reviews assessing relative priorities within a program, with possible recommendations to phase out support for a long-duration effort.

Additionally, proposals to API or for long-duration efforts "may be evaluated by NSF using site visits."

Read together, this is portfolio management stated out loud. Long-running experiments are no longer renewed on their own merits in isolation; they are ranked against the other long-running experiments in the same program, and a favorable individual review does not guarantee continuation if a peer effort ranks higher. Site visits give NSF the evidentiary basis to make that call.

If you run a multi-year effort, the practical response is to stop writing renewal narratives that describe what you have done and start writing them as a case for relative priority: what this effort produces that the alternatives in the same subfield do not, what the phase-down plan looks like if support ends, and what the milestones are that would justify continuing. Assume a program officer is holding your document next to two others.

What to do in the next sixty days

1. Identify your exact program, singular, and pull the target date from its NSF program page today. Do not work from the solicitation for dates.
2. If you target AMO-Experiment or Gravitational Physics, your working date is November 25, 2026. If you target QIS, it is December 8, 2026. Both mean internal review in October.
3. Email the program officer about co-review if your project straddles two programs. Do this before you write, not after you submit.
4. Audit the budget template for a voluntary cost-sharing line and delete it.
5. Scope to about $600,000 total unless you have a specific reason to be an outlier, and be able to defend why you are one.
6. Sequence deliverables so year one produces a result, given termination-for-convenience exposure.

Three hundred awards against a national physics community is not generous. But the structure of 26-523 rewards a specific kind of applicant: one who picks a single program, hits its target date, asks for co-review rather than shopping variants, and writes a budget that looks like it was built for the money that exists rather than the money that used to.