NSF Funded Three Science and Technology Centers, Not Four. The Per-Center Check Held at $30 Million — the Number of Winners Did Not.

The NSF Science and Technology Centers program is the closest thing American basic science has to a moonshot slot machine. Roughly every three years, a handful of university consortia win $30 million over five years, renewable to ten, to try to start a field rather than advance one. Winning one is career-defining. Most institutions never do.

On August 26, 2026, NSF announced the class of 2026: three centers, $90 million over five years, approximately $6 million per center per year, each with the opportunity to compete for up to five additional years.

The number to sit with is not $90 million. It is three.

The three centers

TEMPEST — Center for Transformative Explorations in Multi-Physics and Engineering of Scientific Turbulence Lead: Michigan State University

Turbulence is the canonical unsolved problem in classical physics — the thing that makes fusion plasmas leak, aircraft shake, and weather models diverge. TEMPEST attacks it with a four-way stack: advanced experiments, mathematical modeling, high-performance computing, and artificial intelligence. NSF's framing points the applications at fusion energy and national security technologies.

GENIE — Center for Genome Intelligence Engineering Lead: Northwestern University

GENIE works on the three-dimensional organization of DNA inside the nucleus — how the genome is physically folded, and how that folding degrades with age and disease. The stated target is preventing and reversing cellular decline implicated in Alzheimer's disease, cardiovascular disease, and cancer, through what NSF describes as 3D DNA engineering.

Center for Human and Robot Co-Adaptation Lead: The University of Texas at Austin

Not robots that work around humans — robots and humans that adapt to each other. The center combines robotics, AI, and human factors research toward secure, adaptive systems for homes, hospitals, and workplaces, with an explicit emphasis on assistive robotics and independent living.

Brian Stone, performing the duties of NSF Director, framed the class this way: "These Science and Technology Centers will help American researchers seize opportunities, accelerate scientific discovery, and develop talent and technologies."

Read the three topics together and the portfolio logic is unmistakable: fusion, biotechnology, and robotics — three of the named critical and emerging technology areas in current federal science policy. This is not a random draw from the frontier. It is a deliberate alignment.

The arithmetic that matters

Compare classes:

The per-center award held exactly. NSF did not shave the checks. It cut the number of checks by 25 percent.

That distinction is strategically enormous, and it gets lost every time coverage leads with the total. An agency under budget pressure has two levers: fund the same number of teams at lower amounts, or fund fewer teams at full strength. NSF chose the second. For STCs — where the entire premise is that some problems need large-scale, long-horizon, multi-institution structure that ordinary grants cannot buy — diluting the award would have defeated the instrument. Cutting the count preserves it.

The consequence for applicants is brutal and simple: the bar moved up, not the budget down. You are not competing for a slightly smaller prize. You are competing for one fewer seat at the same table.

The program solicitation, NSF 24-594, anticipates "approximately 5" awards. Three landed. Whatever the internal reasons, plan against the realized number, not the solicitation's estimate.

What the STC competition actually demands

The structural constraints in NSF 24-594 shape the strategy more than the science does:

• Preliminary proposals are mandatory, and only invited teams may submit full proposals. Two-stage competitions with an invitation gate historically compress an enormous applicant pool into a small finalist set.
• A lead institution may submit at most three preliminary proposals per competition. Your university's internal limited-submission process is therefore the first real review you face — and for many faculty, the one they actually lose.
• A PI or co-PI may appear on only one STC proposal per cycle. Senior people are a rate-limited resource. The scramble to lock in a marquee co-PI starts long before the solicitation reopens.
• Awards are cooperative agreements, not standard grants. NSF stays operationally involved for a decade.
• Every center must integrate research, education, and broadening participation, plus knowledge transfer to other sectors and to policymakers. These are not appendices. In a field of finalists whose science is uniformly excellent, they are frequently where the ranking is decided.

The August NSF surge, and what it tells you

The STC announcement did not arrive alone. NSF ran an extraordinary month:

• August 3 — $50 million for two Materials Innovation Platforms
• August 4 — State and Regional AI Infrastructure Hubs launched
• August 17 — 12 notices of funding opportunity totaling over $1.5 billion for foundational research, most with no submission deadline at all
• August 25 — more than $290 million to eight Quantum Leap Challenge Institutes, spanning 36 institutions across 19 states and 30-plus companies, with three new institutes and five renewals at roughly $28–37 million each
• August 26 — the three STCs

That is well over $1.9 billion announced in twenty-four days, and almost all of it flows through large, multi-institution, center-scale instruments.

Set that against the other half of the picture. The FY2027 President's Budget Request proposed $4 billion for NSF, a 54 percent cut from the FY2026 enacted level, and projected 2,100 research grants in FY2027 against 5,800 in FY2025 — while raising average award size by about 14 percent. Congress rejected a nearly identical proposal the prior year and funded the agency at $8.75 billion. Meanwhile NSF's own FY2026 obligations have run well below historical pace, with roughly 6,100 grants and a 46 percent decline documented earlier this year.

The pattern across the budget request and the August awards is the same in both directions: fewer, larger, more centrally directed awards. Whether or not Congress ratifies the FY2027 numbers, NSF's revealed behavior in August 2026 is consistent with that trajectory.

Strategy: three plans, depending on who you are

If you lead a center-capable institution. Start the next STC cycle now, and start it internally. With three preliminary proposals per institution and one-proposal-per-person limits on senior investigators, your bottleneck is not NSF review — it is your own limited-submission committee and your ability to commit named co-PIs eighteen months ahead. Institutions that win STCs almost always ran a disciplined internal down-select a year before the solicitation dropped. Institutions that lose usually assembled a team in the eight weeks after it did.

If you are a researcher who will never lead one. The right move is not to chase a center. It is to become an obviously valuable node in someone else's. STCs are multi-institution by design and carry substantial subaward budgets across a decade. Three specific openings just appeared:

• TEMPEST needs turbulence-adjacent computational, experimental, and ML capability. If you work on plasma diagnostics, high-fidelity simulation, surrogate modeling, or fusion-relevant transport, Michigan State is now assembling a ten-year collaboration network.
• GENIE needs chromatin structure, nuclear imaging, aging biology, and genome-engineering expertise. Northwestern will be building out biomedical partnerships against Alzheimer's, cardiovascular, and cancer aims.
• UT Austin's robotics center needs human factors, assistive technology, clinical deployment partners, and security expertise. Hospitals and independent-living organizations are plausible non-academic partners here, not just CS departments.

Center leadership teams are most receptive to new collaborators in the first twelve to eighteen months, while the research plan is still elastic and the education and broadening-participation components are being staffed. Approach with a specific capability and a specific deliverable, not a general offer to collaborate.

If you were building a proposal around a deadline. Note what else August did: NSF's $1.5 billion foundational research reset moved a large share of core program submissions to accepted-anytime. Combined with the center-scale surge, the agency is simultaneously removing calendar structure from small awards and concentrating money in big ones. A lab that organized its year around a core program deadline and a center bid now has neither anchor. Rebuild the plan around program officer contact, not the calendar.

The bottom line

Three centers, $30 million each, in fusion-relevant turbulence, 3D genome engineering, and human-robot co-adaptation. The award size survived a punishing budget environment fully intact — which is the most encouraging thing in the announcement.

The count did not. Four became three, in a program whose solicitation contemplates five. If you are planning a center bid for the next cycle, plan for a field where roughly one in every institution's three allowed preliminary proposals must be the one, and where a single senior investigator can only be on one team.

That is not a funding cut. It is a selectivity increase, and it demands a different kind of preparation.

Sources: NSF — three new Science and Technology Centers with $90M investment, NSF STC program page and solicitation NSF 24-594, NSF — eight quantum institutes, $290M, NSF — $1.5B foundational research NOFOs, AIP FYI — FY27 budget request for NSF, Science — NSF picks new Science and Technology Centers.