Trump's NSF Funding Cuts Threaten Innovation

Trump's 54% NSF Cut Would Gut US Science Even If Congress Says No

Trump's FY2027 budget proposes cutting the National Science Foundation by 54% to $3.1 billion — but the deeper threat is the agency being dismantled from within.

President Trump's April 3, 2026 budget request asks Congress to cut the National Science Foundation to $3.1 billion — a 54% reduction and the deepest single-year cut proposed for a US science agency since the Cold War. That request will almost certainly be sanded down by appropriators, as it was last year. The real story is that the top line barely matters, because the administration has spent 18 months hollowing out the agency's grant pipeline, oversight board, and internal structure regardless of what Congress appropriates — and the primary beneficiaries are China's basic-research complex and European universities now recruiting displaced American scientists.

The number, and what it actually buys

The White House's FY2027 request, released April 3, 2026 by Office of Management and Budget director Russell Vought, would cut roughly $3.9 billion from NSF's current appropriation. According to the OMB budget document, the reduction targets "climate; clean energy; woke social, behavioral and economic sciences; and programs in low priority areas of science." NSF would retain a $100 million joint DOE–NSF Energy-Water Security line and modest funding for AI, quantum, and critical-minerals research.

That is not the first attempt. In FY2026 the administration asked for a 55% cut — from roughly $9 billion down to $3.1 billion. According to the Congressional Research Service, Congress rejected almost all of it: NSF received $310 million less than FY2025, a 3.4% haircut rather than a beheading, and $4.847 billion more than the president asked for. The Senate committee bill had proposed only a 0.7% cut; the House committee bill would have cut 22.7%. Appropriators split the difference well above the executive's number.

The Center for Strategic and International Studies notes the FY2027 request repeats that pattern alongside a 42% defense increase — the largest defense budget request since World War II, exceeding $1.5 trillion — while cutting $3 billion from basic defense research and $1 billion from applied defense research. The pattern also holds government-wide: under the FY2026 request, according to CRS R&D funding analysis, NSF faced the largest percentage cut of any major agency at -55%, NIH would have fallen $19 billion (-41%), and the Department of Defense would have absorbed 62% of all federal R&D spending.

What the White House is doing without waiting for a vote

Congress writes the top line. The executive branch controls how, when, and to whom the money flows — and that is where the story has moved.

Since Trump's return, NSF has cancelled or suspended roughly 2,000 grants worth about $1 billion. A Nature analysis of NSF Award Search data pegged the 2025 total at 1,996 NSF grants terminated or frozen, alongside 5,843 at the National Institutes of Health. Total federal grants stopped or paused across 2025 topped 7,800, with roughly $1.4 billion in unspent funding still frozen at year-end. Nature also documented that new NSF grants issued in 2025 fell 25% below the 10-year average, and that terminations disproportionately targeted work on misinformation, vaccine hesitancy, infectious disease, and researchers from under-represented backgrounds.

The tempo was set immediately. An independent analysis published by Science found that in the two months after Trump's January 20, 2025 inauguration, NSF made 919 new awards versus 1,707 in the same window of 2024 — a 46% drop, and roughly $400 million in withheld obligations. The engineering directorate obligated $20 million versus $133 million a year earlier. The education directorate made 12 awards, down from 120. The Directorate for Technology, Innovation and Partnerships — created by the 2022 CHIPS and Science Act specifically to compete with China on AI, quantum, and semiconductors — obligated $19 million, one-tenth its year-earlier total. These are the exact program areas the administration says it wants to protect.

Then the agency itself was restructured. According to Science, NSF in April 2025 abolished all 37 of its divisions across eight directorates in a single memo from acting chief management officer Micah Cheatham, cutting senior executive positions from 143 to 59 — a 60% reduction — to match "the needs of the agency" under a shrunken future budget. Eleven external advisory committees were "disestablished" under Trump's February 19, 2025 executive order on the federal bureaucracy, according to a separate Science report; only two committees survived because Congress had explicitly required them. NSF director Sethuraman Panchanathan, a first-term Trump appointee, resigned in April 2025 16 months before the end of his six-year term, writing that he had "done all I can." Chief of staff Brian Stone has run the agency on an acting basis since. The Senate is still processing the nomination of James O'Neill of Texas as permanent director, sent in March 2026.

Then, on Friday, April 24, 2026, the White House terminated all 22 seated members of the National Science Board — the statutory body that approves NSF's programs and $8.8 billion in awards — in a single email. NPR reported that the White House cited the 2021 Supreme Court decision United States v. Arthrex as constitutional grounds, arguing that non-Senate-confirmed appointees cannot exercise the authority Congress delegated to the board. Board members appointed by both parties, including Trump's own 2020 reappointee Roger Beachy, said they had received no reason. Sociologist Alondra Nelson had already resigned in May 2025, writing in Time that the board had become a "ceremonial assemblage" observed passively by DOGE staffer Zachary Terrell — the first policy resignation in NSF's 75-year history. Al Jazeera reported that Roger Beachy told the outlet that no rationale was given for the mass dismissal, and Democratic committee members learned of it from unofficial channels.

The courts have slowed one front — and only one

There is a single clean loss for the administration to log. On May 2, 2025, NSF issued Policy Notice 25-034 capping indirect-cost reimbursement on grants to universities at 15%, matching earlier moves at NIH and the Department of Energy. Indirect costs — facilities, administration, compliance, shared instrumentation — typically run 50-70% at research universities and are the mechanism through which federal grants sustain lab infrastructure.

The Association of American Universities and co-plaintiffs sued three days later. In their complaint and preliminary-injunction motion, the universities argued Congress never authorized NSF to impose an arbitrary cap unmoored from actual indirect costs, and that the 15% floor would "grind vital scientific work to a halt" in biomedical research, cybersecurity, quantum computing, and domestic mineral extraction. The government replied that NSF's organic statute and 41 U.S.C. § 4708 granted the discretion, and that the policy applied only to future awards. On June 20, 2025, Judge Indira Talwani of the District of Massachusetts granted summary judgment against the government, vacating the policy as "arbitrary and capricious, and contrary to law". Parallel NIH and DOE rate-cap policies were enjoined earlier in the same courthouse.

But the injunctions apply to the rate cap alone, not to the underlying budget authority, the grant-termination discretion, the personnel decisions, or the restructuring. The vast majority of the administration's dismantling is out of reach of the Talwani ruling.

Who wins

The visible loser is US higher education. The less visible winners are two.

The first is the Chinese state R&D complex. A September 2025 CSIS analysis drawing on OECD data found China's government-intramural research spending in 2023 was more than 1.5 times US levels. The 2025 Nature Index recorded China's high-quality publication share at 32,122 to the US's 22,083. From 2020 to 2024 the US's count of "high-level" scientists — authors of highly cited papers in top journals — fell from 36,599 to 31,781, while China's rose from 18,805 to 32,511, surpassing the US for the first time. CSIS notes Chinese basic-research appropriations grew 16.3% in the most recent budget cycle, and that Chinese firms can employ 2.3 researchers for every $100,000 of R&D spend — more than double the US figure. The relative-advantage math tilts further with every quarter US grants freeze.

The second beneficiary is European recruitment. Aix-Marseille Université's "Safe Place for Science" program received 298 applications by mid-April 2025, NPR reported — 135 from Americans plus 45 dual nationals, from Johns Hopkins, NASA, Columbia, Yale, and Stanford. CentraleSupélec announced a €3.2 million matching fund; Dutch Education Minister Eppo Bruins wrote to parliament seeking authority to establish a fund to lure senior international scientists. A Nature-cited survey documented by Science drew more than 1,200 US-based researchers citing Trump cuts as reason to consider a move. McGill's Richard Gold, quoted by Science, warned of losing "a cohort of researchers" — the kind of pipeline gap that shows up not in this year's papers but in the next decade's patents.

The startup exposure

The commercial ecosystem the White House says it wants to protect — AI, quantum, semiconductors — is unusually dependent on the funding streams being cut. NSF is a small share of total federal R&D but the dominant funder of university basic research in mathematics, computer science, and the physical sciences.

That composition matters because private capital does not fund the pre-competitive research feeding those industries. CSIS observes that the atomic clock — the physics behind GPS, and therefore behind every precision-guided munition — traces to NSF- and NIST-funded work in the 1960s. Every current US quantum computing company originated in a university group whose graduate students were paid on NSF fellowships. According to Science, the Graduate Research Fellowship Program (GRFP), the largest such program in the country, was cut roughly in half for 2025 — to about 1,000 fellows from a typical 2,000-plus — and NSB members openly discussed corporate naming rights for individual fellows to stretch a shrinking $350 million pot. NSF also cancelled its GRFP grant to Harvard on May 12, 2025, leaving some 260 incoming fellows in limbo as part of Trump's separate campaign against the university.

"If NSF is going to sap research at institutions of higher education, and handicap our competition with (for example) China over AI, it owes more." — Plaintiffs' reply brief, Association of American Universities v. NSF, Doc. 63 (May 2025)

The historical parallel is instructive. Trump's proposal would cut NSF real dollars below the level of the mid-1990s — a period in which China's total R&D was under 5% of US spending. The gap that Trump inherited took thirty years to build. The share it took to erode has taken 18 months.

What to watch

• FY2027 CJS appropriations markup (Senate, autumn 2026) — the Senate committee's opening bid will signal whether Republican appropriators back a repeat of last year's rebuke or move partway toward the White House's number. House Science chair Brian Babin has publicly defended the administration's direction; Senate Commerce-Justice-Science chair Jerry Moran has not.
• O'Neill confirmation vote — if the Senate confirms James O'Neill as NSF director, expect a formal restructuring plan aligned with FY2027 priorities. If the nomination stalls, Brian Stone's acting tenure — and the pattern of policy-by-memo — continues.
• National Science Board replacement appointments — whether Trump appoints new members subject to Senate confirmation (as the Arthrex rationale would require) or leaves the board vacant will determine whether NSB has any independent voice on grants in 2027.
• NSF FY2026 grant-obligation close-out (September 30, 2026) — year-end totals will show whether the pace picked up after the FY2026 appropriation locked in Congress's higher number, or whether the executive is de facto impounding funds.

Diplomat View

The FY2027 budget request is a signaling document, not a forecast. Congress will not enact a 54% cut; last year's 3.4% outcome is closer to the ceiling of what appropriators will tolerate, and defense hawks in both parties know NSF-funded physics underwrites the weapons stack. The operative risk is that the top line becomes irrelevant. The administration has demonstrated across 18 months that grant terminations, indirect-cost caps, division abolitions, board firings, and personnel exits can each be executed without a single vote — and only one of those levers, the rate cap, has been reversed in court. If new NSB appointments never arrive and O'Neill's confirmation stalls, expect governance-by-memo to harden into structure. The forecast reverses only under two conditions: a bipartisan appropriations rider explicitly ring-fencing NSF grant terminations and personnel actions, or a Trump-appointed NSF director with standing to defend the pipeline from inside. Absent either by Q1 2027, the US basic-research lead over China narrows by default, not by choice — and the next decade's frontier-technology companies incubate somewhere else. *