Congress Approved $8.75 Billion for NSF. Then DOGE Terminated 1,752 Grants Worth $1.4 Billion.

The National Science Foundation has $8.75 billion in Congressional backing for fiscal year 2026. It also has 1,752 fewer active grants than it did a year ago — terminated not by Congress, not by the agency's merit review process, but by the Department of Government Efficiency.

These two facts coexist uneasily, and together they define a funding environment that no researcher navigating NSF has ever encountered: a budget that survived political assault from one direction and a research portfolio under active demolition from another.

The numbers tell the story. Congress signed the FY2026 appropriations bill on January 23, 2026, decisively rejecting the White House's proposed 55% budget cut that would have slashed NSF to $3.9 billion. The final figure — $8.75 billion, with $7.18 billion for research activities — represents a 3.4% decrease from FY2024. A cut, but a manageable one. Senate Appropriations Chair Susan Collins called the package "fiscally responsible" while maintaining "essential federal investments." The bipartisan message was clear: American science funding would not be gutted on Congress's watch.

But DOGE was already inside the building.

1,752 Grants and the Anatomy of a Purge

Between late 2025 and early 2026, DOGE operatives working within NSF identified and terminated 1,752 grants worth approximately $1.4 billion. The terminations did not follow the agency's standard processes for award management. They were targeted by category, not by scientific merit.

The heaviest blow fell on the STEM Education directorate (EDU). Of the 1,752 terminated grants, 839 came from EDU — nearly half of all terminations — representing $888 million, or 63% of the total dollar value lost. The Social, Behavioral, and Economic Sciences directorate (SBE) lost another 320 grants worth $91 million. Together, these two directorates accounted for 66% of all terminated grants by count and more than 70% by dollars.

The targeting was not random. DOGE flagged grants categorized under "diversity, equity, and inclusion research" and "studies related to misinformation and disinformation" — categories that, in practice, swept up everything from broadening participation programs that recruited underrepresented students into STEM fields to research on information quality in digital ecosystems. A longitudinal study tracking retention rates for first-generation college students in engineering could fall under the DEI umbrella. A computational social science project modeling how false health claims spread during pandemics could be classified as misinformation research. The categories were broad enough to capture far more than their labels implied.

The scale of the education losses is historically unprecedented. The EDU directorate's entire FY2025 budget was approximately $1.1 billion. Losing $888 million in terminated grants represents the functional equivalent of wiping out nearly an entire year of the directorate's funding — except these were multi-year awards to institutions and researchers mid-project, with students hired, equipment purchased, and partnerships built.

An Agency Without a Director

The terminations unfolded against a backdrop of institutional instability. NSF Director Sethuraman Panchanathan resigned on April 24, 2025. As of early 2026, no permanent replacement has been nominated. Brian Stone, the agency's Chief of Staff, has been performing director duties — a political appointee running an agency whose entire identity rests on the independence of its scientific judgment.

This leadership vacuum matters for a specific reason. NSF's merit review process — the system by which proposals are evaluated and funded based on scientific quality — has been the agency's core differentiator since its founding in 1950. That process is now being fundamentally restructured. The minimum number of external reviews has dropped from three to two, with one review permitted to come from internal NSF staff. Panel discussions are now optional rather than required. Review summaries have been compressed from multiple paragraphs to three-to-five sentences. And program officers have gained substantially more discretion over which proposals get funded.

Each of these changes, individually, could be defended as administrative modernization. Combined with 1,752 politically targeted terminations and an acting director, they create an environment where the traditional firewall between political priorities and scientific funding decisions is thinner than at any point in NSF's 76-year history.

What the Budget Actually Buys Now

The $8.75 billion appropriation is real money, and it will fund real science. NSF expects to make approximately 10,000 new awards in FY2026, supporting more than 250,000 scientists, technicians, teachers, and students. Some program-level changes benefit researchers: the RAPID award ceiling increased to $300,000 (from $200,000), EAGER proposals can now request up to $400,000, and the equipment purchase threshold rose from $5,000 to $10,000.

But the effective purchasing power of the budget has been diminished by the terminations. When $1.4 billion in active grants disappear, the researchers who held them do not disappear with them. They return to the applicant pool. Graduate students whose funding was cut need new support. Institutions that lost mid-project awards seek replacement funding. The demand on the remaining $8.75 billion increases even as the supply holds steady.

NSF funded approximately 8,800 new grants in FY2025 — 20% fewer than the roughly 11,000 it funded in FY2024. The 10,000 new awards projected for FY2026 would represent a recovery, but only if the applicant pool hasn't expanded faster than the award count. Early indications suggest it has.

New Compliance Requirements Add Friction

FY2026 also introduces new compliance obligations that add administrative overhead for applicants and recipient institutions. Research security training is now mandatory for all senior and key personnel — a requirement driven by the CHIPS and Science Act's provisions on foreign influence in federally funded research. Data sharing at the time of publication is newly required, reflecting a broader push toward open science that has been building across federal agencies.

These are not unreasonable requirements in isolation. But they arrive at a moment when institutional research offices are already stretched thin — processing termination paperwork, managing budget uncertainty, and adapting to the merit review overhaul. The cumulative compliance burden falls disproportionately on smaller institutions, where the grants office may be one or two people managing dozens of active and pending awards.

The Two-Track Funding Reality

What makes the current NSF landscape genuinely unprecedented is the structural contradiction at its center. Congress looked at the White House's proposed 55% cut, rejected it, and funded NSF at near-historical levels. That is an extraordinary act of bipartisan institutional defense — one that suggests strong Congressional support for federal science regardless of which party controls the executive branch.

Simultaneously, executive-branch actors operating through DOGE bypassed Congress entirely, terminating $1.4 billion in active awards without Congressional authorization and without following the agency's merit-based processes. The legal basis for these terminations remains contested. Multiple lawsuits have been filed challenging DOGE's authority to terminate grants that Congress appropriated funds for.

For researchers, this two-track reality creates a specific strategic challenge. The traditional advice for navigating NSF funding — write the strongest possible proposal, align with published program priorities, build a track record of prior NSF work — remains correct. But it is now incomplete. Researchers must also consider whether their work falls into categories that could be politically targeted. This does not mean avoiding important research areas. It means understanding the risk landscape and positioning proposals accordingly.

What Grant Seekers Should Do Right Now

For active NSF awardees: Review your award terms and conditions. The revised 2 CFR 200 Uniform Guidance introduced termination-for-convenience clauses that give agencies broader authority to end awards. Ensure your institution's research office is tracking any communications from NSF about programmatic reviews or policy compliance.

For prospective applicants: The 10,000 new awards NSF plans to make in FY2026 represent genuine opportunity. Focus on directorates that emerged relatively unscathed from the DOGE terminations — Engineering (ENG), Computer and Information Science and Engineering (CISE), and Mathematical and Physical Sciences (MPS) were far less affected than EDU and SBE. If your research falls in education or social science, frame proposals around technical outcomes and national competitiveness rather than equity language that could trigger political screening.

For STEM education researchers: Diversify your funding base immediately. The EDU directorate lost 839 grants. Even if new competitions are announced, the backlog of displaced researchers competing for those slots will be intense. Consider state-level research funding programs that have expanded in response to federal uncertainty, as well as private foundations that have increased payouts to offset federal gaps.

For everyone: Complete the new research security training requirement early. Institutions that delay compliance risk administrative holds on pending awards. And watch the director nomination process — whoever is confirmed as the permanent NSF director will set the tone for how aggressively the agency pushes back against political interference in merit review.

The $8.75 billion is real. The 1,752 terminations are also real. Navigating both realities simultaneously — competing for new funding while protecting against politically motivated disruption — is the defining challenge of NSF grant seeking in 2026. Tools like Granted can help you identify where the funded opportunities are and build proposals that compete on scientific merit in an environment where merit alone may no longer be sufficient protection.