The Federal Science Brain Drain: A Generational Loss, in Numbers

Last Updated: June 10, 2026 Reading Time: 10 min

A nonprofit watchdog put a number on what federal scientists have been describing anecdotally for a year. Between September 2024 and February 2026, federal science agencies lost nearly 118,000 employees, about 40% of all federal workforce cuts, even though science workers are a small slice of government. The Partnership for Public Service's June 9 report calls it a "generational loss," and the phrase is precise rather than dramatic: the federal science brain drain emptied both ends of the career pipeline at once. Here is the agency-by-agency data, the grant collapse it's already causing, and what it means whether you're still inside or recently out.

Key Takeaways

• 118,000 science-agency employees left between September 2024 and February 2026; science absorbed ~40% of all federal cuts.
• 10,109 PhDs walked out in 2025 alone, 14% of the government's entire doctoral science workforce, with departures outnumbering hires 11-to-1 at 14 research agencies.
• The losses are top-and-bottom: senior experts took buyouts (USGS career SES posts fell from 27 to 5) while entry hiring collapsed (DOE first-year scientists: 116 down to 12).
• The grant machine is already seizing: NIH competitive awards are down 66%, funding announcements fell from 756 to 14, and FDA on-time drug approvals slipped to 78%.
• The political layer grew while the career layer shrank: non-confirmed political appointees at NIH increased roughly tenfold, and government-wide the political workforce is the largest in 40 years.
• 75% of U.S. scientists told Nature they're considering leaving the country, which is the rebuild problem in one statistic.

The Agency-by-Agency Damage Table

No single source publishes this; we synthesized it from the Partnership report, Science/AAAS's OPM-data analysis, the American Institute of Physics, and Eos. Measurement windows vary slightly by source.

Agency	Staff loss	The expert layer	Main driver
SAMHSA	-41.7% (~900 → under 60 staff)	Most professional staff gone, still managing $3B in grants	Near-total reduction
NPS	-34 to -37% (7,366 people)	—	Buyouts + firings
NSF	-18 to -33%	-40% of PhD workforce	Buyouts; career scientists replaced by appointees
CDC	-22 to -25%	Senior epidemiologists across global health, NIOSH, maternal health	Four cut waves
NIH	~-20% (4,000+)	1,100+ PhD exits in 2025 vs 421 in 2024	DRP, RIF, voluntary exits
FDA (drug centers)	~-20%	~1,000 net departures in Q4 FY2025 alone	Cuts + buyouts
USGS	~-20% (1,641)	Career SES fell 27 → 5	Buyouts + freeze
NOAA	-16 to -20%	Hurricane Hunter leadership among losses; 3,000 vacancies frozen	Buyouts + firings
USDA / Forest Service	-18 to -33%	Recent-graduate scientist cohort lost	Buyouts + firings
NASA	-12 to -24% (4,300)	2,000+ senior staff	Mostly DRP
DOE	~-15%	First-year scientists: 116 → 12	Buyouts
NIST	~-17% (700+)	—	Mixed separations
EPA	~-10% (1,673)	1,000+ research-division scientists targeted	Office closures + firings

Sources: Partnership for Public Service (June 2026), Science/AAAS analysis of OPM data (April 2026), AIP, Eos/AGU.

Why "Generational" Is the Accurate Word

A normal downsizing thins the middle of a workforce. This one removed the top and sealed the bottom.

The top left. The federal workforce was already old (42.5% over age 50 in 2023, versus 33% of the U.S. labor force), so buyout offers landed on a workforce full of retirement-eligible senior experts, and they took them: more than half of separations at DOE, NSF, and USGS came through the deferred resignation program. These are the people who carry the institutional knowledge: one genome-research division lost two-thirds of its staff, and its former director said plainly that there aren't enough people left to do the work.

The bottom never arrived. The hiring freeze that began in January 2025 held new hires at science agencies below 2% of headcount. Departures beat arrivals 11-to-1. The early-career collapse is the starkest number in the dataset: DOE employed 116 physical scientists with under a year of service in 2024, and 12 in 2025.

That combination is what can't be fixed by a future administration posting vacancies. You can hire a workforce; you can't hire a 20-year experience distribution. Pipeline math says the rebuild runs a decade or more, and that assumes the talent wants to come back: in Nature's poll, 75% of U.S. scientists said they're considering leaving the country, with postdocs at 79%.

The Damage You Can Already Measure

The brain drain stopped being theoretical the moment the grant machinery slowed:

• NIH competitive grant awards are down 66% in FY2026 versus the FY2021-2024 average.
• Funding announcements collapsed from 756 (2024) to 125 (2025) to 14 through mid-March 2026.
• The FY2025 grant success rate hit 17%, the lowest in 30 years; early-stage investigators fell from a 29.8% to an 18.5% success rate in two years. At the National Cancer Institute, odds are roughly 1-in-25, versus 1-in-10 before the cuts.
• FDA on-time drug approvals slipped to 78% in late 2025, from a historical 85-90%, after the agency's reviewer corps shrank.
• Federal science grant obligations fell 24% in FY2025, and 36 of 50 states received less federal science funding.

Meanwhile the management layer moved the other way: political appointees not subject to Senate confirmation grew roughly tenfold at NIH, and the Partnership counts the largest political workforce government-wide in four decades. Fewer career scientists, more political supervision of the ones who remain.

If You're Still Inside

The day-to-day translation of these numbers is workload and review pressure: the same statutory missions with 20-40% fewer hands, and decisions that used to be career-scientist calls now passing through political layers. Three practical notes:

• Your retirement eligibility is your option value. With agencies this thin, retention incentives and workload both rise. Know exactly what your annuity looks like now versus two years from now with the FERS Retirement Calculator, so any decision you make is a choice rather than a reaction.
• If a VERA lands at your agency, the eligibility math (age 50 + 20 years, or any age + 25) is two minutes in the VERA Eligibility Checker.
• Document your position description and duties now. Reorganizations move fast, and your RIF retention standing depends on records that are easier to assemble before the music stops.

If You're Already Out

Two clocks matter. RIF-separated career employees carry Reemployment Priority List rights for 2 years (1 year if career-conditional), but only if you register within 30 days of separation; that's your priority claim if your old agency rehires into your specialty, including under any future rebuild. DRP and voluntary departures generally carry no RPL rights, which thousands of scientists discovered after the fact. If your separation involved severance, check what you were owed against the Severance Pay Calculator; underpayments happen in fast-moving RIFs.

Frequently Asked Questions

How many federal scientists have left?

Nearly 118,000 science-agency employees between September 2024 and February 2026 (Partnership for Public Service), including 10,109 doctoral-trained scientists in calendar 2025 alone, which is 14% of the government's PhD science workforce (Science/AAAS analysis of OPM data).

Which agencies were hit hardest?

By share of workforce: SAMHSA (-41.7%), NPS (about -35%), NSF (-40% of its PhDs), CDC (-22 to -25%), NIH and FDA's drug centers (about -20% each), NOAA and USGS (about -20% each).

Why is it called a generational loss?

Because the losses concentrated at both ends: senior experts retired out through buyouts while the hiring freeze cut new scientist arrivals to almost nothing (11-to-1 departures-to-hires; DOE's first-year scientists fell from 116 to 12). Rebuilding an experience distribution takes a decade-plus.

What's the impact on research funding?

NIH competitive awards are down 66% in FY2026, funding announcements fell from 756 to 14, the grant success rate is at a 30-year low of 17%, and 36 states received less federal science money in FY2025.

What rights do separated scientists have?

RIF-separated career employees get 2 years of Reemployment Priority List standing (1 year career-conditional) if they register within 30 days. DRP and voluntary exits generally get none. RPL is the rehire-priority mechanism if agencies rebuild.

Related Resources

• HHS Hires 12K After Cutting 20K: The rebuild-math case study at one department
• The $165.6 Billion Cost of Federal Workforce Changes: The economic accounting behind these cuts
• CDC Workforce Cuts: The public-health slice in detail
• FERS Retirement Calculator: Know your numbers before deciding anything
• VERA Eligibility Checker: Two-minute early-retirement eligibility check

---

This article is general information. Sources: Partnership for Public Service report coverage, FNN (June 9, 2026), Science/AAAS: 10,000+ STEM PhDs lost (April 2026), AIP FYI workforce analysis, Eos: State of the Science workforce, AAU NIH grantmaking data, Nature emigration poll. Figures use each source's measurement window, noted in the table.