The NSF is on Track for its Worst Year in Generations

Throttling of science funding is utterly unprecedented and radical when compared to the past 50 years.
news
National Science Foundation
funding
Author

Noam Ross

Published

June 19, 2026

The National Science Foundation is on track for its worst year in over half a century. NSF grant records go back to 1960, giving us a long baseline to judge the scale of the current slowdown. What that history reveals is not just a departure from recent norms or a pre-COVID baseline, but a reversal of over five decades of growth in American science investment.

In 2025, the NSF made about 8,650 awards.1 The last time it made so few awards was in 1985. This year’s trajectory is far worse. It looks most similar to 1973, when the NSF made only about 1,700 awards in total. While it is possible that the NSF could accelerate its award pace in the final months of the year, the current trajectory is so far below all other years that it would require an unprecedented surge in awards to pull us back up to anywhere near recent levels, especially given the interruptions like the recent pause.

A line graph titled "Cumulative NSF Awards Made by Day of Fiscal Year" shows cumulative NSF award counts from October to September across six decades. A color-coded legend groups lines by decade from the 1960s through the 2020s. Most historical lines rise steeply and cluster between 5,000 and 15,000 awards by fiscal year end. A thick bright red line labeled "2026" runs far below every other year, reaching only about 3,500 awards by mid-June. Year-end labels on the right show FY2024 in teal at the top (around 10,500 awards), FY2025 in dark red (around 9,500), and FY1985 in green nearby. At the bottom, "1973" in gold marks the historical year whose full-year trajectory most closely matches FY2026's current pace.

If we look at dollars rather than awards, the picture is scarcely better. Last year, despite ostensibly spending its full congressional appropriation of $8.2 billion, newly awarded project budgets only amounted to about $6 billion, more than a 25% drop from the previous year and a number last seen in 2019.2 This year so far, new awards’ budgets amount to less than $1 billion - 12% of the overall budget and a number last seen in 1987. For 2027, the president’s budget request is $3.96 billion, a twenty-year setback even if they spend it all on new awards.

A line graph titled "Cumulative NSF Funds Awarded by Day of Fiscal Year" shows cumulative dollar amounts from October to September. A color-coded legend groups lines by decade from the 1960s through the 2020s, with recent years reaching $6–8 billion by fiscal year end. A thick bright red line labeled "2026" runs far below all other years, reaching only about $1.7 billion by mid-June. Year-end labels on the right show FY2024 in teal at the top (around $8 billion), FY2025 and FY2019 clustered near $6 billion, and FY2007 in purple at around $4 billion. A red dot marks the FY2027 President's Budget Request level at roughly $3.5 billion, with "2007" labeled as the closest historical year-end value. "1987" in green at the bottom (around $1.9 billion) marks the historical year closest to FY2026's current pace.

Even these stark numbers don’t paint the full picture, because comparing today’s numbers to the budgets so far back can be misleading. Below I plot the progression of awarded funds again, this time on a per capita, inflation adjusted basis to get a more apples-to-apples comparison across decades. Before the current administration, the NSF was spending $20-30 per American to fund scientific research each year. The 2020s, despite being high in absolute dollars, were far from outlier years. 2009, the year of the American Recovery and Reinvestment Act stimulus, was the peak of scientific investment in recent times. Last year’s throttling pulled us back to about $17, numbers last seen in the 1990s. While an acceleration of funding in the last months of the fiscal year could bring us back up, the current trajectory most resembles the level of investment in science last seen in 1969 during the height of the Vietnam War, and next year’s budget request is $11 per person, a level of investment last seen in 1973.

A line graph titled "Cumulative NSF Funds Awarded Per Capita (2024 dollars) by Day of Fiscal Year" shows inflation-adjusted per-capita NSF funding from October to September. A color-coded legend groups lines by decade from the 1960s through the 2020s. A thick bright red line labeled "2026" runs far below all other years, reaching only about $3–4 per person by mid-June. Year-end labels on the right show FY2024 in teal at the top (around $23 per capita), FY2025 in dark red (around $17), and "1987" in green (around $15). A red dot marks the FY2027 President's Budget Request at roughly $11 per capita, with "1973" in gold labeled as the closest historical year-end value. "1969" in brown at the very bottom (around $5) marks the historical year most closely matching FY2026's current pace.

It is clear from the administration’s actions that they are attempting to end the public role of scientific funding as it has been for over fifty years, despite Congress, the courts, and the public’s support for scientific research. Far from reversal of some recent overreach, or a correction from a drift off course, this is a radical contraction of the American scientific enterprise to a level from before most of today’s Americans were even born.

If we allow this strangulation to continue for long, American science will not easily be recusitated.

Footnotes

  1. These numbers are slightly different than those on our live funding curves page and NSF’s “By the Numbers” page. 8,650 reflects new award records, which can also include institutional transfers and some award renewals, though we attempt to filter these out. In our live curves, we use USAspending accounting data to help deduplicate and disambiguate these cases. The long-term historical data on awards is does not have similar detailed accounting data available.↩︎

  2. These numbers are different than those on our live funding curves page. Long-term historical data on obligations - the formal commitment of funds to projects - is not available at the daily level as it is for more recent years. Instead here we assign the full expected budget value of each award to the date the grant was awarded. This means that the within-award timing of funding is not as accurate. Yet this is in some ways a more honest assessment. It reflects actual overall investment planned to be made for each new award, and is not distorted by the timing of when funds are obligated, which the administration has manipulated through forward-funding grants already in the pipeline rather than awarding new ones.↩︎