A doctoral student who loses her F31 fellowship mid-project does not get to pause her career and resume it when the policy environment improves. She loses her stipend, her health insurance, and her lab placement. She probably leaves science. That is not a hypothetical. The NIH terminated 405 F31 awards in the first half of 2025, and women held 58% of them.
The PNAS study published March 24 confirms what many researchers suspected: the $2.5 billion in NIH grant terminations between February and August 2025 fell hardest on women and early-career scientists. Women led 57.9% of terminated grants with remaining funds, compared to 48.2% for men. Among assistant professors, women led roughly 60% of canceled projects. Among doctoral candidates, 60.2%. The pattern is consistent across every early-career category the authors examined.
I want to be precise about what this means. No post-hoc policy adjustment undoes this damage.
Career Stage Is the Variable That Matters
I spend most of my time writing about energy systems, where I track deployed capacity because announcements are cheap. The same logic applies here. NIH announced a Unified Funding Strategy in December 2025 that factors career stage into funding decisions and lowers score cutoffs for early-career applicants. On paper, that is a reasonable structural reform. But it is an announcement, not deployed capacity. It has funded zero grants so far.
Meanwhile, the damage is already measured. The study's lead author, Diego Oliveira at the University of North Dakota, identified the mechanism clearly: women and early-career researchers held smaller grants at earlier stages of execution, meaning a larger share of their planned research was interrupted. At Harvard, the median canceled grant for a woman investigator was $362,000, compared to $689,000 for men. Smaller grants, more unspent funds, more vulnerability.
Donna Ginther, an economist at the University of Kansas who studies NIH funding demographics, called the training grant terminations a risk of "derailing scientific careers just as they are getting started." She is describing a one-way door. A senior investigator who loses 1 of 3 active R01 grants can absorb the hit. An assistant professor whose only grant disappears cannot.
The Pipeline Arithmetic
The fair counterargument is that these cuts exposed a pre-existing fragility rather than creating a new one. Women were already concentrated in smaller, more precarious award categories. The cuts did not invent that distribution. True. But a system that concentrates its most vulnerable participants in its most fragile funding categories and then eliminates those categories is not neutral. It is a system that selected for maximum damage to the people least able to recover.
The numbers on training grants are stark. Women held 66% of canceled T34 undergraduate awards. The NIH eliminated $56.8 million in early-career programs and $329.6 million in T32 funds. These are not large numbers by federal standards. The estimated $6.3 billion in lost economic output dwarfs them. But the career-stage math is unforgiving: a 28-year-old doctoral candidate who loses her fellowship in year 3 of a 5-year program faces a fundamentally different calculus than a 55-year-old professor who loses supplemental funding.
NIH's spokesperson insisted the agency "allocated its full budget" and that "any other suggestion is false." That response treats the distribution of damage as irrelevant to the question of fairness. It is not.
The Unified Funding Strategy may eventually produce a better pipeline. I will evaluate it the way I evaluate any policy: by its measurable outcomes in the first funding cycle, not by its press release. But the 405 doctoral fellows whose awards were terminated, the assistant professors whose only grants vanished, the training programs that no longer exist: those are sunk costs in the most literal sense. The people affected are not waiting for the next funding cycle. They are updating their resumes.
Structural reform that arrives after the structural damage is triage. Useful triage, perhaps. But no one should confuse it with prevention.
