Trump Shifts $200B Research Budget Toward AI and Solo Scientists

The White House just proposed the biggest rewrite of how America funds science since the postwar era, and the interesting part is not the dollar figure. It is who the checks are made out to. According to reporting in the New York Times, the Trump administration wants to steer the roughly $200 billion a year the federal government spends on research away from universities and consensus peer-review panels and toward individual scientists, AI-driven programs, and ARPA-style teams built for high-risk work.

The mechanics are the tell. A 120-plus-page report from the Office of Science and Technology Policy, titled 'Science: A New Golden Age,' proposes portable awards that follow the researcher rather than the school, 'golden ticket' grants for unconventional ideas, and faster funding paths that supplement traditional peer review. A companion memo, co-signed by OSTP director Michael Kratsios and OMB director Russell Vought, gives every federal agency with at least $3 billion in 2026 research budget authority 90 days to return an action plan for adopting the new practices.

Sitting alongside it is the Genesis Mission, which the White House expanded on July 22 with more than $5 billion in commitments across at least 15 federal agencies, including NIH, NASA, Energy, Transportation and the Department of War. The aim, according to the Kratsios-Vought memo, is to fund research that treats AI as 'a new instrument of scientific discovery,' with access to Department of Energy supercomputers, specialized datasets and 278 selected projects covering chronic disease research, pediatric cancer, drug discovery, semiconductor manufacturing and biological threat detection. The overall framing is competition with China.

The honest caveats are large. A White House report is a directive, not an appropriation, and Congress writes the check. Universities that rely on federal indirect-cost recovery have both the incentive and the lobbying muscle to fight grant portability. 'Golden ticket' bypasses of peer review can just as easily concentrate money in politically favored labs as unlock genuinely unconventional work. What the reporting does not give you is a clear map of which existing programs get zeroed out to free the money, or how portable-grant math will actually apply to the university overhead that currently pays for labs, staff and facilities.

If any of it lands as designed, the winners are individual star researchers, focused research organizations and AI-native scientific tooling shops already circling this space. The exposed side is the big research universities that treated federal indirect-cost recovery as a load-bearing wall of their business model. Whichever side of that shift you sit on, this one rewards reading the actual report rather than the summary.