Experts say US is in danger of losing its status as the center of AI R&D, as GOP tax plan raises costs on graduates and immigration policy deters some
Will Knight / MIT Technology Review :
Context & Ripple Effects
In late 2017, Will Knight reported experts' warning that two Washington-driven cost shocks — the GOP tax plan hitting graduate students and immigration policy that deters foreign researchers — could end the US run as the center of AI R&D. The concern was not abstract: the field ran on foreign-born graduate labor, and policy was pricing both the degree and the visa out of reach.
The arc since then reads as confirmation rather than alarmism. Insiders soon worried that proposed Commerce Department export restrictions on AI would stunt the domestic industry, analysts flagged that national-security and immigration policy risked cutting US companies off from Chinese-origin talent — China had 47% of top AI researchers in 2022 — and by 2025 Zeki measured the balance at zero: AI talent entering the US now equals talent leaving. Microsoft's Eric Horvitz has since tied the same logic to cuts in academic research funding, arguing they push talent abroad.
First-order effects
- Foreign graduate students and early-career AI researchers face a higher net cost of staying in the US — tuition under the GOP tax plan plus visa uncertainty — making Canada, Europe, or returning home relatively more attractive.
- US universities and corporate labs lose access to the pipeline they recruit from, since immigration policy already deters some of the foreign graduates and researchers their AI programs depend on.
Second-order effects
- US companies competing for scarce AI talent face a thinner domestic pool and must bid up salaries or offshore research roles, while rivals abroad gain a recruiting pitch backed by friendlier policy.
- Restrictive policy compounds itself: the same national-security instinct behind the export-restriction debate also narrows the talent funnel from China, so each measure tightens the constraint the industry warned about.
Third-order effects
- If the pattern holds, AI R&D leadership becomes geographically distributed rather than US-centered — a structural shift from one dominant hub to competing national ecosystems, with government policy rather than market pull deciding where researchers cluster.
- The recurring warnings from 2017 through Horvitz suggest a durable tension: every security-motivated restriction on people, money, or technology trades a slice of US AI capacity for perceived control, forcing future administrations to choose explicitly between openness and control.
The trend: US AI research leadership is eroding not through any single defeat but through an accumulating stack of self-imposed costs — tax, visas, export controls, and funding cuts — that redistributes talent toward rival hubs.