China struggles to recruit foreign AI researchers even as it wins the domestic talent race

China has built AI labs that the rest of the world now watches closely. DeepSeek and Moonshot have become internationally known names.

The numbers behind China’s talent shift

A study published by Carnegie China in September 2026 found that China hosted 41% of the world’s top AI researchers in 2025. The study measured elite status by participation in NeurIPS, one of the field’s most prestigious research conferences.

That share was 27% in 2022. The US moved in the opposite direction, sliding from 46% to 34% over the same stretch.

The research points to two forces driving the reversal. Leading Chinese firms are offering competitive pay, while restrictive US visa policies have pushed researchers back to China after time spent overseas.

Retention at the undergraduate level tells a similar story. The share of Chinese undergraduate degree holders who stayed in the country rose to 69% in 2025, up from 57% in 2022.

DeepSeek offers a sharp case study. An analysis by Stanford’s Hoover Institution and HAI found that 53.5% of authors affiliated with the lab had never worked or studied outside China.

Among DeepSeek researchers who did gain US experience, more than 70% eventually came home.

The same research found that one-third of the core contributors to an AI model rivaling OpenAI’s capabilities came out of China’s domestic education system.

Why foreign recruits remain elusive

China’s Ministry of Human Resources and Social Security estimates the country’s AI talent gap exceeds 5 million professionals.

Language is the most immediate barrier. Working in a Chinese lab often means operating in Chinese, which shrinks the pool of viable international candidates considerably.

Geopolitical tension adds another layer. Researchers weighing a move must consider how the rivalry between Washington and Beijing could affect their careers, mobility and future employment options.

Rising anti-foreign sentiment, fueled in part by local youth unemployment, makes it politically awkward to roll out the welcome mat for outside hires.

DeepSeek’s hiring push and what comes next

DeepSeek plans to double department sizes across various roles by mid-2026 and is offering competitive salaries to do it. It has also signaled openness to foreign researchers, with one notable condition: candidates who speak Chinese.

What this means for the global AI race

For the US, the Carnegie China data is an uncomfortable signal. Losing its lead in elite researchers suggests that visa restrictions carry a real cost in a competition both governments treat as strategic.

The Hoover and HAI findings sharpen that point. When more than 70% of DeepSeek researchers with US experience go home, American institutions are effectively training talent for a rival ecosystem.

For China, the picture is mixed. A self-reliant pipeline insulates its AI sector from foreign pressure, but a talent gap above 5 million suggests demand is far outpacing supply.

Policymakers in Washington, meanwhile, may face renewed debate over visa rules. The data suggests that keeping researchers out does not stop them from doing frontier work, it simply changes where that work happens.