Beijing
The political-AI center of the second-largest research economy on earth.
Kernel
Beijing is the city where Chinese AI policy, state-aligned capital, and the principal national AI labs converge. Zhongguancun's Tsinghua and Peking University corridor produces the bulk of China's top-tier ML researchers; Beihang and BAAI provide a state-research interface; ByteDance, Baidu, Meituan, Xiaomi, and most of the country's foundation-model labs (Zhipu, MiniMax, Moonshot, 01.AI, Stepfun) operate primarily here. By 2025 Beijing is, by paper count and patent volume, the largest single AI research center outside the U.S.
Why it rose
Forty years of capital concentration in the political center plus the country's two best universities plus a deliberate state strategy (2017 AI Plan: dominance by 2030). The 2024 wave of indigenous foundation models — Qwen out of Hangzhou but largely Beijing-staffed at the talent level, DeepSeek out of Hangzhou, ChatGLM/Zhipu out of Tsinghua — shows the talent depth. Whether the closed-frontier-lab model survives Chinese open-weights efficiency is the central question of 2026–2028 global AI.
What it is giving the world (2024–)
Open-weights frontier models at a fraction of U.S. training cost (DeepSeek R1 in early 2025). A working alternative to U.S.-stack AI infrastructure for the global south. A multi-trillion-dollar sovereign-tech demonstration: a non-Western city building frontier-AI capability against the most concentrated export-control regime since the Cold War.