Beijing Greenlights First 400,000 Nvidia H200 AI Chips During Huang’s Lunar New Year Tour

Great-power chip controls echo Cold War-era CoCom lists (1949-1994), when embargoed mainframes became bargaining chips in détente; today’s AI GPUs play the same role in Sino-American tech diplomacy. The decision reveals two converging structural forces: Beijing’s push for semiconductor self-reliance after the 2018-2020 ZTE/Huawei sanctions, and its immediate need for cutting-edge compute to train large models before domestic alternatives mature. By granting a quota while insisting on future domestic uptake, China is rehearsing the ‘import-substitution window’ strategy it used with high-speed rail in the 2000s—buy foreign tech, learn, replace. For Washington, the 25 % levy turns export control into revenue, blurring security and industrial policy much like the 1930s Neutrality Act “cash-and-carry” carve-outs. Over a century, the episode may matter less for the chips themselves—today’s top GPU is tomorrow’s scrap metal—than for the precedent: strategic technologies becoming tariffable, quota-managed commodities in a bifurcating world system. Whether this becomes the Sputnik-moment that accelerates a fully sovereign Chinese AI stack, or the start of a managed coexistence akin to U.S.-Soviet grain sales of the 1970s, will shape the next phase of global techno-economic order.