China has ordered publicly owned computing hubs to source more than half of their artificial-intelligence chips from domestic producers, expanding a quota first piloted in Shanghai across the country, according to people familiar with the directive. The instruction, endorsed by agencies including the Ministry of Industry and Information Technology and the National Development and Reform Commission, aims to curb reliance on US suppliers as Washington tightens export controls. Beijing’s push coincides with a construction boom: more than 500 new data-centre projects were announced nationwide in 2023 and 2024, spanning regions from Inner Mongolia to Guangdong. Ensuring that these facilities run largely on home-grown semiconductors is intended to fortify the indigenous supply chain and protect strategic computing capacity from external policy shocks. Local chipmakers are moving quickly to capitalise. Cambricon Technologies is seeking to raise about US$560 million to accelerate development of its AI processors, while other domestic component suppliers have seen valuations surge on expectations of steady orders from state-backed computing clusters. The mandate threatens to chip away at Nvidia’s dominance—analysts estimate the US group controls roughly 80% of the global AI-GPU market and still supplies about 15% of China’s AI-chip demand. The policy underscores the geopolitical stakes in advanced computing and could reallocate billions of dollars in data-centre spending toward Chinese semiconductor firms over the next several years.