China drafts $295 billion plan to build national AI data center grid running on 80% homemade silicon

China is drafting a plan to invest approximately 2 trillion yuan ($295 billion) over five years to build a nationwide AI data center network, with at least 80% of underlying technology, including AI c... Read original →

China’s proposed $295B AI data center grid isn’t just infrastructure—it’s a full-stack stress test for forced localization. Technically, SMIC’s N+2 (7nm-class) lacks EUV scalability, and domestic HBM shortages cap Huawei Ascend output regardless of design prowess. The 80% domestic mandate effectively bans NVIDIA/AMD from state-linked projects, binding China Mobile and China Telecom to immature silicon ecosystems. Near-term, operators face soaring CAPEX and subpar performance-per-watt; long-term, if CXMT fails to deliver HBM within 18 months, the entire grid risks ‘chips without memory’—a structural bottleneck. Strategically, U.S. rivals will likely accelerate Southeast Asia-based AI clouds to isolate Chinese tech. Over the next 24 months, the real battle shifts from chip architecture to materials, equipment, and advanced packaging—the silent, fragile links in China’s semiconductor chain.