China expands tech replacement drive to include AI chips amid US curbs

What changed

For the first time, China has put artificial intelligence processors into its official “secure and reliable” technology assessment framework. The China Information Technology Security Evaluation Centre (中国信息技术安全评估中心) and the National Secrecy Science and Technology Evaluation Centre (国家保密科学技术评估中心) released a new batch of certifications that create a category for “AI training and inference chips,” with approvals valid for three years. It has been reported that these approvals are widely viewed as the definitive procurement catalogue for party and government agencies, central state-owned enterprises and other state-linked customers under the Xinchuang (信创) initiative — Beijing’s long-running campaign to replace foreign hardware and software in sensitive systems.

Why it matters

The move signals a clear policy pivot from earlier efforts that focused on CPUs and databases toward the infrastructure powering generative AI. Who stands in the crosshairs? Initially Intel, AMD and some Western database suppliers; now Nvidia and other foreign providers of AI compute are squarely targeted. The push has gained urgency after successive rounds of US export controls that have curtailed Chinese access to advanced GPUs, and Beijing is accelerating certification and procurement pathways for home‑grown alternatives — but can those domestic chips scale fast enough?

Geopolitical context and implications

This is more than an industrial policy tweak. By folding AI processors into the Xinchuang procurement ecosystem, Beijing is institutionalising an avenue for state buyers to favour domestic vendors, shaping investment, supply chains and product roadmaps. Reportedly, the certification list will influence what central agencies and state-linked firms buy next, intensifying the tech bifurcation between China and the West as export controls and countermeasures reshape the global semiconductor landscape.