China Adds Homegrown AI Chips to 'Secure and Reliable' Procurement List for the First Time — Nine Options Added as Move Away From Nvidia Continues

Why It Matters

The list formalizes China’s shift away from Nvidia‑based solutions, giving home‑grown chips preferential access to billions of dollars in public contracts and accelerating the country’s quest for AI hardware self‑sufficiency.

Key Takeaways

• China certifies nine domestic AI chips for state procurement
• Huawei, Alibaba, Biren, Hygon, Iluvatar, MetaX, Moore Threads approved
• Nvidia's domestic share drops as Chinese firms claim 41% of AI servers
• SMIC runs 93% capacity, limiting production of certified AI chips
• Cambricon absent from list, possibly chose not to submit

Pulse Analysis

China’s security agencies have added nine home‑grown AI processors to the Anke V3.0 certification list, turning them into the default choice for government and state‑owned enterprises under the Xinchuang programme. The roster features Huawei’s Ascend 310 and 910, Alibaba’s T‑Head Zhenwu M530 and M890, and chips from Biren, Hygon, Iluvatar, MetaX and Moore Threads. By creating a dedicated “AI training and inference chips” category, Beijing signals a systematic move to replace foreign hardware—most notably Nvidia’s GPUs—with domestically vetted alternatives.

The certification comes as Chinese chipmakers already command roughly 41 % of the nation’s AI server shipments, delivering 1.65 million of the 4 million AI GPUs installed in 2025. Huawei alone shipped about 812 000 AI chips and projects $12 billion in processor revenue for 2026. Morgan Stanley projects the domestic AI‑chip market could reach $67 billion by 2030, with local supply covering about 76 % of demand. The exclusion of Cambricon, a former certified player, underscores that participation remains voluntary and competitive.

Despite the policy push, production capacity remains a bottleneck. All certified vendors rely on SMIC, whose most advanced stable node is an N+2 process comparable to 7 nm, and the fab reported utilization above 93 % for 2025 after spending $8.1 billion on capital expenditures. With limited wafer slots, scaling the certified chips to meet government‑driven procurement targets will be challenging. Nonetheless, the move deepens China’s strategic autonomy in AI hardware and could accelerate investment in next‑generation process technologies to close the gap with global rivals.