China Aims for AI Independence

A top Beijing advisor has suggested China should shift its focus to developing its own AI chips, rather than relying on foreign technology like Nvidia's accelerators. This push for self-sufficiency is motivated by concerns that Asian companies risk over-dependence on US tech. China has been increasing domestic AI accelerator production. Companies like Huawei and Cambricon are ramping up production, with Huawei potentially producing 1.2 million Ascend 910B chips in Q4 2025. SMIC is also increasing its production of 7nm-class chips.

Despite progress, challenges remain. SMIC's yields for advanced chips are still lower than those of TSMC. Huawei leads China's AI chip efforts, fostering a domestic ecosystem spanning chips and computing frameworks. Alibaba is also developing its own AI chip for AI inference tasks, manufactured by a Chinese company. This is to counter Nvidia's market restrictions in China. China is building an AI ecosystem around open-source models and locally optimised hardware.

This move signifies a broader geopolitical competition, with China aiming to establish its own standards and alliances in the AI sector. Sanctions have spurred domestic investment, with companies like SMIC increasing 7nm production. DeepSeek has begun optimising its AI models for domestic chips, further distancing itself from the Nvidia ecosystem.