How US export curbs are forcing China to redesign its AI chip industry - South China Morning Post

Under sustained US export controls on advanced semiconductors, Chinese AI chipmakers are striving to build a self-reliant silicon ecosystem capable of breaking Nvidia’s dominance. The core of this str... Read original →

Sustained U.S. export controls are forcing China’s AI chip sector to fundamentally rearchitect its hardware strategy. The technical ripple effect is clear: reliance on NVIDIA’s CUDA-locked GPU ecosystem is giving way to ASIC-centric designs—Huawei’s Ascend and Cambricon’s MLU chips prioritize inference efficiency over programmability. Soaring compliance costs compel firms to rebuild domestic EDA, packaging, and testing capabilities, while advanced-node restrictions push adoption of Chiplet-based workarounds. NVIDIA may temporarily retain share via A800/H800 variants, but homegrown LLMs from Alibaba and DeepSeek will increasingly optimize for domestic silicon, creating a closed-loop stack. Over the next 12–24 months, China’s AI chip landscape will accelerate its 'GPU deprecation' trajectory—not by matching NVIDIA’s specs, but by leveraging policy-backed vertical integration to dominate end-to-end AI infrastructure despite a generational performance gap.