Chinese Firms Leave Nvidia for Local AI Suppliers, Survey Shows - Yahoo Finance

According to a Bloomberg Intelligence survey, Chinese companies are increasingly shifting away from Nvidia's advanced AI accelerators toward domestic semiconductor solutions, reflecting the impact of ... Read original →

U.S. restrictions on Nvidia’s H20 chips have not stifled China’s AI infrastructure buildout but instead triggered a tipping point in domestic substitution. Technically, Huawei’s Ascend and Cambricon’s MLU architectures are forcing a software stack migration away from CUDA, increasing deployment latency but cementing long-term autonomy. Compliance-wise, hyperscalers like Tencent and Alibaba are diversifying supply chains to mitigate secondary sanctions, elevating ChangXin and SMIC as critical fallbacks—though sub-3nm scaling remains bottlenecked by EUV access. Nvidia may respond with Southeast Asia rerouting or downgraded SKUs, yet its China revenue will irreversibly shift toward edge AI. Within 18 months, local AI accelerator adoption will exceed 60%, igniting consolidation among Chinese chip startups and redirecting global memory capacity toward AI-optimized HBM3e. Geopolitics has turned China from the world’s largest import market into its most insulated, yet scale-rich, semiconductor proving ground.