Nvidia’s AI chip sales in China stall, as local chipmakers like Huawei take the lead - SRN News

The competition between the U.S. and China in developing artificial intelligence is intensifying, particularly in the hardware and computing power arena. Chinese companies like Huawei are overtaking g... Read original →

U.S. export controls on AI chips have backfired, accelerating China’s domestic substitution rather than stifling its compute ambitions. Huawei’s Ascend 950 now rivals the H200 in inference workloads, slashing AI cluster deployment costs by over 30% and eroding Nvidia’s mid-tier sales. Technologically, this spurs rapid co-development in Chinese EDA, advanced packaging, and compiler stacks—though sub-3nm nodes remain bottlenecked by EUV restrictions. Compliance-wise, multinationals face dual regulatory burdens: adhering to U.S. BIS rules while meeting China’s localization mandates for AI infrastructure, inflating operational overhead. Strategically, AMD may leverage foundries in Taiwan, China to launch a MI300X variant for the residual market, while Nvidia doubles down on CUDA lock-in via software moats. Within 18 months, Chinese AI chips will dominate Southeast Asia, cementing a bifurcated global hardware order: foreign chips for training, domestic ones for inference.