Nvidia’s China AI-chip share falls as domestic suppliers grow

Nvidia shipped 678,000 AI accelerators in China from January through June, down 44% from a year earlier, according to IDC data.[1] Huawei shipped 572,000 accelerators, up from more than 300,000 a year earlier, and IDC put its share of China’s overall AI-accelerator market at 28%.[1] The figures point to stronger domestic competition, but the market-share measures are not stated on identical terms: Reuters described Nvidia’s 34% share as belonging to the data-centre AI-chip market, while Huawei’s 28% figure covers the overall AI-accelerator market.[1]

US export controls have restricted China’s access to Nvidia’s most powerful AI chips, limiting which products Nvidia can sell there.[1] Beijing has promoted domestic alternatives and required Chinese companies to seek approval to import compliant Nvidia chips.[1] Across the market, total AI-accelerator shipments reached 2.008 million units in the first half.[1] GPU shipments fell 26% to 991,000, while non-GPU accelerator shipments rose nearly 80% to 1.016 million.[1] Non-GPU accelerators include chips designed for specific AI tasks, such as running trained models.[1]

Huawei, Alibaba chip-design unit T-Head, Baidu’s Kunlunxin and Cambricon accounted for 95% of the non-GPU segment.[1] IDC said government agencies, internet companies and telecoms providers were adopting these accelerators.[1] Nine of the 11 Chinese chipmakers tracked by IDC increased shipments from a year earlier.[1] IDC also said competition was shifting from the performance of a single chip toward system-level capabilities, including how chips work in larger computing clusters.[1] It forecast total accelerator shipments in China would grow 19% that year, even as GPU shipments fell 9%.[1]

IDC did not specify which Nvidia products accounted for the company’s 34% market share.[1] It also did not say whether its reported share included B200 and B300 chips, which are barred from export to China but have been smuggled into the country.[1] That leaves an important limitation for buyers and market analysts comparing vendors or estimating the supply of particular products. Nvidia did not respond to a request for comment.[1]