Nvidia has denied claims it is preparing a China-only LPU, leaving rumour merchants with egg on their keyboards.
According to Reuters, the chipmaker said a report by The Information claiming it would ship a language-processing unit to Chinese customers by year-end was incorrect.
“The reporting in The Information on Nvidia’s LPU is incorrect. We have no LPU sales in the China market today, and no China-specific LPU product in our roadmap,” an Nvidia spokesperson said.
Reuters reported in March that Nvidia was preparing a China-compatible version of its AI chips as it sought to compete in the fast-growing AI inference market.
The Information later claimed, citing two Nvidia employees, that small-batch shipments of a China-focused AI chip were planned before the end of 2026. It said several customers had already placed orders.
According to that report, the chip was a version of Nvidia’s LPU, built using technology licensed from startup Groq. The processor was said to work with GPUs to speed up AI chatbot responses.
The publication claimed the chip complied with US export-control rules. It said Nvidia had tweaked software so the LPU could run with processors available in China.
Nvidia’s China business has been all over the shop lately. Washington approved H200 AI chip sales in May to a limited group of Chinese firms, including Alibaba, Tencent and ByteDance, although deliveries only recently began.
Nvidia has been flogging its Vera CPU to Chinese customers too, after US restrictions made its next-generation Vera Rubin AI system unavailable in the market.
Nvidia chief executive Jensen Huang said in May that the company had “largely conceded” China’s AI chip market to Huawei Technologies, which has become its main domestic rival.
