NVIDIA Corporation (NASDAQ: NVDA) plans to begin limited shipments of a newly tailored artificial intelligence chip to Chinese customers by the end of 2024, according to a report citing The Information.
The chip is a modified version of Nvidia’s Language Processing Unit (LPU), developed using licensed technology from Groq, and designed to accelerate AI model response times when paired with compliant processors in China. The move follows restrictions that prevent Nvidia’s next-generation Vera Rubin architecture from being exported to China under U.S. regulations.
Engineers reconfigured the LPU’s software to ensure full compliance with export controls while maintaining performance. Several Chinese firms have already placed orders for the device, which will be available in small-batch quantities by year-end. Nvidia is also scaling shipments of its cleared H200 processors, with packaging partner Amkor Technology completing more than one million units in the second quarter.
Separately, Nvidia is developing two lower-power gaming chips specifically for the Chinese market. The company’s efforts come as domestic demand for AI inference hardware grows, with competitors such as Huawei Technologies advancing its Ascend 950DT chip. Earlier this year, Nvidia CEO Jensen Huang acknowledged that the company had largely conceded the region to Huawei amid regulatory constraints.
Chinese tech giants are also accelerating domestic chip production through 2027, while local AI firms including Moonshot AI and DeepSeek have faced computing bottlenecks that have led to service restrictions and fee increases.
Nvidia’s shares were down 0.5% in afternoon trading following the report.
