Nvidia to Ship AI Chip for China Market by Year-End Amid Regulatory Challenges

Nvidia is set to ship a new AI chip specifically designed for the Chinese market by the end of 2026, as reported by The Information. This chip, based on language processing unit technology licensed from Groq, aims to navigate U.S. export restrictions while meeting the high demand for AI hardware in China.

Chinese companies have already placed orders for the chip, but it still requires approval from Chinese regulators for import. The new chip is an LPU variant optimized for AI inference workloads, intended to complement Nvidia's existing GPUs rather than replace them. It targets applications such as chatbot interactions and other AI services.

Nvidia has previously shipped compliant chips to China, including the H20, which was designed to meet U.S. export control thresholds while remaining appealing to Chinese buyers. The new LPU follows a similar strategy, balancing compliance with performance to secure orders from Chinese firms.

The Biden administration has tightened export controls on advanced AI chips, limiting the computational power of processors sold to Chinese entities. Although some exports of more capable chips have been allowed under a case-by-case approval framework, the new LPU must clear both U.S. export controls and Chinese import regulations.

Nvidia's push into the inference market is crucial, as inference involves running trained models at scale for end users, presenting a larger long-term market opportunity compared to training. With Chinese companies like Baidu developing their own inference capabilities, Nvidia's shipment of an optimized chip is not only about revenue but also about maintaining its competitive edge in the market.

The planned shipment timeline gives Nvidia approximately four months to secure the necessary approvals to proceed.