Nvidia denies plans for China-specific AI chip

https://live-hls-7agy.livepush.io/live_abr_cdn/nsitWAl1pTZtwca/emnmZbKZcO8RfwJR/index.m3u8

Nvidia has denied a report suggesting it plans to ship a language processing unit specifically designed for Chinese customers by the end of the year.The company described these claims as incorrect, despite ongoing U.S. export controls impacting its operations in China.

China strategy update

Earlier reports indicated that Nvidia was preparing to release small batches of an AI chip for Chinese customers and had received orders.

This involves a language processing unit that complements its GPUs for enhancing AI chatbot responses.

The AI chip purportedly complied with U.S. export regulations without sacrificing processing power, according to The Information.

Nvidia quickly disputed this, insisting there are no current LPU sales or products on its roadmap for China.

Recent changes in the market highlight Nvidia’s fluctuating approach to its Chinese operations.

Washington approved sales of Nvidia’s H200 AI chips to selected Chinese firms like Alibaba, Tencent, and ByteDance, though actual deliveries have only just begun.

Despite some shipments, regulators have instructed these companies to keep the hardware outside of mainland China.

Nvidia is also promoting its Vera CPU to Chinese clientele.

The company has admitted to largely conceding the AI chip market in China to rival Huawei Technologies.

Market opportunities emerge

Surging demand for inference hardware in China presents potential opportunities for Nvidia.

Chinese AI firms are experiencing significant capacity shortages, prompting Moonshot AI to halt new registrations.

Local companies have secured domestic chip production until 2027, providing Nvidia a temporary gap to explore.

Regardless of whether Nvidia proceeds with its China-compliant inference chip, its current strategy highlights a shift from its traditional high-volume chip operations.