en.Wedoany.com Reported - Nvidia has informed Chinese customers that its Arm-based Vera server CPU could arrive as early as August, with orders now being accepted. Previously, during Computex, the company stated that Vera systems would begin shipping through system integrators and cloud partners this fall.
Meanwhile, shipments of the H200 AI GPU to China remain frozen. Nvidia CEO Jensen Huang said months ago that the company's market share in China had effectively dropped to zero. Amid a global server CPU shortage, Nvidia's notification to Chinese buyers that they can obtain the chip in August indicates a favorable position in the allocation queue for this product line. Nvidia expects the Vera product line to generate $20 billion in revenue by the end of its fiscal year in January.
According to Reuters, citing three sources familiar with the matter, Chinese cloud companies have been testing over 300 Vera servers, and at least one major cloud provider plans to place orders. One source said initial deployments will be limited to these companies' overseas data centers. If successful, Vera will reach Chinese buyers that Nvidia's GPUs cannot. Compared to GPU accelerators, server CPUs face much looser U.S. export restrictions. The U.S. approved about 10 Chinese companies to purchase the H200, but none have been delivered because Chinese officials did not approve the purchases to support domestic chipmakers.
Vera CPU deployments are limited to overseas data centers because, while Chinese cloud providers want to acquire the hardware, placing U.S. chips in domestic data centers could trigger scrutiny from Beijing. The Vera CPU was originally the CPU component of the Vera Rubin superchip, first unveiled at the GTC event last year. Nvidia launched it as a standalone product at GTC San Jose in March, alongside a rack design capable of housing 256 liquid-cooled Vera CPUs, supporting over 22,500 concurrent CPU environments. At Computex, Nvidia stated the chip is now in full production, completing tasks 1.8 times faster than x86 processors in agent workloads. Its predecessor, Grace, has shipped nearly 2.5 million units to date.
Server CPUs are under supply pressure as AI workloads shift from training to inference and agent execution. Agent AI heavily relies on the main processor for tool calling, code execution, and data processing, and due to the explosion of agent applications, CPU demand has exceeded supply. Intel has given Chinese customers delivery lead times of up to six months, while AMD says the global CPU market is tight, with demand exceeding expectations and supply constraints expected to persist.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
