LONDON, June 12 (Parliament Politics Magazine) – Nvidia has officially initiated a sales pitch to Chinese clients for its new Vera central processing unit, signaling a significant pivot to recover market share in the region. According to three sources familiar with the matter, the company informed major Chinese customers that they may begin placing orders for the new AI data center processors as early as August.
This outreach follows months of stalled shipments of the H200 AI chip to the Chinese market. CEO Jensen Huang previously noted that Nvidia’s market share in China effectively fell to zero, driven by U.S. export controls on advanced hardware and Beijing’s emphasis on domestic technological self-reliance.
Entering the server CPU market
The introduction of the Vera processor increases competition with established firms such as Intel and AMD, both of which are currently racing to increase their supplies of server CPUs for AI data centers. Nvidia’s new offering is the company’s first standalone central processing unit specifically designed for agentic AI, which refers to systems capable of performing tasks autonomously.
The chip is currently in full production and is optimized for the background computing processes that AI agents require. According to the company, the Vera chip runs up to 1.8 times faster than comparable processors currently available from its primary rivals.
“Most of the chips are initially going into the large, ready-to-install racks favoured by hyperscalers, with the simpler two-processor servers expected to ramp up later,” stated a report from the research firm SemiAnalysis.
Client interest and market strategy
Some Chinese firms have already shown interest in the new hardware. One major Chinese cloud company is reportedly planning a pilot order of more than 300 servers, each containing two Vera CPUs. The client intends to deploy the systems for testing purposes before deciding whether to place larger, official orders based on the performance results.
However, widespread adoption may face challenges. These include potential issues regarding software ecosystems and compatibility, as well as the technical constraints involved in migrating existing workloads from domestic AI chips to the new architecture.
Nvidia projects approximately $20 billion in revenue from Vera chip sales by the end of its current fiscal year in January. The global shift in the AI industry from model training to inference computing the process of answering queries has created a notable CPU shortage. Intel recently notified Chinese customers of delivery lead times extending up to six months, while AMD has flagged that demand for the global CPU market is outpacing forecasts.
Navigating export regulations
Selling standalone CPUs in China could prove to be less complex than distributing graphics processing units, which are subject to stricter U.S. export controls. While Washington has previously licensed about 10 Chinese firms to purchase the H200 GPU, none of those units have been delivered, as Chinese officials have withheld approval to support domestic suppliers.
Because the Vera CPU utilizes Arm technology, its release places Nvidia in direct competition with Intel and AMD’s long-standing x86 architecture. The chip features a specialized design built on a 3-nanometer process node, incorporating 88 physical cores capable of managing 176 threads through spatial multithreading.
To manage technical and regulatory hurdles, sources indicated that Chinese clients plan to initially deploy the Vera chips only in their overseas data centers for testing. This strategy allows the firms to evaluate the hardware while navigating the current international trade environment.
