Key Moments
- Nvidia has informed Chinese customers that Vera CPUs for AI data centers could be available as early as August, and is already inviting orders, according to three sources.
- A major Chinese cloud provider is preparing an initial order for more than 300 dual-Vera-CPU servers, while broader adoption remains uncertain due to software and migration challenges.
- Nvidia targets $20 billion in Vera revenue by the end of its fiscal year to end-January, positioning the Arm-based CPU in direct competition with Intel and AMD.
Vera Launch Targets China After GPU Shipments Stall
Nvidia has begun pitching its new “Vera” central processing unit to Chinese customers, telling them the AI data center chip could be ready for delivery as soon as August and that they can now start placing orders, three people familiar with the discussions said.
The push highlights how the company is rapidly shifting focus to this new product line in an effort to revive its sharply weakening position in China, where shipments of its H200 graphics processing unit, its second-most powerful AI chip, have been on hold for months.
Nvidia’s chief executive Jensen Huang said in October that the company’s market share in China had effectively dropped to zero, citing the impact of U.S. restrictions on advanced chip exports and China’s drive to build up domestic capabilities in strategic technologies.
The Vera rollout also intensifies Nvidia’s rivalry with established CPU suppliers Intel and AMD, both of which are working to expand server CPU availability for AI data center workloads.
Vera Positioned for Agentic AI and Data Center Compute
According to the sources, some Chinese customers have expressed interest in Vera, Nvidia’s first standalone CPU designed specifically for agentic AI – systems that can execute tasks autonomously. The sources requested anonymity because the talks are confidential.
Now in full production, Vera is designed for the core computing tasks that support AI agents behind the scenes. Nvidia says the chip delivers up to 1.8 times the performance of comparable processors from competing vendors.
When Vera was introduced in March, Huang projected that it would grow into the company’s next multibillion-dollar product line. At that time, Nvidia also said large cloud players such as Alibaba and ByteDance were collaborating with it to deploy Vera, but it did not indicate whether ordering had commenced.
Nvidia declined to comment. Alibaba and ByteDance did not respond to requests for comment.
Early Chinese Cloud Demand and Deployment Plans
One major Chinese cloud provider is preparing to order more than 300 servers configured with two Vera CPUs each, one of the sources said. The company intends to first use these systems for testing before deciding whether to commit to larger, formal orders, the person added.
Another source said it is still unclear whether this early interest will scale up significantly. Key factors include software ecosystems and compatibility, along with the challenges of moving existing workloads that have been developed for domestic AI chips.
Pricing, Configurations, and Supply Dynamics
Industry research firm SemiAnalysis estimated that a single Vera processor will be priced “well north” of $20,000 before any volume discounts. A fully built rack incorporating 256 Vera chips would cost roughly $10 million, with the final figure depending on the memory configuration, according to the firm.
SemiAnalysis said initial shipments are primarily going into large, turnkey racks favored by hyperscale customers, while simpler servers carrying two processors are expected to scale up later.
Nvidia is projecting $20 billion in revenue from Vera sales by the close of its fiscal year ending in January.
| Item | Detail |
|---|---|
| Earliest availability for Chinese clients | As soon as August (per sources) |
| Initial cloud test deployment | >300 servers, 2 Vera CPUs per server (source) |
| Estimated single Vera CPU price | “Well north” of $20,000 (before bulk discounts) |
| Estimated cost of 256-chip rack | Around $10 million (depending on memory configuration) |
| Target Vera revenue | $20 billion by end of fiscal year to end-January |
AI Market Shift Fuels CPU Demand
Chinese interest in Vera is emerging as the global AI industry shifts its focus from training models to inference workloads – responding to user queries – an area where CPUs and custom accelerators are mounting stronger competition against GPUs.
This transition has contributed to a global shortage of CPUs. Reuters reported in February that Intel had informed Chinese customers that lead times for server CPU deliveries could extend up to six months. AMD stated last month that the global CPU market is “tight,” with demand running ahead of expectations and supply limitations likely to continue.
Arm-Based Design and Regulatory Implications
Vera is built on Arm technology, placing Nvidia in direct competition with Intel and AMD, whose processors are based on the long-dominant x86 architecture.
Selling CPUs into China may face fewer regulatory barriers than selling GPUs, which are subject to stricter U.S. export controls. Washington has authorized about 10 Chinese firms to purchase Nvidia’s H200 GPU, but none of those orders have been fulfilled so far. Chinese authorities, seeking to support domestic chip makers, have not granted approval for those shipments.
One of the sources said Chinese customers plan to initially deploy Vera chips only in their overseas data centers for testing.
TradingPedia.com is a financial media specialized in providing daily news and education covering Forex, equities and commodities. Our academies for traders cover 
