NVIDIA has reportedly reached out to several Chinese clients to adopt its upcoming Vera CPUs & there is a lot of interest building up.
Vera CPUs Will Be Available In August, NVIDIA Tells China As It Attempts To Revive Its Lost Revenue
NVIDIA's GPUs have faced a hard time in China. The company saw US regulations restrict them from selling Hopper chips, while the more advanced Blackwell generation is entirely banned. NVIDIA's CEO stated a few months back that their China share has dropped to Zero, and while the US Government did resume Hopper GPU sales, they have yet to see any reasonable revenue. More so, China itself has pushed against the use of NVIDIA GPUs at domestic data centers and AI factories, with even consumer-oriented options being pushed out of the market.
While the restrictions on GPUs continue to apply, NVIDIA is now trying to find an alternative to resume its revenue crunch in China. That's where Vera comes in. The NVIDIA Vera CPUs are in volume production, & the company has hand-delivered the first samples to major US-based AI firms. But it's not just US firms that are showing interest in Vera.
According to Reuters, NVIDIA has talked with potential Chinese customers about the adoption of its Vera CPUs. The company has told clients that Vera is all set for availability in August this year, and they can start placing orders right now.
NVIDIA has told Chinese clients that its new "Vera" central processors for AI data centres could be available as soon as August and that they can begin placing orders, three sources familiar with the matter said.
One major Chinese cloud company plans to place an order for more than 300 servers, each containing two Vera CPUs, one of the sources said.
Reuters
What makes Vera such a potent product for AI data centers, especially Agentic AI, is its standalone option in the form of Vera MGX racks. These racks feature up to 256 Vera CPUs, connected using fast NVLink I/O & offer up to 90% faster performance in AI applications versus standard x86 offerings. Each Vera CPU has 88 Olympus "Arm" cores, 164 MB of L3 cache, 166 MB of L2 cache, and supports up to 1.5 TB of LPDDR5X SOCAMM2 memory or up to 400 TB of unified memory across the rack. The platform supports PCIe Gen6, CXL 3.1, and comes in air/liquid cooled options with TDPs ranging from 250W up to 450W.
As per the report, multiple Chinese companies have shown interest, with one planning to place orders of 300 Vera servers, that's two Vera CPUs per server. The client has told NVIDIA that it will first place orders for a test unit before placing the full order.
Current price estimates suggest that a single Vera CPU will cost "north of $20,000," and a full rack will be priced around $10 Million. Earlier reports suggested that Alibaba and ByteDance were also showing lots of interest in getting NVIDIA Vera CPUs for their data centers.
Meanwhile, AMD & Intel are doubling down in the battle against NVIDIA's Vera, which aims to take away major CPU revenue and market share from them.
In its latest benchmarks, AMD shows how its EPYC lineup trounces Vera and next-gen Venice chips will offer massive performance advantages over the same chip. Intel is also preparing its Xeon onslaught against NVIDIA Vera, and the custom silicon threat is also rising with major AI firms developing their own CPUs for Agentic AI. NVIDIA has already said that they are set to become the largest CPU supplier in 2026, and with Vera entering volume ramp, it will be a fierce showdown in the Data Center-oriented CPU markets.
About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech's Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.
Follow Wccftech on Google to get more of our news coverage in your feeds.
