Nvidia starts pitching “Vera” CPUs to Chinese AI customers ahead of a reported August availability window

Nvidia has begun a sales push in China for a new family of central processing units it is marketing for AI data centres, according to three sources familiar with the discussions. The people said Nvidia has told Chinese clients that its “Vera” processors could be available as soon as August and that customers may be able to begin placing orders around that timeframe.

The outreach suggests Nvidia is using a familiar approach in enterprise AI hardware, preparing customers for lead times and ramp schedules ahead of formal mass shipments. It also indicates the company is prioritizing the Chinese market for its next wave of computing platforms, even as the broader AI chip supply chain has required careful planning for customers operating under shifting export and procurement constraints.

The sources did not provide additional specifics about Vera’s pricing, performance targets, or how it compares with Nvidia’s existing CPU lineup or with competing server processors used to feed accelerators for AI training and inference. They also did not detail whether the August estimate would apply to all Vera configurations or only select customers and system partners.

In practical terms, CPUs are the “general-purpose” workhorses inside servers. In AI data centres, they coordinate data movement, run orchestration software, and handle workloads that do not map directly to specialized accelerators like GPUs. For large customers, access to a new CPU generation can matter because it affects system design, server configuration, power and cooling planning, and software validation timelines.

For Nvidia, which is best known for GPUs and networking used in AI systems, expanding CPU offerings is part of building more complete platform stacks. A CPU positioned for AI data centres can help simplify deployments for large customers, particularly if customers can standardize on Nvidia’s broader compute and interconnect ecosystem rather than mixing components from multiple vendors.

Nvidia has not publicly confirmed the timing or commercial availability details described in the discussions relayed to the sources. The company also has not, in the context of this report, outlined a launch date, customer order process, or shipment schedule for Vera. As a result, the August window should be treated as a customer-facing estimate rather than a company-issued product rollout date.

China remains a high-stakes market for advanced computing, where customers often want clarity on delivery timing and component availability. If Vera does become orderable in August, it could influence how data-centre operators schedule server procurement and how system integrators plan their builds for the second half of the year.

What to watch next is whether Nvidia issues an official product announcement or update that aligns with the reported August availability, and whether customers or partners describe system validation progress for Vera-based server platforms. Any further disclosure about commercial order timing, initial configurations, and supply readiness would be the most direct indicators of how quickly Vera could move from sales conversations into shipped deployments.