NVIDIA has identified Anthropic and OpenAI as some of the initial users of its Vera chip.
During his Computex keynote, Jensen Huang took some time to read out a list of notable guests. Among the first significant users of Vera, Nvidia's new in-house processor, are Anthropic, OpenAI, SpaceX, and Oracle, as stated by the Nvidia CEO to the audience in Taipei on Monday.
Nvidia, which established its success on graphics chips, aims to also be recognized for its CPU. Vera is positioned as a complete redesign and serves as the successor to Grace, Nvidia's previous data-center processor. It utilizes 88 of Nvidia's proprietary "Olympus" cores, marking a shift from the off-the-shelf Arm Neoverse cores used in Grace, and the company has announced that it is now in full production.
The emphasis is that Vera is a CPU tailored for the era of AI agents—software that plans and executes tasks instead of merely responding to prompts. Nvidia asserts that this chip handles those agent-based workloads more swiftly than the x86 processors from Intel and AMD and is matched with up to 1.2TB/s of memory bandwidth.
Initial independent benchmarks indicate that Vera outperforms Intel's Xeon and AMD's EPYC in several categories, which is a comparison Nvidia is keen on highlighting. The significance of the named customers is as crucial as the technology itself. Anthropic and OpenAI are the two labs whose computing needs have significantly influenced the current development, and Nvidia mentioning them as launch users signals the direction of the chip's future.
According to Nvidia's reports, the first units were delivered by hand in May, prior to the public announcement in Taipei. For Nvidia, entering the CPU market is a strategic maneuver. Its GPUs are already central to nearly all major AI systems, and the data-center processor has traditionally been the one major component they have sourced instead of developed in-house. Creating their own CPU eliminates reliance on others and allows the company to market both the CPU and GPU as a cohesive set, which aligns with Huang's description of the Vera Rubin platform as the largest product launch in the island's history.
Huang has referred to Taiwan as the "epicenter" of this endeavor throughout the year. Oracle Cloud Infrastructure is poised to be the first hyperscaler to implement Vera on a large scale, with wider availability anticipated across other prominent cloud services in the latter half of 2026. This timeline is the key figure to monitor.
While Nvidia made announcements regarding full production and a lineup of prominent names in Taipei, they did not disclose pricing or specify how many units have been committed by the listed labs. The guest list was shared, but the financial details were not.
Other articles
NVIDIA has identified Anthropic and OpenAI as some of the initial users of its Vera chip.
Nvidia announces that among the first recipients of Vera, its proprietary CPU designed for AI agents now in full production, are Anthropic, OpenAI, SpaceX, and Oracle.
