Nvidia ships Vera CPUs to SpaceX, OpenAI

Nvidia has moved its new Vera processor from launch-stage marketing into customer hands. In a company blog post published on May 18, Nvidia said its first Vera CPU systems had been delivered to Anthropic, OpenAI, Oracle Cloud Infrastructure and SpaceXAI, marking the first announced customer shipments of the chip. The deliveries came two months after Nvidia unveiled Vera at its March 16 GTC event as a processor built for “agentic AI” workloads rather than conventional server use cases. Elon Musk, whose SpaceXAI operation was among the named recipients, responded on X with the line “Vera nice.” ### Which companies actually received the first systems? Nvidia named four early recipients in its May 18 post: Anthropic in San Francisco, OpenAI in Mission Bay, SpaceXAI in Palo Alto and Oracle Cloud Infrastructure in Santa Clara. The company said the first three deliveries were made on Friday, followed by Oracle on Monday. Ian Buck, Nvidia’s vice president of hyperscale and high-performance computing, hand-delivered the systems, according to the company. Nvidia described the move as the point at which Vera went “from announcement to production.” ### What is Vera, and how is Nvidia describing it? Nvidia introduced Vera on March 16 as “the world’s first processor purpose-built for the age of agentic AI and reinforcement learning,” according to its press release. (blogs.nvidia.com) The company said Vera is designed to handle the CPU-heavy parts of AI systems, including orchestration, tool use, code execution, retrieval and sandboxed workloads that sit alongside GPU processing. The chip has 88 custom Nvidia-designed Olympus cores, 1.2 terabytes per second of memory bandwidth and 50% faster per-core performance under constant load, Nvidia said in its May 18 blog. In March, Nvidia also said Vera delivered twice the efficiency and 50% faster performance than traditional rack-scale CPUs. ### Why are OpenAI and Anthropic attached to the Vera Rubin platform? (investor.nvidia.com) OpenAI and Anthropic were already among the companies publicly tied to Nvidia’s broader Vera Rubin platform before this week’s delivery announcement. In Nvidia’s March 16 newsroom release, Anthropic Chief Executive Dario Amodei said the platform would help support “increasingly complex reasoning, agentic workflows and mission-critical decisions,” while OpenAI Chief Executive Sam Altman said it would let the company run “more powerful models and agents at massive scale.” (blogs.nvidia.com) Oracle Cloud Infrastructure was also listed in Nvidia’s March 16 press release as one of the customers collaborating to deploy Vera. That earlier release named Oracle, but not OpenAI, Anthropic or SpaceXAI, as a Vera deployment partner. The May 18 delivery post added those three as first announced recipients of physical systems. (nvidianews.nvidia.com) ### What does SpaceXAI mean in Nvidia’s announcement? Nvidia’s May 18 post referred to the Palo Alto recipient as “SpaceXAI.” The company did not define the term in the lines available from its blog, but it used that name in the delivery announcement and in the caption accompanying Buck’s handoff to partners. Musk’s public response came after that announcement. (investor.nvidia.com) A post on X cited by market and industry accounts quoted him as saying “Vera nice,” echoing the product name after SpaceXAI was listed among the recipients. ### What comes next for Vera shipments? Nvidia said on March 16 that Vera is part of the broader Vera Rubin platform, which it described as being in full production across seven chips and multiple rack-scale systems. (blogs.nvidia.com) The company also said a new Vera CPU rack integrates 256 liquid-cooled Vera CPUs and is built on Nvidia’s MGX modular reference architecture. Nvidia’s May 18 post said the first systems had already reached four named customers, and Oracle Cloud Infrastructure, OpenAI, Anthropic and SpaceXAI were the initial sites it identified. The next public milestones are likely to come through Nvidia customer deployment updates and the company’s continuing rollout of the Vera Rubin platform announced in March. (blogs.nvidia.com) (nvidianews.nvidia.com)