Nvidia wants to own every chip inside its AI data centers

Nvidia is exaggerating Vera Rubin announced the new Vera Rubin chipset system this week, revealing new performance standards for GPU and CPU combo Before the competitor AMDThe annual product event is in San Francisco on Thursday.

During a lengthy technical workshop last week at the company’s headquarters in Santa Clara, California, Nvidia executives bragged to a small group of journalists about increasing the chips’ system capabilities in terms of power and efficiency. Biggest takeaway: Nvidia, which has long specialized in making graphics processing units, is increasingly trying to position itself as a supplier of CPUs that can power AI agents.

While GPUs are still… Main devices Used by companies to train and run their AI models, the industry’s shift toward more complex agent systems has increased demand for central processing units, which can coordinate data flows, networking, and other software tasks. This is likely one of the reasons why Nvidia has been keen to promote itself as a supplier of complete AI systems, rather than just AI chips.

Vera Rubin is Nvidia’s successor Hybrid chip Grace Blackwell’s system represents the backbone of its near-term future supporting the AI ​​industry. It’s designed to provide one CPU for every two GPUs. In a single Vera Rubin NVL 72 super-chip system, there are 36 Vera CPUs for every 72 Rubin GPUs. Nvidia also sells the Vera CPU as a standalone product, and has done so He reportedly told Chinese agents These could be ready as soon as August.

The CPU chip that Nvidia is using for its new Vera Rubin hardware system.

Courtesy of Nvidia

Nvidia executives emphasized that the new Vera Rubin NVL72 racks — a collection of chips packed into a single liquid-cooled platform — are much more “plug-and-play” than some of its previous products. During a brief tour of Nvidia’s data center lab in Silicon Valley, Nvidia executives shared that OpenAI already has one Vera Rubin rack in use.

Nvidia CEO Jensen Huang did not show up at the workshop in Santa Clara last week; He was in Japan to announce a new chip maker Partnerships with a number of Japanese companies To develop artificial intelligence for robots. Instead, the briefings were moderated by Ian Buck, Nvidia’s longtime vice president of accelerated computing and the engineer behind the company’s CUDA software.

“We are on a roadmap to develop new architectures, not just GPUs but CPUs,” Pak told reporters. “We will continue to innovate, because we either do this or die in Silicon Valley.”

The meetings were held at Hwang’s Executive Briefing Center, where multiple nearby offices were filled with bags of Taiwanese snacks that the CEO had brought back from his recent trip to Computex, a massive annual semiconductor trade show in Taipei, an Nvidia spokesperson told WIRED.