Nvidia Wants to Have Every Chip Inside AI Data Centers

Nvidia is hyping has launched its new Vera Rubin chip this week, revealing new performance indicators GPU and CPU combo ahead of the enemy AMDThe annual event in San Francisco on Thursday.

At a long technical conference last week at the company’s headquarters in Santa Clara, California, Nvidia executives boasted to a small group of reporters about the chip’s increased power and capabilities. The biggest takeaway: Nvidia, a long-time GPU maker, is trying to position itself as a supplier of CPUs that can power AI assistants.

Even GPUs still great equipment that companies use to train and run their AI models, the industry’s shift to more complex, operating systems has increased the demand for CPUs, which can handle data flow, networks, and other software tasks. This is probably one reason why Nvidia has been keen to promote itself as a supplier of complete AI systems rather than AI chips.

Vera Rubin is the successor of Nvidia instead hybrid superchip Grace Blackwell and represents the future framework for supporting the AI ​​industry. It is designed to provide one CPU for every two GPUs. In one Vera Rubin NVL 72 high-end processor, there are 36 Vera CPUs for each of the 72 Rubin GPUs. Nvidia is also selling the Vera CPU as a standalone product, and it has he reportedly told Chinese customers this can be ready as soon as August.

CPU chip Nvidia is using its new Vera Rubin hardware.

Courtesy of Nvidia

Nvidia executives confirmed that its new Vera Rubin NVL72 racks – multiple chips crammed onto a single liquid-cooled platform – are more “plug-and-play” than its previous devices. During a brief tour of Nvidia’s data lab in Silicon Valley, Nvidia executives shared that OpenAI already has a Vera Rubin rack in use.

Nvidia CEO Jensen Huang did not appear at the conference in Santa Clara last week; was in Japan announcing the new chipmaker cooperation with several Japanese companies creating AI for robotics. The brief was instead led by Ian Buck, Nvidia’s vice president of accelerator technology and the architect behind the company’s CUDA software.

“We are on the map to develop new architectures, not only for GPUs but also for CPUs,” Buck told reporters. “We’re going to keep innovating, because it’s do or die in Silicon Valley.”

The meetings were held in Huang’s brief headquarters, where several desks nearby were piled high with bags of Taiwanese snacks the CEO brought with him during his recent trip to Computex, the largest annual semiconductor show in Taipei, an Nvidia spokesperson told WIRED.