At its annual GTC conference this week, Nvidia announced new chips with a focus on supporting the development of AI technologies.
CEO Jensen Huang revealed Blackwell Ultra which will begin shipping in the second half of this year, as well as Vera Rubin graphics processing unit which is forecast to ship in 2026.
Nvidia’s new Blackwell Ultra family of chips will be capable of producing more tokens per second, generating more content in the same amount of time as its predecessor.
Blackwell Ultra will be offered in a version with two paired to an Nvidia Arm CPU, called GB300, and a version with only the GPU, called B300.
It will also come in options with eight GPUs in a single server blade and a rack version with 72 Blackwell chips.
With the Blackwell Ultra-based GB300 NVL72 rack-scale solution, for example, AI factories can achieve up to 50X the output for AI reasoning.
“In the last 2 to 3 years, a major breakthrough happened, a fundamental advance in artificial intelligence happened. We call it agentic AI,” said Huang. “It can reason about how to answer or how to solve a problem.”
The company says that cloud providers can use Blackwell Ultra to offer a premium AI service and make as much as 50 times the revenue from the new chips as the Hopper generation, reported CNBC.
The top four cloud companies have deployed three times the number of Blackwell chips as Hopper chips, claimed Nvidia.
As for the Vera Rubin, the system has two main components: a CPU, called Vera, and a new GPU design, called Rubin.
Vera is Nvidia’s first custom CPU design. Previously, Nvidia used design from Arm. The custom Vera design will reportedly be twice as fast as the CPU used in last year’s Grace Blackwell chips, said Nvidia.
When paired with Vera, Rubin can manage 50 petaflops while doing inference, more than double the 20 petaflops for the company’s current Blackwell chips. Rubin can also reportedly support as much as 288 gigabytes of fast memory.
After Vera Rubin debuts next year, Nvidia plans to release Rubin Ultra around a year later.
The chip architecture after Rubin will be named after physicist Richard Feynman. Nvidia’s Feynman chips are expected to be available in 2028.
Nvidia also used the event to announce new laptops and desktops using its chips, including two AI-focused PCs called DGX Spark and DGX Station that will be able to run large AI models.
Nvidia also announced a platform called Isaac GR00T N1 that will “supercharge humanoid robot development.” The company is working with Google’s DeepMind and Walt Disney Co. on the project, which will also be open to external developers.
The company also touted a partnership with General Motors that will integrate AI into its next-generation vehicles, advanced driver-assistance systems and even factories.
