xAI

Dell and Nvidia are building supercomputer racks for xAI

Published

on

Dell Founder and CEO Michael Dell has announced that his company is building supercomputer racks for Artificial Intelligence (AI) firm xAI.

There are many of the server racks that are shown in the image and ready for use. Prior to this announcement, Dell shared an image of a close view of the rack on social media site X, revealing a multi-slot system including dedicated cooling.

Advertisement
Advertisement

These new racks are part of Dell AI Factory, a solution designed for enterprise customers featuring NVIDIA AI chipsets out of the box. The AI solution includes Dell’s AI infrastructure and services with the Nvidia AI Enterprise software platform. It also features NVIDIA Tesnsor core GPUs, NVIDIA Spectrum -X Ethernet networking platform, and NVIDIA BlueField DPUs.

xAI has designed and developed the Grok large language model (LLM), first released in November last year. It competes against ChatGPT, Gemini, and other generative AI chatbots to provide text replies for users.

Advertisement

Earlier this year, the company announced Grok 1.5 and Grok 1.5 Vision with performance and visual data processing capabilities. However, the company is testing Grok 2 with an advanced GPU setup.

Dell supercomputer racks for xAI (Image Source – Michael Dell/X)

The AI firm plans to build a large data center for advanced AI. Last month, xAI raised $6 billion in series B funding to build advanced infrastructure and improve product performances.

Earlier this month, xAI founder Elon Musk revealed a plan to buy a new GPU and build a computing infrastructure. The AI company will initially use a 100,000 H100 liquid-cooled computing system within a few months. Musk confirmed that the next generation of computing power will come from 300,000 B200 Nvidia AI chipsets with CX8 networking by next summer.

Advertisement

These are some serious upgrades that xAI has planned for the future. Reacting to Michael Dell’s announcement, the xAI chief also said that Dell is assembling half of the racks for the supercomputer. It means the other half of the racks are pending and may be built on a newer AI Chip.

Advertisement
Comments
Exit mobile version