Surging Demand for AI Servers

Dell Technologies Reports Soaring Demand for AI Servers

It’s official. Demand for AI servers isn’t subsiding, which is good news for Nvidia.

On Thursday, the last two major makers of AI servers reported their financial results— Dell Technologies and Hewlett Packard Enterprise —and pointed to soaring demand for their AI-related offerings.

Dell Chief Operating Officer Jeff Clarke said the backlog for orders for the company’s AI servers nearly doubled to $2.9 billion during the company’s fiscal fourth quarter, which ended January. “The demand for these things is in excess of the supply,” he said on a call to discuss the earnings. “That continues certainly into next year.”

Dell shares closed up 32% on Friday. Nvidia was up 4%.

The executive also said Dell’s flagship PowerEdge XE9680 Rack Server, which incorporates Nvidia graphics processing units, was the “fastest-ramping” solution in the company’s history. AI orders are spread across several newer and old GPUs, he said.

Hewlett Packard Enterprise Echoes Strong Demand for AI Systems

HP Enterprise expressed similar thoughts about trends in the market. “AI server demand remains very strong,” CEO Antonio Neri said on the call with investors. “We are capturing the explosion in demand for AI systems,” adding lead times for GPU orders are “still elevated” at 20 weeks or more.

The companies’ comments are important because some skeptics have keyed in on falling lead times for Nvidia’s H100 as a sign of weakening demand. Dell’s and HP’s remarks show orders are transitioning to other newer products like the H200 and the GH200 and that demand will stay robust over the longer term.

Arm AI Superchip Driving Revenue Growth

Last month, chip designer Arm highlighted Nvidia's GH200 AI Superchip data-center systems as a major revenue driver in the cloud server market. Major players such as Dell Technologies, Hewlett Packard Enterprise, Lenovo, Quanta, and Supermicro are utilizing the GH200 to tackle complex challenges.

Rapid Release of Chips by Nvidia

Nvidia is speeding up its chip release schedule, moving from a two-year product cycle to a one-year cadence for its AI datacenter portfolio. The B100, successor to the H100, is set to be launched in the upcoming quarters.

Dell's Hint at an Unreleased B200

During a recent call with investors and analysts, a Dell executive mentioned a B200 chip from Nvidia that is expected to require 1,000 watts per GPU. This chip, not included in Nvidia's October roadmap, hints at a faster product development pace by the company.

Nvidia has not provided details on the upcoming B100 or officially announced the B200 chip. More information may be revealed at the upcoming developer conference.