Nvidia plans to make its system that powers ChatGPT available in the cloud

Nvidia says it will make it's brand of generative AI available through Oracle Cloud, Microsoft Azure, Google Cloud Platform, and other cloud providers.

artificial intelligence machine learning and modern computer and picture id1209989402 100903158 lar

Nvidia recently announced fourth-quarter earnings, and all things considered, they weren’t that bad. They beat expectations even though sales were down. There was no panic on the conference call, no layoffs.

But amid all the talk about earnings and projections for 2023, CEO Jensen Huang dropped a surprise bombshell onto the earnings call with the announcement of DGX Cloud. It’s a deal to make its DGX systems available through multiple cloud providers, rather than installing the the necessary hardware on premises.

Nvidia sells GPU-based compute systems called DGX Pods. The same processors, networking, and Nvidia’s comprehensive AI Enterprise software stack from the Pods will be available through your browser, rather than sinking six or seven figures into hardware for your data center.

“AI supercomputers are hard and time consuming to build,” Huang told the conference call with Wall Street analysts. “Today we're announcing the Nvidia DGX Cloud, the fastest and easiest way to have your own DGX AI supercomputer. Just open your browser.”

Nvidia DGX Cloud will be available through Oracle Cloud infrastructure, Microsoft Azure, Google Cloud Platform, with others on the way, he said. Notably absent is AWS.

Through these participating CSPs, customers can access Nvidia AI Enterprise for training and deploying large language models or other AI workloads. At the pre-trained generative AI model layer, the company will be offering customizable AI models to enterprises that want to build proprietary models and services.

Generative AI is AI that's able to generate original content, the most famous example being ChatGPT, which runs on NVIDIA GPUs.

“We're going to democratize the access of this infrastructure and with accelerated training capabilities really make this technology and this capability quite accessible,” said Huang. “Our goal is to put the DGX infrastructure in the cloud so that we can make this capability available to every enterprise, every company in the world who would like to create proprietary data.”

That was about all he said. Nvidia reps declined to comment further but said details would be made available at Nvidia’s upcoming GTC conference in March.

Anshel Sag, principal analyst with Moor Insights & Strategy, doubts that DGX technology is really ever going to be designed for the masses, but he does think it will live up to Jensen’s promise to democratize access to AI technology more than it has in the past.

“I think this might be more of a software solution leveraging what the company already has on the hardware side, making it more accessible to anyone already used to using the cloud for AI workloads,” he told me.

What is Nvidia Researching?

Nvidia’s earnings were overall positive, even though consumer sales were way down. The data center business continued to do well in the company gave good guidance for the first quarter of 2023.

Notably, its R&D expenses have exploded in the past year. In Q4 of 2021, R&D was about $1.5 billion. This previous quarter, it was just under $2 billion. Going back through the historical earnings reports, there’s just no precedent for that level of a rise.

Nvidia’s R&D has steadily risen over the years but at a much slower pace. We’re talking a 33% increase in one year. Even with the Grace CPU, the inevitable Hopper successor and its networking efforts, that is a significant increase in R&D and it begs the question, what are they working on?

Copyright © 2023 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022