Nvidia moves into AI cloud leasing to spread new technology: Report
Nvidia Corporation CEO Jensen Huang on Tuesday unveiled the company’s plans to rent out powerful and expensive supercomputers used to develop artificial intelligence technologies such as ChatGPT to almost any company.
While that access doesn’t come cheap — $37,000 a month for eight of Nvidia’s flagship A100 or H100 chips — making it available to a wider range of enterprise customers could fuel the AI boom that has sent Nvidia shares up 77 percent this year. It is about five times more valuable than longtime rival Intel Corp.
The Santa Clara, Calif.-based company already dominates the field of artificial intelligence chips and has helped partners such as Microsoft Corp build massive systems for ChatGPT creator OpenAI’s services to answer questions with human-like text and generate images from prompts.
At Nvidia’s annual software developer conference on Tuesday, Huang said the company is working with partners such as Oracle Corp to offer access to Nvidia’s DGX supercomputers with up to 32,000 Nvidia chips to anyone who can log in with a web browser.
“The iPhone moment of artificial intelligence has begun,” Huang said in his virtual keynote, referring to Apple Inc.’s opening to smartphones.
Nvidia also partnered with Microsoft and Alphabet Inc to offer its supercomputers, which are used to create new AI products, as a service, Huang said. Nvidia unveiled new chips and software on Tuesday designed to make products like chatbots much cheaper to run on a daily basis after they’re built on supercomputers.
These products “are years ahead of the competition,” said Hans Mosesmann, a semiconductor analyst at Rosenblatt Securities. “Nvidia’s leadership on the software side of AI is not only monumental, it’s accelerating.”
Nvidia is also working with AT&T Inc to make trucking more efficient, working with quantum computing researchers to speed up software development and working with industry giant Taiwan Semiconductor Manufacturing Co to speed up chip development, Huang added.
Nvidia’s new rental service, called DGX Cloud, could give many more developers access to tens of thousands of its chips at once. Biotech company Amgen Inc and software company ServiceNow Inc have started using the service, Nvidia said.
Nvidia also launched AI Foundations, a service that helps companies train their custom AI models. Several large owners of stock image databases plan to use the service, which would avoid legal questions about the copyright of images used to generate AI content.
Huang also announced technology that speeds up semiconductor design and manufacturing. The software uses Nvidia’s chips to speed up the step between the software-based design of the chip and the physical manufacturing of the lithography masks used to print it onto the silicon.
Those calculations can take a traditional computing chip two weeks to complete, but Nvidia said Tuesday that its chips and software can handle the task overnight, reducing the power used by the task from 35 megawatts to 5 megawatts.
Nvidia said it is working with ASML Holding, Synopsys Inc and TSMC to bring it to market. TSMC will begin preparing the technology for production in June, Huang said.
Read all the Latest Tech News here.
