The largest cloud computing service of the world, Amazon Web Services (AWS), is considering deploying new artificial intelligence chips from Advanced Micro Devices Inc., even though no final decision has yet been arrived at by the company, said a Reuters report quoting information from am AWS executive.
The comments were made at an AMD event when the chip maker unveiled its AI market strategy in response to rival Nvidia Corp.'s dominance of the sector.
While AMD revealed some technical details for an AI chip that could, in some cases, outperform the best current offerings from Nvidia in some criteria, the news pushed shares lower because AMD did not identify a flagship customer for the hardware.
A strategy for luring large cloud computing clients by providing a menu of all the components required to assemble the kinds of systems that would power services resembling ChatGPT, but letting clients choose which they prefer using open-standard connections, was described by AMD Chief Executive Lisa Su in interviews with Reuters.
"We're betting that a lot of people are going to want choice, and they're going to want the ability to customize what they need in their data center," Su said.
While Dave Brown, vice president of elastic compute cloud at Amazon, stated that AWS is exploring using AMD's new MI300 chips in its cloud services, AWS has not publicly committed to doing so.
"We're still working together on where exactly that will land between AWS and AMD, but it's something that our teams are working together on," Brown said. "That's where we've benefited from some of the work that they've done around the design that plugs into existing systems."
While Nvidia does sell its processors separately, it is also asking cloud service providers if they would be interested in offering a full system that Nvidia created under the brand name DGX Cloud. Nvidia's initial system partner is Oracle Corp.
According to Brown, AWS turned down Nvidia's request to collaborate on the DGX Cloud product.
"They approached us, we looked at the business model, and it didn't make a lot of sense," Brown said.
According to Brown, AWS prefers to build all of its own servers. In March, AWS began selling the Nvidia H100 chip, but only in conjunction with its own systems.
(Source:www.reuters.com)
The comments were made at an AMD event when the chip maker unveiled its AI market strategy in response to rival Nvidia Corp.'s dominance of the sector.
While AMD revealed some technical details for an AI chip that could, in some cases, outperform the best current offerings from Nvidia in some criteria, the news pushed shares lower because AMD did not identify a flagship customer for the hardware.
A strategy for luring large cloud computing clients by providing a menu of all the components required to assemble the kinds of systems that would power services resembling ChatGPT, but letting clients choose which they prefer using open-standard connections, was described by AMD Chief Executive Lisa Su in interviews with Reuters.
"We're betting that a lot of people are going to want choice, and they're going to want the ability to customize what they need in their data center," Su said.
While Dave Brown, vice president of elastic compute cloud at Amazon, stated that AWS is exploring using AMD's new MI300 chips in its cloud services, AWS has not publicly committed to doing so.
"We're still working together on where exactly that will land between AWS and AMD, but it's something that our teams are working together on," Brown said. "That's where we've benefited from some of the work that they've done around the design that plugs into existing systems."
While Nvidia does sell its processors separately, it is also asking cloud service providers if they would be interested in offering a full system that Nvidia created under the brand name DGX Cloud. Nvidia's initial system partner is Oracle Corp.
According to Brown, AWS turned down Nvidia's request to collaborate on the DGX Cloud product.
"They approached us, we looked at the business model, and it didn't make a lot of sense," Brown said.
According to Brown, AWS prefers to build all of its own servers. In March, AWS began selling the Nvidia H100 chip, but only in conjunction with its own systems.
(Source:www.reuters.com)
Homepage
Send to a friend
Printable version
Share
