Intel, AMD form x86 group in effort to combat Arm

Intel, AMD vs. Arm, not Nvidia

Nvidia leads the is at the top of the AI chip market, beating both Intel and AMD. But experts see the x86 group’s efforts as less of a play against Nvidia and more an effort to combat Arm’s expansion.

“The x86 is under attack, but not from Nvidia — from Arm,” said Keith Townsend, president at The CTO Advisor, a Futurum Group company.

The focus of this advisory group is less on the AI accelerator chipsets and more on the common compute use cases for consumers and enterprise, Townsend said.

While cloud providers don’t release numbers, several have their own Arm-base processors, according to Russ Fellows, an analyst at the Futurum Group. Arm processors are prevalent on client devices, with huge companies like Apple and Google using them in their phones and computers.

“Arm is definitely getting a lot of traction in a lot of places,” Fellows said.

Arm architecture is almost nonexistent in the enterprise but common in the hyperscaler market, according to Gold. AWS, Google, Microsoft and Oracle all build their Arm own chips to optimize their cloud environments.

“How do you do that?” he asked. “You do that with Arm because that’s the business Arm is in.”

This isn’t solely about Intel and AMD losing ground to Arm, Gold said. This is also about what customers want to run as well as being able to take advantage of both Intel and AMD.

Good for enterprise customers

In the past, modifications had to be made to run applications on Intel if it was built for AMD, and the same is true the other way around, Gold said. Allowing applications to run on either chip without modification would be good news for users and application developers.

There is a saying that “x86 is x86,” meaning it didn’t matter if one bought Intel or AMD, Townsend said. However, in terms of interoperability, the type of x86 server does matter.

Intel has dominated the relationship with OEMs, Townsend said. With Dell, for example, users can purchase nearly any Intel server but are more limited on AMD versions.

The two chipmakers won’t smooth over every issue in interoperability, but they can help with several.

“What it will smooth over is that if I have a mix of [x86 servers], now I can get a little bit closer to a consistent operating model across my environment,” he said.

Intel’s issues and future competition

Intel has had several financial setbacks, falling behind Nvidia in supplying AI accelerators and losing ground to AMD in market capitalization. In August, Intel said it would lay off about 15% of its workforce, or about 15,000 jobs.

The x86 is under attack, but not from Nvidia — from Arm.

Industry experts don’t see the formation of the advisory board with AMD as being directly tied to its financial issues.

“What motivated them to do this is customers have been asking for this basically forever,” Townsend said.

The partnership with AMD doesn’t mean that the competition between the two CPU vendors will cool, Townsend said.

“I don’t think anyone is putting their knives away,” he said.

Developer cloud and MinIO

In other collaboration news, MinIO is going to provide enterprise object storage to the Intel Tiber AI Cloud, a developer cloud for AI workloads in Intel-based environments, including using the Gaudi accelerators. With MinIO storage, the Tiber cloud can be scaled up for better proof of concept.

The Intel Tiber AI Cloud allows developers to take test code with Intel accelerator and provide more options to consider, Fellows said. Intel has been known to foster the developer community and provide easy-to-use APIs to ensure their chips are used in application development, he said.

“This is really way more about fostering developers to even consider using [Intel’s] AI hardware,” he said.

MinIO’s role is that it is S3 compatible, and S3 is a ubiquitous interface, Fellows said. Users can run it in Intel Tiber Cloud, see that it works and theoretically run it in any cloud.

Adam Armstrong is a TechTarget Editorial news writer covering file and block storage hardware and private clouds. He previously worked at StorageReview.