What AMD and Intel’s Alliance Means for Data Center Operators

AMD and Intel this week announced plans to collaborate together to improve and shape the future of x86. For data center operators, it’s a move that can result in improved software and hardware performance, and easier management of their IT infrastructure, analysts say.

The two rival chipmakers on Tuesday (Oct. 15) announced the creation of an x86 ecosystem advisory group to ensure architecture compatibility, simplify software development and develop innovative and scalable solutions.  

"This announcement probably surprised some people," said Daniel Newman, CEO of the Futurum Group, a research and advisory firm. "It’s unprecedented to see these two companies work together in any capacity and agree on anything other than x86 as an architecture."

AMD and Intel’s large tech partners have signed onto the effort, including Broadcom, Dell Technologies, Google Cloud, Hewlett Packard Enterprise, Lenovo, Microsoft, Oracle, and Red Hat.

At issue is the increased competition from Arm-based chips and the fact that while AMD and Intel use the same x86 instruction set architecture (ISA), they’ve historically done some things differently – such as security or the implementation of AVX-512, a workload-specific accelerator that enables AI and HPC workloads to run faster.

Related:Broadcom’s ‘Nvidia Moment’ Has Arrived. Now It Needs to Deliver

“The whole notion of commonality of x86 was starting to go down these two paths that weren’t 180 degrees out of phase, but they started to kind of drift into their own kind of specialized world,” said Matt Kimball, vice president and principal analyst at Moor Insights & Strategy.

Potential Impact of the x86 Ecosystem Advisory Group

That divergence has forced software makers like Broadcom’s VMware and Nutanix to modify some of their source code to support Intel and AMD chips. The same goes for hardware makers like HPE and Lenovo that offer management software and cloud service providers that build software stacks and management platforms for their Intel and AMD environments, Kimball said.

In its announcement Tuesday, Intel and AMD said the new advisory group will work to create a unified set of instructions and architectural interfaces to enhance compatibility, predictability and consistency across x86 products, while providing future direction on architectural enhancements.

More specifically, they said the group’s goals are threefold: enhance hardware and software compatibility while accelerating new cutting-edge features; simplify architectural guidelines to enhance software consistency and standardize interfaces across Intel and AMD x86 products; and enable greater, more efficient integration of new capabilities into operating systems, frameworks and software.

Related:AWS Launches Trainium2 Custom AI Chip, Data Center Upgrades

“If this advisory group – and Intel and AMD – truly works to its potential, what you will see is software that runs more efficiently and can run across both platforms more easily and has more capabilities built into it,” Kimball said.

IDC analyst Ashish Nadkarni agrees, saying it removes friction between AMD and Intel, making it easier for independent software vendors to optimize their applications. They don’t have to worry about nuances between Intel and AMD.

“What it means for enterprises is it removes any friction or inconsistencies between the experience you get if you were running x86 apps on AMD or Intel,” Nadkarni said.  

Why the x86 Group Can Benefit Data Center Operators

If the advisory group is successful with its goals, enterprises and other data center operators will be able to manage their infrastructure more easily, their CPUs will run more efficiently, and they will have better performing software, Kimball said.

“It’s easier, more cost-effective and more performant,” he said. “It’s all goodness. There’s really no downside to this for data center operators.”  

For example, enterprises that deploy a mixed AMD and Intel server environment have become more common in recent years because they feel one chip handles one workload better, while another chip handles a different workload better, and in some cases, they want to use both processors to run general-purpose workloads too, he said.

Related:Biden Tightens Chip Controls on China as Clock Ticks Down

However, those enterprises today may have a hard time live migrating a VMware or Nutanix virtual machine (VM) from an AMD-powered server to an Intel-powered server because there are differences underneath in the microarchitecture, Kimball said.

More specifically, they won’t be able to live-migrate a VM from an AMD server to an Intel server if the VM uses AMD’s secure encrypted virtualization because Intel has its own security for VMs. That compatibility problem goes away if Intel and AMD settle on a unified security solution, he said.

Cloud service providers also benefit from Intel and AMD’s interoperability efforts. With better compatibility, they can better optimize their software stacks and management platforms, which reduces operational costs, Kimball said.

“They can simplify the development, testing and rolling out of their software stacks by having a common set of instructions,” he said. “That would be a huge win, and it also means they could deliver more highly optimized instances to their customers.”

Brewing Battle Between x86 and Arm

AMD and Intel’s alliance was a surprise for many. But industry analysts said their partnership makes sense and is much needed, given the threat that Arm poses in both the consumer and data center space.

While x86 processors still dominate the data center space, Arm has made inroads with cloud providers Amazon Web Services, Google Cloud and Microsoft Azure building their own Arm-based CPUs and startups like Ampere having entered the market in recent years.

Intel and AMD’s partnership confirms how strong Arm is as a platform in the PC, data center and smartphone markets, the Futurum Group's Newman said. But the two giant chipmakers still have the advantage of having a huge installed base and significant market share. Through the new x86 advisory group, AMD and Intel can benefit by making it easier for data center operators to leverage x86, he said.

“This partnership is about the experience of the x86 customer base, trying to make it stickier and trying to give them less reason to potentially move off of the platform is valuable,” Newman said. “x86’s longevity will benefit meaningfully from less complexity and making it easier for customers.”