AMD Hires Intel Exec Who Led Client AI Team For ‘Meteor Lake’ CPUs

John Rayfield leaves his job as head of client AI at Intel for a new role at AMD as the two chip companies bring to market a new class of processors that accelerate AI workloads on PCs. In his role at Intel, Rayfield led development of low-power AI technologies, including the neural processing unit, for the chipmaker’s upcoming Core Ultra ‘Meteor Lake’ processors.

AMD has hired an Intel executive who led the development of low-power AI technologies for the semiconductor giant’s upcoming Core Ultra “Meteor Lake” processors.

John Rayfield was vice president and general manager of client AI at Intel until this month, when he left to join rival AMD as corporate vice president of AI silicon, according to his LinkedIn profile.

Rayfield did not respond to a request for comment. Neither did AMD or Intel.

[Gelsinger: Intel Is Prioritizing AI Sales Enablement For Partners]

Rayfield (pictured) left Intel as both the semiconductor giant and his new employer seek to compete with a new class of client processors that accelerate AI workloads on PCs, which the two companies expect to become a top focus for individuals and organizations alike amid broader hype for generative AI applications.

In December, Intel is set to debut its Core Ultra processors, which the company has said will “usher in the age of the AI PC.” As outlined at last week’s Intel Innovation event, the development of processors that can enable a new wave of AI-powered PC applications is a key part of the chipmaker’s strategy moving forward.

“We see the AI PC as a sea change moment in tech innovation,” Intel CEO Pat Gelsinger said last week.

Rayfield’s departure comes on the heels of the departure of another key AI executive, Kavitha Prasad, who left Intel for a position at Amazon Web Services. Prasad led Intel’s data center AI strategy until she left in July to become general manager of the perimeter protection security business at AWS, CRN reported Monday.

Rayfield’s Purview Included NPU Development For Meteor Lake

As the head of client AI at Intel, Rayfield oversaw development of hardware and software for low-power AI, according to a biography on a Microsoft web page.

His purview included the neural processing unit (NPU) in Intel’s upcoming Core Ultra processors, which were previously known under the Meteor Lake code name.

The NPU is part of Intel’s three-prong strategy for processing AI workloads with the Core Ultra chips.

Intel said last week that the NPU is a low-power AI engine best for “sustained AI and AI offload” while the chip’s CPU is meant for providing fast responses in “lightweight, single-inference low-latency AI tasks” and the GPU is best for “AI infused in the media, 3-D applications and the render pipeline.”

In the spring, Rayfield was a public-facing spokesperson for the Core Ultra processors and Intel’s AI PC strategy, which supposes that organizations will want to increasingly move AI workloads from the cloud to personal computers to lower costs, improve security and reduce latency.

“What we’re seeing is there’s lots and lots of reasons to migrate more of this across to the client, which is a really different compute model,” he said in a May Intel video.

Rayfield became general manager of client AI in June of last year, but he had been leading development of the NPU before then for nearly two years when it was referenced as the VPU, which previously stood for vision processing unit. Intel’s VPU technology came from its 2016 acquisition of Movidius, and it was originally designed for computer vision applications in edge devices.

Prior to joining Intel as general manager of VPU intellectual property in November 2020, Rayfield was CTO at British chip designer Imagination Technologies for a little over a year.

According to his bio on Microsoft’s website, Rayfield “has been involved in AI accelerator development for several years” and in low-power compute technologies for most of his career.

AMD Beat Intel To Market With Chips For AI PCs

Rayfield is joining AMD as the company moves forward with its own AI PC strategy, which kicked off earlier this year with the launch of the chip designer’s Ryzen 7040 series processors for laptops.

The processors each include an AI engine that is designed to enable “accelerated multitasking, increased productivity and efficiency and advanced collaboration,” according to AMD’s website. Development of the AI engine originated from AMD’s $49 billion acquisition of Xilinx last year.

“I think there are a number of areas that we expect to play a pretty key role in really defining a new set of user experiences over the next 12 to 18 months,” said Matt Unangst, senior director of AMD’s commercial client and workstation business, in an interview with CRN last month.

One significant application AMD plans to accelerate with its Ryzen AI technology is Microsoft Copilot, the $30-per-user, per-month add-on that brings a variety of generative AI capabilities to Windows 11 and Microsoft 365. Copilot will also take advantage of Intel’s upcoming Core Ultra chips, Intel said last week.

“There will be a significant portion of Microsoft Copilot that can and will take advantage of our Ryzen AI technology,” Unangst said.