Intel will create a rugged version of its Meteor Lake chip for potential use in billions of IoT devices next year following the deployment in December of Meteor Lake for PCs, Fierce Electronics has learned.
On Tuesday, CEO Pat Gelsinger announced Intel Core Ultra, the product name for Meteor Lake, for PCs at Intel Innovation. It ships Dec. 14 and will appear in an Acer Swift laptop.
While it makes sense that Meteor Lake would serve as a processor for all manner of client devices, not only PCs, the company has so far not formally announced the processor will work for small IoT devices at the edge in various consumer, retail and industrial uses where AI capabilities are useful.
With AI on practically any device at the edge, developers will be able to process data, including bulky video, without the need to send data to a server in the cloud. As such, it will be more economical, Intel believes.
Core Ultra adds a neural processing unit (NPU) to a GPU and CPU for its AI capabilities.
In an interview with FE at Intel Innovation on Wednesday, Sachin Katti, senior vice president and general manager of Intel’s network and edge group, confirmed Intel will expand Meteor Lake far beyond the PC to edge devices next year.
“We will take that chip for use in edge devices, things like Nook and mini computers,” Katti said. “They have the same AI needs.” Asked when it will happen, he answered, “next year.”
Also, they will be useful for the factory floor where they will be hardened to withstand high temperature and vibration, Katti said.
Intel will be able to offer the rugged Meteor Lake at a “significantly better TCO” than buying a discrete CPU from a competitor for general purpose uses and AI, he said.
In an assembly line, Meteor Lake with AI would be “perfectly suited” to be used to find machinery defects in an assembly lines.
Katti said an industrial AI processor will be timed well with more manufacturing returning to the US. AI in manufacturing will help lessen worries over labor shortages. “In the West, labor is not cheap, but AI will help run operations more efficiently than before,” he said.
Meteor Lake with its AI capabilities would find its most obvious advantages in retail, manufacturing and heathcare, he said. “Logistics and manufacturing are already using robots,” he said. Banks branches already operate with ATMs, in many places without tellers or bank employees.
Katti said broadly used, widespread automation with edge devices equipped with AI capabilities is coming and how it is used will be a problem for the entire ecosystem of application vendors, customers and even governments.
“We look at how to augment humans and how to make them productive,” he said. “Ultimately, we provide capabilities and these things have to be built by ecosystems. We provide technology and always have been a part of how it disrupts and how we [deal with it]. “
Automation empowered by AI chips will replace some jobs, and will pose long-term decisions on society broadly. Many businesses are only in the early stages of thinking about the implications.
“The question we all want to engage in as society becomes more automated is that there’s going to be surplus time all of us have and what should governments be thinking about to keep people productive.”
More in the short-term, companies will be grappling with AI decisions that are coming down fast.
“The world of AI is changing very fast,” Katti added. “Here at Intel, we can barely keep up with it. It’s a question of how should business leaders think about it and prepare for it. The pace of change is quickening.”
RELATED: Intel readies its AI PC neural processor, a new take on AI everywhere
Analyst Jack Gold of J. Gold Associates said Meteor Lake's use in the IoC space will open new opportunities for businesses. "Meteor Lake has a lot of the capabilities necessary to power an Edge solution... Think of a small NUC-like device that sits in a retail store, or runs a machine tool, etc. They will need to run some AI workloads going forward like image recognition or customer service, so with its NPU, Meteor Lake (Core Ultra) is a good fit.
"And since it is an SoC built on Foveros that includes chiplets, Intel will be able to add custom-to-purpose chiplets to an SoC as required. That gives Intel, and its customers, many potential solutions to edge-based processing requirements."