For engineers and designers using sensors in applications that operate at the edge, the news about the new Panther Lake processor from Intel will be important.
No, Intel is not making its own sensors to work in chiplet fashion or to be plunked right on the board next to its new Panther Lake processor, but Intel is supporting the huge array of sensors already made by Texas Instruments and many others with the ability to add AI functionality for things like robots, industrial machines, satellites and many other devices--some complex and some simple.
“AI at the edge is a critical area for Intel as it requires less in GPU and more in CPU acceleration, where Intel can be a major player as we move to an inference-based AI market over the next two to three years,” analyst Jack Gold of J. Gold Associates told Fierce. “Think NPUs attached to a CPU with integrated GPU.”
“It’s not just about humanoids, which seem to get lots of headlines but frankly are probably a much longer-term project if they are ever real. But there a huge number of AI edge-based systems that will emerge in retail, manufacturing, medicine, industrial equipment and more where Intel already has a pretty significant installed base and where Intel can leverage that if they have compelling or at least competitive products.”
Arm has also targeted that area at the lower end and giant Nvidia has interest in the space as well, Gold said. “While Nvidia has targeted the edge space, they have a long uphill battle here as they have minimal market share and need a much stronger CPU story, which is partly why they announced a partnership with Intel.”
On Sept. 18, Nvidia announced a $5 billion investment and strategic partnership with Intel to co-develop new data center and PC products by combining Nvidia AI technology with Intel’s CPUs. Intel will design custom x86 CPUs for Nvidia’s AI platforms. Also, Intel will make system-on-chip processors for PCs that integrate Nvidia RTX GPU chiplets.
What will AI with Panther Lake do for designers at the edge? Intel recently showed off several applications, including a humanoid robot attached to a Panther Lake PC which gave the robot movement directions, gesturesand more. There was also a retail application, a traffic management app and one for medical, giving doctors the ability to text or speak directions for an AI to find a pertinent medical scan for a patient.
While Panther Lake at the edge will be critical to Intel, as Gold put it, the question remains if it becomes critical for developers. Given the long tradition with developers using Intel x86, it’s fair to say Panther Lake will be important to developers, yes, if not exactly critical. Intel already works with about 4,000 companies working in the edge application arena.
RELATED: Intel’s Panther Lake processor hits the AI edge—and humanoids!
Panther Lake is the code name for the Intel Core Ultra series 3, which is being made on the 18A process at Ocotillo Fab 52 in Chandler, Arizona, near Phoenix. Panther Lake will be shipping later this year.
Potentially more important than the actual processor, at least for developers, will be a developer’s reference board containing the Panther Lake processor for building edge applications for industry and business use. Pricing and availability will be announced after the first of the year. The board will be separate from a Intel’s Robotics AI Suite and will be optimized to run the Suite’s software, frameworks and benchmarks.
Note: The discussion of “AI at the edge” has provoked an old debate about what constitutes the edge in computer frameworks. It is easy to see the edge as the edge of the network, but most engineers think of edge computing as those applications and devices that work without the constant connection to the cloud or even a data center. With AI compute performed in a white box PC running Panther Lake, many functions of applications will be performed without needing the resources of a data center.