Arm Pushes Edge AI to a Higher Level, Marries Embedded with AI and IoT

By Ken Briodagh

Senior Technology Editor

Embedded Computing Design

April 22, 2024


Arm Pushes Edge AI to a Higher Level, Marries Embedded with AI and IoT

One of the major threads running through this year’s embedded world was the growing entwining of IoT and embedded computing. It seemed that almost every player, especially at the edge, was touting their ability to bring AI and IoT to the Edge, with more power and embedded tech.

In a recent announcement, ARM brought that confluence into the big leagues by releasing a new NPU purpose-built for edge applications, and a new IoT Reference Design Platform for accelerating IoT sensor systems.

The Arm Ethos-U85 NPU is designed to deliver a 4x performance improvement for high performance edge AI, according to the company, and is ideally suited to factory automation and cameras for AI computer vision applications. The IoT Reference Design Platform is called Corstone-320 and ARM has said it will unite embedded IP with virtual hardware with the goal of speeding IoT deployments in voice, audio, and vision systems across verticals.

This is a big deal for embedded, IoT, and AI companies. The ARM ecosystem is huge, reportedly more than 15 million developers, and that kind of muscle means that new solutions will have extensive and streamlined support through ARM’s software enablement and toolset. Those tools offer speedy development and prototyping, quicker time-to-market, and simpler scaling for IoT and AI at the edge.

Developers are being forced to work in every arena from hardware to software, cloud to edge, in many deployments in the modern marketplace. Tools like these are designed to help make that process more approachable and efficient for companies and engineers alike. Complexity is the order of the day, and simplification is the “killer app” that most developers need to bring their solutions to the market.

Of course, simplification can’t come at the cost of effective and powerful solutions, so access to integration tools and AI and embedded frameworks can make all the difference.

ARM’s Ethos-U85 is its third generation NPU for edge AI, and according to the announcement, it is the company’s highest performance, most efficient Ethos NPU ever. It’s four times as powerful and 20 percent more efficient than its predecessor. Ethos-U85 includes the ARM toolchain and adds support for AI frameworks like TensorFlow Lite and PyTorch, Transformer Networks, and Convolutional Neural Networks (CNN) for AI inference. It works with Armv9 Cortex-A CPU to accelerate ML tasks and reportedly bring power-efficient edge inference into a broader range of higher-performing devices.

The new IoT Reference Design Platform, Corstone-320, brings together ARMs Cortex-M85 CPU, Mali-C55 Image Signal Processor and the new Ethos-U85 NPU, to deliver the performance that AI applications need to run real time image classification and object recognition, enable voice assistants with natural language translation on smart speakers, and many other IoT and AI applications. According to the announcement, the platform includes software, tools, and support, including Arm Virtual Hardware that will let designers speed time-to-market by enabling software development to start ahead of hardware availability through virtualization.

This kind of innovation is required for the newest and most critical applications in embedded systems, and ARM is clearly setting itself to be among the leaders in enabling this confluence of IoT, AI, and embedded.


Ken Briodagh is a writer and editor with two decades of experience under his belt. He is in love with technology and if he had his druthers, he would beta test everything from shoe phones to flying cars. In previous lives, he’s been a short order cook, telemarketer, medical supply technician, mover of the bodies at a funeral home, pirate, poet, partial alliterist, parent, partner and pretender to various thrones. Most of his exploits are either exaggerated or blatantly false.

More from Ken