September 09, 2026
News
In a massive set of releases this week, ARM has announced a determination to build the agentic AI computing platform for every use case.
The company said that it sees the world moving from a focus on generative AI to use cases that are much more agentic. What’s more, it says that central computing is not the way forward and instead is predicting a “computing continuum” at the edge, using AI tools and agentic apps to observe, measure, and take actions autonomously.
If this prediction is accurate, it will drive a new system-level challenge, ARM says. Workloads will become more distributed, models and compute requirements will need to be more diverse, systems will have to get more heterogeneous and autonomous. ARM says that embedded will need a platform that can be trusted to deliver consistency for developers, and flexibility for product and application differentiation, and it plans to be the provider of that platform.
This new slate of announcements can be broken into three parts, all aimed at Agentic AI dominance.
Arm CSS for Mobile 2 is the company’s new AI-native compute platform and it’s designed for the system level demands of agentic AI and cinematic mobile graphics
The new Arm Mali G2-Ultra NX is an advanced mobile GPU, and offers dedicated neural accelerators designed to unlock a new class of graphics experiences
The new Arm C2 CPU cluster combines Arm C2-Ultra, Arm’s most powerful mobile CPU, and C2-Pro CPUs with two SME2 units to support responsive on-device AI
CSS for Mobile 2 includes the new Arm Mali G2-Ultra NX GPU, the Arm C2 CPU cluster with SME2, enhanced system IP, physical implementations and a developer-ready software ecosystem in a platform optimized for the most sophisticated edge and mobile devices. This also means that ARM’s silicon partners will have a faster path to differentiated products and a shorter time-to-market. Developers will get to make use of the common platform for build and deployment at scale and create better experiences for their customers along the way.
The Arm CSS for Mobile 2 is the big news, although the power and flexibility of the new Arm Mali G2-Ultra NX and C2 CPU cluster are exciting and interesting in their own right.
Mali G2-Ultra NX is built for neural graphics, with neural accelerators integrated into the shader cores, allowing neural graphics workloads to run alongside graphics and compute workloads. This integration reuses the GPU memory system, coherent caches, and control structures, helping reduce data movement and improve efficiency across the pipeline, Arm says. Three neural technologies that are part of Mali G2-Ultra NX demonstrate what this enables:
Neural Super Sampling (NSS) reconstructs higher-resolution images from lower-resolution renders, reducing the amount of conventional rendering work needed to produce a high-quality final image.
Neural Frame Rate Upscaling (NFRU) generates intermediate frames to increase frame rates and deliver smoother gameplay.
Neural Super Sampling and Denoising (NSSD) combines neural upscaling with denoising to improve image quality in demanding ray-traced scenes with complex lighting and shadows.
The C2 CPU Cluster is built for responsive agentic AI, Arm says and the features are tailored to agentic use cases.
Performance: 15 percent higher single-thread performance and 12 percent faster application launch than the previous generation
Responsiveness: Up to 1.7 times faster at execution for AI models supporting voice, personal memory, and reasoning, with the C2 CPU cluster and SME2 helping orchestrate responsive on-device agents.
Efficiency: Supports up to two SME2 engines to deliver up to 70 percent performance uplift vs C1 CPU cluster.
Intelligence: C2-Ultra enables foreground responsiveness, while C2-Pro enables sustained performance across longer-running AI loops.
This is exciting news from Arm, and a great sign for the industry. Hopefully it will push companies away from generative tools that don’t have good business cases, and into AI applications that will serve customers’ needs.
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.
At Embedded Computing Design, he covers, AI, Edge Computing, Data Centers, Automotive, Industrial, Smart City, IoT and IIoT, Semiconductors, Healthcare, and lots more. He hosts weekly programs on YouTube, including the technology unboxing feature DevKit Weekly, and his news show ICYMI, and, along with Tiera Oliver, hosts the Embedded Insiders and Embedded Executive podcasts.
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.
Muck Rack: https://muckrack.com/ken-briodagh
Industry Moderator: https://www.microelectronicsus.com/speakers/ken-briodagh-1
Apple Podcasts: https://podcasts.apple.com/us/podcast/embedded-insiders/id1359230366
Spotify: https://open.spotify.com/show/5X07QpW7U2CM9ZbGxl4p40