Arm launches first edge AI platform with billion-parameter models

Arm has launched its first edge AI platform, the Armv9 system, designed to democratize access to AI development tools for businesses of all sizes. The platform will be available through Arm Flexible Access, the company’s low-barrier entry program, enabling both startups and large-scale manufacturers to experiment with advanced AI capabilities without prohibitive costs.

Why this matters: The move addresses growing demand for real-time AI processing at the edge, where data is created, rather than relying on centralized cloud computing that can’t deliver the privacy, speed, and energy efficiency many industries now require.

Key details: The Armv9 edge AI platform combines Arm Cortex-A320 CPU and Arm Ethos-U85 NPU to enable on-device AI models with over one billion parameters.
• The Cortex-A320 will be available to partners in November 2025, with the Ethos-U85 rolling out in early 2026.
• The platform features larger memory systems, advanced security, and support for modern programming frameworks.
• Target applications include smart cameras, industrial automation, and advanced human-machine interfaces that enable interactions through vision, voice, and gesture recognition.

In plain English: Think of this as giving smartphones and smart devices their own mini-brain instead of having to phone home to a data center every time they need to make a decision. The CPU (central processing unit) acts like the device’s main thinking engine, while the NPU (neural processing unit) is specifically designed to handle AI tasks like recognizing faces or understanding speech.

What they’re saying: Paul Williamson, senior vice president and general manager of Arm’s IoT Business, emphasized the shift toward edge computing.
• “AI has moved beyond the cloud, and the next evolution is happening where data is created,” Williamson said. “Industries now need real-time intelligence, privacy and energy efficiency that centralized compute can’t deliver.”
• “Innovation depends on access,” he added. “By bringing the Armv9 edge AI platform into Arm Flexible Access, we’re offering no-cost or low-cost access to our latest technology optimized for IoT, with flexible terms that let both startups and large silicon users experiment and scale.”

Looking ahead: Williamson indicated Arm will focus on developing edge devices running adaptive AI models as the technology becomes more sophisticated.
• “As AI becomes more capable, accessibility becomes even more essential,” he said. “Arm Flexible Access will continue expanding to give all developers the same opportunity to innovate on the world’s most advanced compute platform.”