Arm Holdings has launched its latest chip designs, called Lumex, aimed at enhancing AI capabilities in top-tier smartphones and next-generation PCs.

The British semiconductor firm describes Lumex as its most advanced compute subsystem (CSS) platform to date.

It combines the capabilities of its high-performance CPUs, Scalable Matrix Extension version 2 (SME2), GPUs, and system IP. This integration is intended to expedite the introduction of AI-enabled devices to the market, facilitating experiences such as desktop-class gaming, real-time translation, and more intelligent assistants.

The Arm Lumex CSS platform will support real-time on-device AI applications, including voice translation and personalisation, with SME2-enabled Arm CPUs reportedly achieving up to five times faster AI performance.

In a blog post, Arm Client Line of Business general manager and senior vice president Chris Bergey said: “Partners can choose exactly how they build Lumex into their SoC – they can take the platform as delivered and leverage cutting-edge physical implementations tailored to their needs, reaping time to market and time to performance benefits.

“Alternatively, partners can configure the platform Register Transfer Level (RTL) for their targeted tiers and harden the cores themselves.”

The Lumex designs are available in four variants- C1-Ultra, C1-Premium, C1-Pro, and C1-Nano.

The C1-Ultra is tailored for flagship performance, suitable for large-model inference and generative AI tasks. The C1-Premium offers similar performance with improved area efficiency, targeting sub-flagship mobile segments and voice assistants.

The C1-Pro focuses on sustained efficiency for video playback and streaming inference, while the C1-Nano is designed for extreme power efficiency in wearables.

According to a Reuters report, Arm is set to hold a launch event in China in September 2025, to showcase the new designs, as many of the leading handset manufacturers are based there.