On the same day Apple revealed its new iPhone 17 lineup, chip designer Arm is introducing its next generation of processors bundled into its new Arm Lumex platform. These processors will likely make their way into premium Android phones coming in 2026 — and improve their AI capabilities without draining device batteries faster, Arm says.

«It’s pretty amazing how there’s been kind of this insatiable amount of performance being asked for [by our customers], and a lot of it’s around AI, as well as some graphics workloads,» said Chris Bergey, senior vice president and general manager of Arm’s client line of business.  

Arm’s processors have traditionally been the centerpieces of holistic systems-on-a-chip, which power smartphones. For instance, Arm’s previous top-end central processing unit, the Cortex X925, released last year, is featured in Samsung’s Exynos 2500 chipset, which powered the Galaxy Z Flip 7 and MediaTek’s Dimensity 9400 chipset, found in the Oppo Find X8 Pro. 

It’s likely that Arm’s new chips, either on their own or bundled into the company’s Arm Lumex platform, will power premium Android phones and other devices next year. But it faces more competition as Qualcomm (which used Arm CPUs in older chips like the Snapdragon 8 Gen 3) has shifted to its internally designed Oryon CPUs in its latest silicon — which was the subject of its own tech licensing clash over the last few years between the two chip companies. 

Arm’s new CPU range has shifted with a new naming paradigm. The successors to the X925 are two different chips: the highest-end C1-Ultra, which boasts 25% greater performance over its predecessor, and the next-most-powerful C1-Premium (no performance improvement given). The successor to the A725 chip is the C1-Pro, which has 12% greater efficiency. 

These CPUs also benefit from an evolution of their chip architecture called Scalable Matrix Extension version 2, which enables better AI performance. 

The new Arm Lumex platform combines these chips with the new Mali G1-Ultra GPU (which Arm says has 20% better performance and twice the ray tracing as its predecessor) for a system that can be plugged into larger chipsets. The end result: up to 5x improvement in AI performance, 4.7x lower latency for speech-based workloads (think live translation) and 2.8x faster audio generation, Arm says. 

The goal is faster performance without increasing the battery drain when running AI tasks that tax processing power. For example, a yoga tutor demo app running Arm’s new chips saw a 2.4x boost in text-to-speech to give faster feedback to users, the company said in a press release. 

As a supplier of chips and technology, it’s ultimately up to the phonemakers using Arm’s silicon to decide how much (or little) of the advancements they integrate into their devices. But as demand for generative AI like ChatGPT increases, so too does the drive to get that functionality working on phones as efficiently as possible, rather than relying on slower responses going to and from the cloud. So what will consumers see with devices running C1-Ultra chips and Arm’s new technology?

«I think what they’re gonna see is the ability to run amazing on-device AI and to do so with significant power savings, significant performance increases, and also third-party support,» Bergey said. «Not just first-party devices, but also third-party devices.»