Qualcomm’s New Chip Powers AI Wearables
▼ Summary
– Qualcomm has announced a new “Snapdragon Wear Elite” chip designed for AI wearables like pendants, pins, and potentially display-free smart glasses.
– The chip features dual AI processors: a low-power eNPU for basic functions and a more powerful Hexagon NPU capable of handling two billion parameters on-device.
– It offers significant power efficiency improvements, including 40% less power for GPS and support for fast charging that provides 50% battery in about 10 minutes.
– The Elite chip includes enhanced connectivity (satellite, 5G, UWB, Bluetooth 6.0), a 5x CPU performance boost, and supports Android, Wear OS, and Linux.
– Qualcomm’s investment signals ongoing industry interest in AI wearables, with companies like Google, Apple, and OpenAI also reportedly exploring the category.
The landscape of personal technology is shifting, with a new wave of AI-powered wearables poised to enter the market. Qualcomm’s newly announced Snapdragon Wear Elite chip is designed to be the engine for this emerging category, moving beyond traditional smartwatches to power devices like AI pins, pendants, and potentially even smart glasses without displays.
Qualcomm describes the Elite as a “wrist plus” processor, meaning it is intended to exist alongside, not replace, its existing W5 Plus chip for smartwatches. The company is targeting manufacturers who want to build compact, always-connected gadgets that leverage on-device artificial intelligence. For more complex augmented reality glasses, Qualcomm expects makers to utilize its dedicated AR platform instead.
A significant upgrade comes from its fabrication on an advanced 3nm process, which contributes to greater efficiency. The chip features a dual-tier AI processing system: an eNPU for low-power tasks like listening for wake words and an enhanced Hexagon NPU for heavier computational workloads. Qualcomm states this Hexagon unit is capable of running AI models with up to two billion parameters directly on the device, processing roughly ten tokens per second.
While retaining a co-processor design similar to its predecessor, the Wear Elite improves power management to allow the main chip to handle more functions. Specific gains include GPS tracking that uses 40 percent less power. For charging, it supports 9V quick charging, promising around a 50 percent battery top-up in just ten minutes. Qualcomm also estimates an overall 30 percent improvement in “days of use” for battery life, a general metric that translates to longer periods between needing to plug in.
Connectivity receives a major boost with new support for satellite communications, 5G networks, ultra-wideband (UWB), and Bluetooth 6.0. Performance sees substantial leaps, with the CPU reportedly five times faster and the GPU now capable of driving 1080p animations at 60 frames per second. Beyond Android and Wear OS, the chip will also support Linux, a move aimed at assisting startups that wish to develop devices running their own proprietary software.
This development signals continued industry faith in AI wearables, despite the absence of a major commercial success in forms like pins or pendants. When a foundational component supplier like Qualcomm invests in creating specialized silicon for this niche, it indicates underlying demand from device makers. This aligns with a broader industry push; Google has publicly discussed building an ecosystem of AI hardware that includes wearables, Apple is rumored to be considering its own AI wearable, and the collaboration between Jony Ive and Sam Altman suggests OpenAI may also be exploring a physical AI device.
(Source: The Verge)
