Previous Qualcomm Snapdragon Wear chips were mostly designed for smartwatches. The new Snapdragon Wear Elite, which was first announced for MWC 2026 in Barcelona, is believed to offer more platforms than that. The chip is built on a 3nm process node and includes Qualcomm’s Hexagon NPU. That neural processing unit is built to handle low-level AI tasks, though Qualcomm has also included an additional eNPU AI accelerator for low-power AI use cases.
The chip maker further claims that it has increased the CPU by five times in single-thread performance compared to the previous W5 Gen 2 chip. The chip maker has also improved the maximum frame rates you can achieve from the Wear Elite’s GPU. These upgrades could make the next-generation smartwatch a little faster when loading apps. However, Qualcomm’s main goal is to introduce new use cases for its platform, whether through pins, pendants or AI-focused hubs.
This new chip promises to be able to handle 2 billion parameter AI models on the device. To put this in perspective, Google’s smallest AI model, Gemma, is a 270 million parameter model. This means that the chip is technically capable of handling much smaller conversational models. How this plays out in reality is still to be determined. Additionally, Qualcomm claims that it has enhanced image stabilization for the smaller cameras. It supports cameras capturing images and videos at 1080p and 60 fps. These can be useful for AI vision models. Additionally, any type of AI vision model will likely need to run on the cloud, requiring an always-present Internet connection. The need for a constant 5G or Wi-Fi connection has stymied previous efforts at AI wearables — even if you ignore AI’s tendency to give inconsistent answers or outright lie about what it sees.
A future full of AI wearables
John Kehrli, Qualcomm’s senior director of project management, told Gizmodo that the chip maker is already in talks with multiple companies, all of whom are trying to design some type of AI wearable that eventually makes sense. Kehrli mentioned how work is being done on a variety of form factors beyond AI glasses, such as Meta’s Ray-Ban SmartGlasses and AR glasses. There’s also Razer, which is proposing players they’ll want a Project Motoko gaming headset with two camera lenses so the AI can see what you’re playing and offer (often incoherent) commentary.
Then there’s a device like the Looki L1, a self-described “personal AI wearable.” It may look like the Nickelodeon Splat logo, however it is designed to hang around your neck and provide commentary or simply record your life with the help of the built-in camera that can capture 1080p video or photos. That device is currently running on Qualcomm’s W5 Gen 2 chip.
Can Qualcomm chip overcome human flaws?
So far, the highest-profile examples we have of AI wearables have been ridicules and utter failures. Humane famously raised $240 million to create an AI-focused PIN, which requires a constant internet connection and produces excessive heat to perform the most basic functions. Heuman eventually disintegrated and sold most of its assets to HP. Other devices, like the Ploud AI Pin, are recording-only devices that rely on apps and cloud-based AI for transcription.
Then there was Friend, another VC-backed startup that wanted to put an AI companion around your neck. Its million-dollar New York City ad campaign ran afoul of skeptical graffiti artists, so the company eventually moved away from AI hardware to a more chatbot website interface.
Kehrli said Qualcomm is not envisioning a single use case for this AI-ready wearables chip. The next device may come in a form factor that no one would have thought of. He said, “What may make sense to you may not make sense to me.” In the end, we may find ourselves overloaded with tools with very specific use cases. Some companies are not getting that memo.
We still don’t know what the fuck OpenAI and famous designer Jony Ive are doing. However, recent leaks from The Information suggest that it could be similar to a smart speaker with built-in cameras to help process information. Similarly, Bloomberg claims that Apple is working on its own AI pendant that is equivalent to a human AI pin, with an AI-enhanced Siri built in. It is difficult to assess technology from description alone. These tools aren’t going to spark instant happiness, whether in the Marie Kondo sense or in the gadget hobbyist sense. Not having a clear use case from the start makes it much less likely that regular users will be willing to stick a camera around their neck.
<a href