Google Set To Acquire Ultra-Low-Power Eye Tracking Startup For Headsets & Glasses
Google is set to acquire AdHawk, a startup that makes ultra-low-power eye tracking sensors for headsets and glasses, Bloomberg reports.


Google is in "final talks" to acquire AdHawk, a startup that makes ultra-low-power eye tracking tech for headsets and glasses, Bloomberg reports.
According to the report, the agreement to acquire is set to be completed this week, and is valued at $100 million, with an additional $15 million if certain performance targets are met. If it goes through, AdHawk employees will join Google's Android XR team.
Meta considered purchasing AdHawk in 2022, the report claims, but decided not to go through with the acquisition.
Google's Android XR platform is set to debut in Samsung's standalone headset later this year. Like Apple Vision Pro and PlayStation VR2, Samsung's headset features internal cameras pointed at your eyes, combined with infrared illuminators that help them see. The tracking is done using computer vision algorithms that analyze each frame and output the size and position of your pupil.
But sampling and processing multiple cameras each frame is a relatively expensive process, from a power, compute, and component cost perspective. Instead, AdHawk's eye tracking system uses a MEMS micromirror (a tiny, electrically actuated mirror) to rapidly sweep a beam of infrared light over your eyes, and a series of simple and inexpensive photosensors measure the intensity of the reflection of this infrared light off your eye.
It's a somewhat similar approach to Inseye Lumi, the low-power Quest 2 & 3 eye tracking addon that was recently "paused" in favor of bringing the technology to smart glasses.
The result is eye tracking at a much higher update rate than camera-based systems, using a fraction of the power.
AdHawk says its eye tracking currently runs at 250Hz tethered or 500Hz untethered, with a latency of 4 milliseconds. But as with Inseye Lumi, the tradeoff is accuracy. AdHawk claims 1 degree of accuracy, which is better than Inseye's 2 degrees. The accuracy of camera-based eye tracking, though, is measured in tenths of a degree.
This means that interface elements on devices with this kind of eye tracking, at its current quality at least, might need to be larger and more simplified than what we see in visionOS and Android XR today. This could make it suitable for future lower-cost Android XR headsets and, one day, eventual AR glasses. Reporting from October suggests true AR glasses from Google are very far off, however, with the company focused on supporting Samsung's simpler smart glasses for now.