2022 BRN Discussion, page-2871

Stand-alone AR/VR devices are very resource-intensive (VR more than AR), as they are a mobile platform with heavy GPU (graphics processing unit) and moderate CPU requirements.

SoC (system-on-a-chip) processors like Apple's M1 and Qualcomm's Snapdragon XR2 are both ARM-based CPUs and each SoC also has a GPU, their specs are different. The Snapdragon XR2 is currently used in the Oculus Quest 2 VR hardware.

I would venture to guess the majority of the heating and power consumption are drawn by the GPU on the chip itself. While I don't think that Brainchip would alleviate those types of requirements for AR/VR processing, I do believe there is one application it will be able to solve swimmingly.

While more so used in AR applications than VR (although the Quest devices support it), are both the camera tracking in the user's defined play space as well as tracking the user's hands as if they were controller input. All of this tracking is done by the images captured by several cameras mounted on the device.

I think this is where Magik Eye's technology (https://www.magik-eye.com/downloads/news/20200817/PR%20BrainChip%20MagikEye%20FINAL%2014%20August%202020%20Manny%20Edits%20per%20SL.pdf) is going to be valuable for playspace mapping, hand tracking, and gesture recognition, by leveraging Akida to reduce the required power needed for this processing.

My opinion is that this may, in turn, reduce demand on the GPU and alleviate heat generation and power consumption that engineers are trying to solve while building the next generation of AR/VR devices.