As I see it, the only absolute upper limit for VR is the resolution and frame rate of the VR headset displays, as well as the quality of the optical stack. Rendering of the graphics can be done by anything from a single high-end GPU in a PC up to a beefy server in the cloud—although in that case, of course, network latency and video compression will impact the experience.
unless it’s a niche thing (simulation, high end gaming), the future is undoubtedly free headsets like the Quest 3 and Apple Vision- for convenience in handling and setup, conformity and freedom of use.
VR needs to at least match the bleeding edge in graphics, which is desktop GPUs. Those desktop GPUs have large wattage and cooling demands. Therefore wireless VR headsets must use the mobile or laptop versions of those GPUs, which perform worse than the desktop equivalent. On top of this VR needs to draw the same scene twice (one for each eyeball). I cannot see how VR will ever overcome this with current hardware. They need to be able to put a desktop GPU in the wireless headset without needing to power it from AC, and thats just for starters because didnt Carmack say we need something like 240hz and 4k+ (for each eyeball) for indiscernible photo realism? Id say were maybe 10-20 years away from just having the GPU power to do this, let alone making software for it
The work to make or even use a proper VR app or game is so so much more than the flat equivalent and there are only some added utility for spatial input. Tech can certainly improve some of that...
But a VR video call is solved... You can do it mostly out of the box with an AVP but who is going to buy a $3k device for yourself and anyone you want to call and then have a couple calls and never use it because its not worth the hassle to strap it on.