You certainly can do somewhat useful AR on a phone. See e.g. some of the ML photo and sound identification tools (e.g. Merlin, Google Lens) that have very quietly gotten quite useful.

But AR that is really a killer app almost certainly needs a fashionable, light, easy to use headset and that's a hardware problem to a significant degree that probably doesn't have a quick fix. Yes, you can do something that's effectively a prototype that's more like a "headset" but that's almost certainly not a mass market device that most people will walk around wearing.

(There's also the social glasshole issue but my guess is that goes away with a useful enough device even if some people hate the idea.)

I think market will be there it’s just like 5 years from now and most people seem to know this such as Roblox and Epic (creative 2.0). It will be speed up by generative AI

> What AR/VR needs is a mass market killer app

I really think we're just waiting for the resolution to increase and it to become thinner and lighter and better at hand tracking.

The Quest 2 is already a better experience watching 1080p surround-sound content than most people's TV's -- if you're watching by yourself. But streamers haven't quite caught onto that yet.

But once the screen becomes sharp enough to do real work and hand tracking improves dramatically, I think we're going to start seeing some dramatic shifts.