3 decades might be how long it takes until this is running locally in your glasses, although we may hit some hard limit in silicon before we get there at all.
But AI models are already running on tablets (not necessarily on Nvidia hardware) and I expect some phone to ship with them within a year (maybe as a stunt, I guess it would be a few years more before this is practical).
But AI models are already running on tablets (not necessarily on Nvidia hardware) and I expect some phone to ship with them within a year (maybe as a stunt, I guess it would be a few years more before this is practical).