I think the market will not price in Tesla as an AI play until Full Self Driving is viable, which, honestly, it is not right now.
The issue is that FSD is a long tail problem. It is trivially easy to get ~80% of situations. Decently hard to get 99.9% of solutions, but those .1% are all really hard edge cases. I suggest paying attention to youtube channels like dirty tesla to get a sense of where the tech is. It is certainly improving, but the jury is out on if it will be true true self driving.
I think if there is hope for FSD being something monetizable, it will first be geofenced. There already are AI cars being used in Vegas and SF. So it does seem possible for Tesla to allow just those locations to be used for robotaxis.
One thing I think people aren't talking about is tesla's investment into Dojo. I do question, if that is a good idea, since nVidia is making some intense strides into development hardware. This decoupling from nVidia could be why Tesla is not currently tracking, as they are redoing the research and product dev nVidia is currently doing.
I've been following Gary Black, and he puts TSLA at a 320 price target in 6-12 months. I Think that is reasonable. That is dependant on the effect of advertisement, the cybertruck launch, Twitter positive cash flow, and the roll out of the compact model. If we can get a compact model, with decamillion deliveries, I would personally value that at like 500$/share. I think that happens within 4 years.
If we get true FSD, that is monetizable, then you are looking at 4 digit share price numbers. And more if it gets implemented into optimus. But, of course, that is very speculative.