Yes: 50 mio robo-vehicles/year is a staggeringly wild assumption. Granted.50M will never happen, unless you think Tesla achieves well over a 50% worldwide market share. At some point (even if it doesn't happen in the next couple of years), robotaxis will arrive and car ownership will level off and gradually decline.
Piper Sandler believes global auto sales will peak ~2030 and decline from there. They are probably accurate within 1-2 years.
I'll eat my hat if Tesla builds more than 10-11 total GFs.
But ... we don't know how the transportation market will look like in 20-30 years.
There is a path for Tesla to get robotaxi cost down to 5 cent/mile. Lets say 7 cent/mile price. With some extrapolation of growth globally, and more and more people joining the (perhaps lower) middle class globally there might be a huge market for 7 cent/mile transport.
If you include vans and mini-buses you can perhaps get the price per customer below 1 cent/mile. At 1 cent/mile almost all people can afford some kind of robo-transport in 20-30 years. So around 9 billion customers. (depending on population growth)