You said "And no Tesla has EVER driven without a driver with Tesla assuming liability for crashes." and I gave a counterexample. I figured you wouldn't count the Austin robotaxi rollout, since there is a safety monitor in those vehicles, just not always in the driver's seat. I also assumed you meant "on public roads", since Teslas already drive themselves from factories to staging/shipping areas, but that's on private roads. And their cybercab demo last year lacked safety drivers, but again that was on private roads.
Are you willing to stake money on your prediction? I'll bet you any amount up to $1k that Tesla will have self-driving vehicles on public roads without a safety monitor in the car by the end of 2026. This could be in the form of their robotaxi service or FSD unsupervised (since many Teslas are insured by Tesla).
"Are you predicting that Tesla will never get rid of their safety drivers in their Robotaxis?"
That is exactly what I'm saying. Because vision only non-redundant technology stack is never going to be reliable enough.
"Because as soon as that happens, won't your prediction be incorrect? "
yes, that is how predictions work.