.... what?
Tesla has no external facing data center service, and it multiple steps from even being capable of offering one.
Meanwhile XAI is
actually buying server space from one of the many companies
actually offering it.
How is billions of miles of car driving data applicable for solving general robotics AI for a thing that isn't a car?
All inputs on things like how humans behave around crosswalks, how other cars behave around other cars/objects/people, etc wouldn't be relevant to a walking robot.... and outputs (walking, hand manipulations, etc) would all be entirely different outputs/controls than what the car does- and with vastly more needed precision... if your car parks 25 mm further or nearer an object it's not an issue-- if your robot hand misses a fine motor control thing by 25mm it's a complete failure.
I mean there'd be SOME overlap in things like generally recognizing objects exist and they're X distance away, but that's not an especially hard AI issue and tons of folks have solutions that do that without Teslas data....And Optimus cameras/speed/etc are all different locations than the car...
So I don't see the 'billions of miles of driving data' thing being
nearly the massive advantage for optimus development that it is for FSD development.
They share the same FSD computer... the cameras are different in placement (and I'd expect # in the final product)- and they for sure wouldn't share ALL the same SW as that'd make no sense.
Keep in mind they still don't have
basic autopilot working on Cybertruck 6 months after first delivery despite being far more similar to other FSD-enabled cars than Optimus is... so it's not like they just toss the FSD stack onto Optimus and everything works right...