This is different. The problem of providing offshore satcom with LEOsats is a quantified engineering problem. IOW, the industry knows quite well what technologies will solve the problem and Starlink has those technologies in place. The only unknown is how fast, accurate, and reliable the laser-based intersat comms will be.
None of that is true for the self-driving car problem. That problem still contains a multitude of unknowns, including unknown unknowns.
Cybertruck is yet another kind of problem. I don't know what the issue with that is but I'd guess it's about manufacturing capacity.