>The question is when would self-driving systems be proven and accessible. I think by 2040.

I think sometime between 2028 and 2040. Depending on how hard snow and rain are to solve and how challenging scaling is (my guess is scaling will simply be driven by rate of manufacturing which means 3 years to fill market penetration after full self driving is solved).

I honestly think it will be solved faster than some believe due to exponential rates of data collection for autonomous fleets in combination with better computing capabilities and better AI architecture. I still place it in the late 2020s at the earliest though for mass market and well later than that if we're talking reliable snow driving, etc...