The limitation on ownership is cost. Autonomy increases the cost of individual vehicles, but offers improved and cheaper alternatives to ownership. Unless autonomous systems end up being very cheap, they should be expected to reduce private ownership, with the most obvious effect being reduction of the number of multi-vehicle households.
How cheap is cheap?
tesla S generation MobileyeQ3 is an old 40nm 42mm2 chip. Yet for autonomous drive applications it outperforms the Nvidia Tegra 20nm 126mm2 chip. Roughly thats a foundry cost reduction of about 12x cheaper for the old mobileye solution compared to the old Nvidia solution.
Point is not a pissing match between mobileye and Nvidia, but that GPU based systems are significantly more expensive than dedicated systems. Tesla model 3 generation sensor suite seems remarkably well suited to MobileyeQ4, (a 3.5watt ASIC) there is no mobileye inside that anymore, but it is reasonable to expect the cost of autonomous drive, compared to nannied up non autonomous drive to be comparable to the cost difference between a stick shift transmission and an automatic transmission. Which implies that in USA greater than 95% of cars will gain autonomous capability. that will significantly increase vehicle ownership. I would guess that 95% of females involved in car purchasing in USA would really value the convenience of facebook use while driving. That means ubiquitous autonomous drive capable cars for USA.