Looking at it superficially, most recent big USA cities are built on a grid pattern which should be possible to program for and as long as a taxi was restricted to this clearly defined area then it could be made to work.
Support from the local authorities to keep mapping and temporary road restrictions up to date, Google Maps style cameras to continually monitor local conditions and update a central database, and probably a net of data transmitters on local street furniture (5G?) to maintain real time awareness of traffic conditions could make this viable.
Reduce the vehicle footprint to match the tiny cars from Japan and thus reduce both road occupancy and increase ease of finding curb/kerb space and you could have viable autonomous transport.
Outside that very limited environment, especially in old European towns and villages, you have a whole different level of problem.
Unsuitable infrastructure and no money to update it.
Almost certainly not enough revenue to justify the service provider funding upgrades to the infrastructure.
So the second reasonable target is enhanced long journey cruise control on major routes. Again a reasonably well defined target with limited variables.
Again there is the requirement for accurate data including traffic and roadworks problems.
The major routes would also benefit from autonomous only lanes to prioritise organised traffic flow.
There is, of course, the range problem for electric vehicles.
Again for it to work efficiently there are infrastructure costs to be funded.
In all of this I don't see Tesla having a massive advantage over conventional vehicles with enhanced cruise control except in city centres where the infrastructure could be effectively redeveloped to allow only small footprint electric autonomous vehicles, bicycles (including electric) and pedestrians. However Tesla doesn't seem to be developing a suitable vehicle at the moment.
As an additional point, any redevelopment costs need to be weighed against redevelopment to just restrict city centres to public transport and specially designed service vehicles. Which then raises potential issues for anyone living in the centre, of course, which in turn requires more infrastructure for storage/transfer points to more conventional transport.
So I don't see a Tesla based solution coming to a town near me any time soon.
There will be enough problems trying to recycle all the petrol/diesel vehicles off the roads to reduce emissions without more massive infrastructure development in a continuing climate of austerity coupled with cheap labour, which incidentally is why providers such as Uber are making specially designed vehicles such as the London Taxi less relevant.