Self-driving is expected to bring down ridesharing costs. As costs go down, ridership goes up. As ridership goes up, traffic goes up. Currently I can get around by bicycle, rideshare, bus, subway, trolley, and rail. With the exception of the subway, all of those are slowed down by increased vehicle traffic. The tipping point you're talking about may be city streets becoming gridlocked.
Then there's the bit rot problem. Most of this self driving tech is software. Over time, software becomes old, eventually is abandoned, and has to be rewritten. Assuming the cityscape was changed by 2030 (this is impossible, but just for argument's sake) we could have broken legacy tech by 2050. But we will either be stuck with it, or middle-aged people will have to start driving 30+ year old cars as taxis, as no young people will have gotten a drivers license, and all the self driving cars won't have steering wheels.
A vehicle miles added tax would, in smart policy districts, allow the curbing of excessive car use, especially for a company like Uber or Lyft. There's also congestion tax, which many cities are already implementing. Some cities just straight up banned non-electric past a certain year, so we're already moving slowly to deal with traffic. I think we'll have the tools we need by the time it's a huge problem. In Boston, the traffic is pretty awful, there are just many cars shoved into a small area with infrastructure based on cow paths from original settlements.
Then there's the bit rot problem. Most of this self driving tech is software. Over time, software becomes old, eventually is abandoned, and has to be rewritten. Assuming the cityscape was changed by 2030 (this is impossible, but just for argument's sake) we could have broken legacy tech by 2050. But we will either be stuck with it, or middle-aged people will have to start driving 30+ year old cars as taxis, as no young people will have gotten a drivers license, and all the self driving cars won't have steering wheels.