For as long as mankind has gathered en masse, access to mass transit has depended on proximity to a station or a port, and bringing transportation to an area has meant massive capital expenditures to expand these centralized hubs. If the means of transportation is rail, then expansion has additional capital costs for laying tracks, and if the rail is a subway in an urban area, costs go through the roof ($2.23Bn per mile for the 2nd Avenue Subway, according to CityLab).
Autonomous vehicles change the game. Today, around 80% of the cost of an Uber or Lyft ride goes to the driver. Take out that cost, and fares could be around a fifth of what they are now. The cost might go up a bit to account for what will initially be a higher cost for the newer technology, and for the need to cover expenses that are currently paid for by the driver, such as fuel, insurance, maintenance, etc., but with the size and scale of the companies operating in this space I expect robust fleet management to create efficiencies and drive prices back down to around the same level. Driverless vehicles will eliminate the cost of driver acquisition and qualification as well, bringing costs down even further. So let’s say the end result is that ride costs around 25% of what they are today. Based on data reported by SherpaShare, the average Uber and Lyft trip cost around $13 in the first half of 2015. 25% of that is $3.25, which is in line with public transportation — and that’s for a ride in which you take up the entire capacity of the vehicle from point A to point B. Currently, if you share the ride with one other person by using UberPool or a Lyft Line, you get 25% off of the base fare. That brings the current average cost of a ride down to $9.75, or $2.43 in a driverless world . If you use a service like Via, which uses SUVs to fit up to 6 passengers at a time and groups riders by pickup and drop off area, that cost goes down to under a dollar each.
We suddenly have the ability for mass transit to diffuse and blanket an entire city, picking you up at your door or at your corner . Some local density will still be required to make the service economical, but there won’t be any capital costs to spreading to a new area as the population shifts there (or as political demand requires it). The implications of this are profound.
First, you will have no such thing as a neighborhood unserved by transit, just like today you have no such thing as a neighborhood unserved by electricity. Pickup times will be higher in areas that have less density or less affluence, but everybody in an urban area will be able to get service.
This in turn will affect real estate development and property values, as the spikes in value of being near public transit will be smoothed out (there will still be many other factors remaining that influence property values of course). When you search for an apartment in New York, one of your top criteria is what subway is it near, and how far is it from the stop — ie how far are you going to have to walk every morning, or in the cold of winter, or when it rains? Geography — especially traffic, walking proximity to neighborhood amenities, etc. — will always be a consideration, but if you don’t have to worry about the distance to the subway, the universe of possible housing options opens by an order of magnitude. The same is true on the other end of the commute — the universe of options for where to locate your office multiplies.
Second, transit will scale like a variable cost, increasing gradually with demand, rather than in a binary fashion when the decision is made to build new stations. As long as roads exist, the density of transit stations won’t have to be planned in advance; supply will perfectly match demand. It will be possible to hold large-scale cultural or sporting events in different parts of the city, without having to worry about whether there is public transportation or parking at the site .
For at least the first phase of the shift to autonomous vehicles, in which the roads are still predominantly being used by human-driven cars, autonomous point-to-point mass transit will probably not be the sole means of transportation for longer trips or trips from the suburbs into the urban core. It will most likely be the “last mile” which connects the rider to more traditional station-to-station transit. The important part is that this “last mile” can in fact extend for much larger distances, so for the first time development that expands the boundaries of urban development will be able to proceed independently of plans for transit infrastructure, rather than being dependent on it.
The expansion of mass transit in a zero fixed-cost world will be so far removed from what it is today that it won’t even be recognizable. Except for the corridors which are so dense that they can only be served by traditional, fixed-infrastructure subways carrying thousands of people per train, routes will dynamically adjust according to rider demand. At scale these will be roughly constant, but will dynamically adjust to things like the weather, holidays, major events, road closures, traffic and construction, etc, which current mass transit is notoriously challenged at handling. They will pop up when density in a given area reaches a tipping point, and they will go away when no longer needed. I am very excited for this world to arrive.
 The one significant outlier is NYC, where the average fare is $29 (or $7.25 at 25%). However a good part of this is because of additional regulatory burdens placed by the city’s Taxi and Limousine Commission, such as the requirement to have additional insurance, which may not apply in a driverless world.
 Nevermind the other benefits of this form of mass transit: a clean, comfortable vehicle that sends you an alert before it arrives and handles payments automatically.
 It’s possible that if you had too many rides getting dropped off at the same location, and not enough pick ups in the surrounding area, that the overall system could become unbalanced, but there are various ways to solve this.