The Uber IPO filing (and Lyft just before it) made me think about all the speculation that they’re waiting for self-driving cars to come along and save their bacon.

Which made me think a bit about what might change when self-driving cars become practical and widespread. So let me put on my speculative futurologist hat for a moment and do my best Isaac Asimov imitation (and if you’re wondering why, try and read this absolutely wonderful story about self-drive cars, which he named automatobiles).

The thing is, like most big changes in technology, a self-driving car isn’t just going to substitute the driving and navigation functions of a human driver. Effectively a self-driving car is a form of robot. It will have a fair bit of processing power, high speed access to live information (not just navigation and maps) and the ability to process commands. The moment you think of it as an intelligent mobile entity rather than just a driverless car you realize it may change how we look at transportation.

For instance, if a self-driving car – let’s call it an automat in honor of Asimov – can accept instructions, plan a route and be at various points at specific times, keeping track of vacant seats and so on, why should we each own one? Simpler to hail one and have it show up at my door or go to a designated pickup location. If I want to go somewhere quickly I can specify that I want the entire car to myself and am willing to pay a premium for door to door service, urgent service and exclusivity. If not, I can book a seat on a scheduled service like a bus and walk a bit to a standard waiting point. There are all kinds of permutations and combinations possible to trade-off cost, convenience and exclusivity.

All of which sounds a lot like Uber, so people might think that’s when Uber comes into it’s own.

However, think about this. Why would the rest of the transportation industry wait around and keep buying regular cars, leaving automats for Uber and Lyft to buy? Taxi companies, car rental companies, bus companies can all shift to automats. If less people buy personal cars maybe even the car makers and dealers will shift to this kind of car rental model instead of selling them. Owning a fleet of vehicles takes a very different business model from that of the ride-hail app companies, but it is pretty much the same model that all the conventional transport companies have had all along.

If Uber and Lyft were to continue doing what they do now, it would hinge upon using excess capacity from private cars and selling that capacity at slightly above the variable cost so that owners are incentivized to make their cars available. So, if people continued to buy cars in an automat era and decided after they’d been dropped to work that the car could be made available for public use, then potentially a ride-hail app ecosystem could continue. Potentially it could be priced lower than regular cabs and transport services if owners are willing to recover just slightly more than the variable cost. However if the ride-hail companies owned their own fleet, their economics are no different than the other players in the field so they won’t be able to sustain a different pricing model.

The fundamental problem is that ride-hail companies haven’t changed the economics of transportation. Automats will, but they’ll change it for everyone so that won’t confer any advantage to the ride-hail companies.

Here’s the important thing to remember about Uber (or Lyft, Didi, Grab… you name a company) – at the end of the day, it’s just an app. Not unique, not hard to replicate and certainly not worth 120 billion dollars.