One Big Reason Robotaxis Won’t Replace Owning Cars

Support CleanTechnica’s work through a Substack subscription or on Stripe.

I’ve been covering the robotaxi debate for a decade. I still remember a discussion Chris DeMorro and I had 12 years ago about which would end up being a bigger deal — “Tesla D” or Tesla Autopilot — and a subsequent poll. I used to be much more bullish about robotaxis, but now no longer see them as a very disruptive idea, for a few reasons. However, a comment today from a reader really helped me zero in on one key point.

First, though, let me emphasize that I do think robotaxis will take over the taxi market. Eventually, they will be good enough and cheap enough that paying a human to drive a car instead of using robotaxi software just won’t make sense. In fact, I don’t think we’re far off from this.

Additionally, I think that potentially lower costs from robotaxis and not having the limitations of human drivers will lead to robotaxis expanding this portion of the market a little bit. Robotaxis will end up eating into some portion of normal owner-driven transportation as well.

However, that is not what a lot of the robotaxi hype centers around. A lot of the robotaxi hype assumes that robotaxis will make much of the car ownership market go away, that most people (or at least many of them) will stop buying cars and start using robotaxis instead. There are a few reasons I just don’t see that happening. Let’s dive into those now.

Cost Per Mile (… or Human Mile)

The comment a reader, Steve Shaw, made earlier today is as follows: “You are suggesting that a robotaxi will be cheaper per mile than owning a similar car?” Steve didn’t delve into the thoughts that question triggered in my mind, though he did make some other good points. Thinking about answering that question gave me the following light bulb thoughts.

Consumer cars are already getting the same self-driving capabilities as robotaxis. You will be able to have your own car drive you wherever you want, and more safely than you would do it. BYD is already saying it will take on crash liability whenever its self-driving tech is being used. Tesla “Full Self Driving” will easily drive you on most trips now, even if it’s still not able to be unsupervised due to edge cases and bad weather concerns. The tech needs to improve, competition needs to come to market in the US, and costs need to come down, but capabilities have clearly improved.

Now, people easily fall into the trap of thinking, “But robotaxis will be in use much more, so will have much lower $/mile cost.” But there’s at least one serious flaw in this thinking. Much of the lifespan of a car and its parts is about miles driven, not how quickly miles are driven. Yes, some things degrade or get worn out due to time factors, but most of a car degrades and gets worn out from use. That’s even more true of a vehicle being used by a bunch of customers rather than one owner who cares a lot about the condition and preservation of their high-cost asset. Perhaps a robotaxi will drive 100,000 miles a year compared to you driving 10,000 miles a year. But then that car might be out of service after two years whereas your car could last 20 (going through different owners, most likely). Tires are still going to need to be replaced, as well as other parts that crumble or get beat up from wear and tear. Also, if the point of the vehicle is customer service as a robotaxi, you need the car to remain in a certain minimal condition or else people won’t be happy with the service. You might be fine using or passing on your car to a teenage son or daughter after it’s got 100,000 miles on it and is somewhat “worn out,” but is a car in that condition ideal for robotaxi service?

Aside from the seats, doors, trunk, or other parts of the car that might get visibly damaged and worn out from human use, there are all of the components of the drivetrain, HVAC, computers, etc. And this brings us back to that question from Steve about per mile cost. Let’s modify that slightly: let’s consider per human mile cost.

Think about it: how can a robotaxi be cheaper than a personally owned car with the same self-driving capability? The former is going to drive much more to perform the same number of trips, or to drive people the same number of miles/kilometers. Whereas a personal car only takes you where you want to go, only takes you from one destination to another directly, a taxi or robotaxi has to get to you from some starting point that is not your driveway, take you to your destination, and then drive somewhere else to wait for another passenger to request a ride. That’s extra miles being tagged on before and after your trip. That means robotaxis will rack up a lot more miles (and wear and tear) to perform the same driving tasks.

As I stated yesterday, for what would be an A–B–C–D sequence of trips in a normal car, this is what a robotaxi would do:

• • Drive from point Z to point A (to pick you up),
  • Then from point A to point B (your first destination),
  • Then from point B to point Y,
  • Then from point X back to point B (the car won’t park where it drops you off and then provide its next trip starting from that same exact point),
  • Then from point B to point C (your second destination),
  • Then from point C to point W,
  • Then from point V to point D (your third destination),
  • Then from point D to point U.

That’s a lot of extra driving compared to a personally owned car with the same self-driving capabilities. Cost per mile of human transport will go up significantly.

Some might contend that robotaxis will have more flexibility than humans and can just wait at each destination for another direct transport route. But that ignores basic facts of human transportation patterns, something that carsharing, bikesharing, and taxi service have had to contend with in the real world. “There’s always going to be disparities between where people are and where they want to go. People early in the morning will be going to the airport, but planes won’t be landing to bring them elsewhere. Same thing with people going into a city core in the morning, or leaving it late afternoon/early evening,” JoeNCA clearly summarizes.

So, that is the one big reason robotaxis won’t replace owning cars. The cost per mile of human transport will be higher. But there are other reasons as well. I won’t expound on them like this since I’ve already done so before, but here’s a list of some of them:

• People own cars for status reasons, and they won’t stop buying their BMWs, Porsches, Audis, Volkswagens, Fords, Cadillacs, or even Hyundais, Kias, and Chevrolets just because robotaxis exist.
• People treat their cars as extensions of their homes, with all kinds of personal stuff in them, as mobile storage spaces, etc.
• People want immediate access to a car — they don’t want to have to wait 5 to 15 to 30 minutes for a car every time they want to go someplace. Some people are more flexible in this regard, but even retirees may fear not having a vehicle when they need one immediately if they rely on taxis instead of a personal vehicle. (Florida doesn’t require retaking driving tests to keep your license specifically to entice retirees to the state, including millions of people who can afford to take taxis everywhere but want to retain independence and status with their own personal car.)
• People like different types of vehicles and are not going to be happy all using a few standard robotaxi vehicles.

Consider this final point: Most people buy cars well beyond what they need in terms of size and capability just for the odd chance they need more space or capability. Logically, they could buy vehicles that fit their normal needs and just rent or hire a ride on the odd occasion they need something more. That could save them a ton of money. But people choose instead to overbuy and pay more. As many intro courses on economics will teach you, humans are not “rational actors.” Economics and various analyses will try to assume humans are completely rational actors, but the real world is different. And even rationally speaking, it’s very hard to make the case vast quantities of people will switch from personal vehicle ownership to robotaxi use. We shall see.

That said … there are some potential cases and applications that could lead to market expansion. I’ll come back to those in another article.