How good a business is running a robotaxi?


Most of the major players want to run a robotaxi business -- Uber style ride service using robocars. Yet some have started to wonder if this is the best business model, or if it's even a good one, while companies invest billions in it.

In this new article on I investigate some of these questions and why the players are investing these sum, and what sort of profits they might make.

Read about it at Some Say Self-Driving Robotaxi Isn’t A Business; Billions Are Being Bet That It Is


Shuttling people seems like a decent market, but the TAM increases greatly when rides are 80-90% cheaper. The car can run errands for me, or delivery services for businesses now become feasible. I would think that opens up all sorts of commercial opportunities.

Interested in your perspective of why UBER/Lyft did not impact the rental car business as much as you expect Robo-taxis to do.

I have not seen a direct study? Have you? Of course both were crushed by the pandemic. I would be very surprised if there aren't a number of people who previously would have rented car who used Uber instead. I have done so, but my own experience is just one data point. I also now use Uber exclusively instead of airport parking.

Brad - I was referring to pre-pandemic stats. if you look at 2010-2019 rental car revenues ( for example) they are flattish to up. And this did not happen by increasing prices. The revenue/car/day was also flattish to up. And during the same period, with the exception of Seattle, the personal car ownership in 20 largest cities in US did NOT decline. In fact it went up.

Intuitively one could make a case that the ride-hailing services should have impacted this (you gave some of your personal example usage, I do the same). But they didn't. Your article makes the case that self-driving vehicles will do so. Very interested in your perspective of why ride-hailing did not have the impact that you are predicting for self-driving? And what is the different with self-driving vehicles?

What you would want to do is look at numbers for the rental car industry as a function of the amount of air travel going on. Generally, car rental should track air travel and a few other things. You could look at revenue, or number of rentals which the public companies probably report. Ideally broken down by airport. You want to remove non-airport rental because people probably aren't going to replace a non-airport rental with an Uber trip. The bread and butter of car rental, though, is people who arrive on a business trip, rent a car at the airport, and drive it a small amount to their hotel and meetings and then back to the airport. For simple trips, Uber can be a better choice. Of course, in spread out cities, Uber is expensive. In more dense cities it's the other way (nobody would rent a car for a business trip to Manhattan.)

rental market vs uber
brad - your comments are sensible. But they don't address the fundamental question: Why did ride hailing impact the rental and the personal car ownership? What was the reason? And why would robo-taxis will have a different answer as you suggest as the fundamental opportunity of robo-taxis?

My understanding is that ride hailing has affected car ownership in denser urban areas, places where more people already don't own cars. In places where car ownership is considered essential and is almost universal, I doubt it did much.

Where it had success was in areas where "Should I own a car?" was already a reasonable question, and it moved some people over the line. In particular people who put good value on getting their own time back, or who hate looking for and paying for parking. But it was more of a service that would be "better, not cheaper."

For my airport trips, with airport parking at $15/day, and the drive in your own car also costing 50 cents/mile all-in, it didn't take too many days in the parking lot to make it a no-brainer to use Uber to the airport. But if you started valuing your time on the drive at a decent rate, like the rate you earn money, it becomes very easy to see Uber as the cheaper option to the airport. Add on the fact it takes you to the gate, while driving takes you to a parking lot to get a shuttle, and that win was also easier.

If you owned a robocar, you wouldn’t have to pay to park at the airport.

If ridehail has had impact on car ownership in cities, it has been hard to see. I am sure you know Bruce Schaller's work on this:,vehicle%20ownership%20rates.

Maybe the impact just hasn't shown up yet, but ridehail is no longer young per se.

At some level perhaps we are all missing the optionality or some other value of ownership. Even before ridehail, I remember visiting friends in Manhattan, and watching people a couple of times a week race to shift their parked cars to the other side of the street. People I knew would own a car they would park all week long, and then maybe drive to the countryside on the weekend. Even then (again, pre-ridehail) I'd point out that they could ditch their car, walk once a week 3 blocks to the Hertz garage, and rent a car for the weekend for much less than the ownership cost. They'd always agree with the numbers and then proceed to never do it. It amazes me that even in cities like NYC and London, what with ridehail and limos and taxis and public transport personal vehicle ownership rates are still above 50% of households. I must be missing something beyond the pure economics.

I never expected Uber, as better taxi, to change car ownership a great deal, though I did expect it to take it down a bit in areas that already have "wealthy non car owners." I mean Manhattan's low rate of car ownership relies heavily on its extensive taxi fleet. To make a dent you do need people to be already debating between car ownership and non-ownership. Then, the existence of good non-car alternatives, including Uber, carshare etc. should move the needle.

I did not expect traffic to go down of course, other than through services like UberPool. UberPool has not been as successful as many of us hoped, and the pandemic shut it down entirely. I tried to use it when I could and quickly learned it was not very good. The detours and time delays were much too large for the savings you got.

I do believe I would use an Uberpool style service if it promised minimal delay. I believe that is possible but requires multi-leg trips with instant transfers. Ie. something like the chart I have at Future of transit where you find the common sections of the route for multiple travelers at the same time, and give them solo rides to the convergence point and from the divergence point, and shared ride along the common path. This is only practical with robotaxi though, as it's not efficient to have human drivers do short 1/2 mile trips unless you can get them to do a lot of them so they keep in constant work and don't deadhead too much. Robots don't mind waiting and even deadheading.

Even Hertz was too slow at non-airport locations. Zipcar and other carshare were much better. You really want to make it frictionless. But even that walk is friction, particularly if you want to haul gear or multiple people.

Lyft/Sixt have an interesting plan underway I would like to see how it does. Right now rental cars are very expensive though.

A friend is trying to build a startup which delivers rental cars to you. That's a good idea, but will people pay the cost, and the advance notice needed. It's sold as a subscription not a daily rental.

The reality is the private car is really good. People like it. Its main downsides are cost (but it's cheaper than Uber,) finding and paying for parking in places with scarce parking, and time spent driving. Robocars fix #3 (taxi or privately owned) and #2. Cost includes not just the cost of parking, but the cost of a garage.

A common misconception I have seen is people imagine that because cars are parked >95% of the time, that's wasteful and sharing will win. It is wasteful, but we can afford the convenience it brings in many cases. It is not as wasteful as people think, though. For cars that get heavy use, they wear out by the mile more than by the year, so they don't really "sit idle." If a car is a consumable with a life of 200,000 miles over 20 years, if you get your usage above 10K miles/year, you can treat it like a consumable rather than an asset.

Good points all. Very good point on the 95% idle meme. As a speaker at a conference I went to years ago said (I wish I could recall his name, to give him credit), "I only use my toothbrush 0.5% of the day, but I sure as hell am not renting it out to you the rest of the day!" (grin)

Damn, Brad, it sounds like you’re finally accepting that the robotaxis are not necessarily going to win over the personally owned robocar. Combined with your more recent fondness for a level 3 service, it looks like you’re coming around.

I’ve always thought you greatly overestimated the benefit of robotaxis compared to personally-owned robocars. The overhead is just too high to justify the cost savings for normal daily usage, at least in places where parking is not a premium.

We’ll have robotaxis, but we’ll also have personally owned robocars, and rental robocars. I think the rental robocar is perhaps the most promising. Rent a car by the day or by the hour and use it however you like during that time period. Not a great option for your daily commute, but it can be a good one for the road trip, where your needs are significantly different from your main commuter vehicle.

Of course, the commute itself is very gradually starting to go away. If that really takes off, maybe this calculation changes. If you only need to drive sporadically, and not 5 days a week at the same time as everyone else, a robotaxi model might make more sense.

Personally I find working at the office so much better than working at home. But there are generations younger than I am that are more likely to see this differently.

No, I think car ownership has many advantages and is very popular. See a comment above where I outline what they are.

Some of those advantages are stronger because parking is subsidized heavily in most places, and even in city cores it is still subsidized, just less. If people paid the true market cost of parking car ownership would be less attractive. Of course, for private robocars, parking will also be much cheaper. (Less true at home, because even if you don't provide a garage or driveway at your home, it probably costs just as much to do it nearby, while in the city a distance of a mile can mean a large difference in parking cost today.)

The Robotaxi offers you:

  • Some savings on parking, particularly for urban people. But even suburban people can reclaim garage space for other uses. Around here space is now $500 per square foot!
  • Dynamic vehicle on demand. Cheap vehicle for solo trips. Luxury when on a date. SUV or truck or minivan when you need it. Most people buy the SUV which they need 5% of the time but drive it 100% of the trips.
  • No time depreciation for people who drive less than 10K miles/year, which is a large number.
  • The vehicle is constantly upgraded. For not much more you can mostly ride in new vehicles.
  • Somebody else worries about everything -- maintenance, charging, repairs, cleaning etc. Many people will pay quite a premium for this.
  • Two vehicles at once when needed
  • Cost-effective one-way trips. While you can send your private car to the airport to pick up a guest, it must do the round trip, while a robotaxi has somebody else pay for the return leg.

You can get some of the benefits of robotaxi while still owning a car, and that will in fact be common for families who switch form 2 cars to 1 car + robotaxi. But there is a cost for this. And as it reduces your miles on the your owned car, it starts moving that into the time depreciation column.

There are many types of people. Some will maintain car ownership. Others say "I can't imagine why anybody would own a car in the robotaxi world." Most are somewhere in between.

I’m not sure where you get “parking is heavily subsidized in most places” from. Maybe in some cities the costs of parking is paid for with general tax funds, but for the most part the costs are borne by the property owners that benefit from the spots (directly or indirectly). It’s pretty close to free market except where government programs come in to intentionally subsidize it for low income individuals.

Most suburban people don’t use their garages for their cars anyway. And those that do benefit from the convenience that goes along with that. If you fill your garage with stuff, then you’ve gotta go outside in the weather to get to your car. If you use a robotaxi, then you can’t keep personal items in the car. (Garage helps if you’re worried about theft of those personal items.) Other than that, the savings is minimal, really. Even for city-dwellers (or city-workers), the cost of parking is basically just the cost of your car driving to a cheap parking spot. And robotaxi companies will have to pay for cheap parking spots too. The vast majority of the cars will need to be parked all at once overnight. So you don’t save much if any actual space. Just possibly relocate it to a cheaper area (which private robocars can also do).

Dynamic vehicle on demand is great. Buy the robocar that you need for 95% of your miles, and rent a robocar (or robotaxi) for the other 5% of the time. (With that said, for a lot of people, including many people with children and/or pets, I think they’d prefer the SUV 95% of the time and are willing to pay the cost of it.) Speaking of pets, I guess we’ll have pet-free robotaxis and pet-allowed robotaxis? That’s going to be a mess. You’ll have to accommodate people with service animals or emotional support animals, but you’ll also want to accommodate people with allergies. Any idea what Waymo is doing here? (Update: Looks like they currently allow only “service dogs and [certain] miniature horses.” No pets. No ESAs. No lizards or rabbits or fish or hermit crabs, though maybe they don’t enforce the rule in the case of some animals. Presumably they’d have to change this if they want their service to be usable as the only car for the masses. And I don’t see what they do about people allergic to dogs.)

Robotaxis might work for people who don’t travel by vehicle much. City dwellers and some on-campus students are two good candidates.

Constantly getting upgraded? I don’t know about that. Are traditional taxis constantly getting upgraded? These cars will last several years, if not longer, on average. Of course you can pay more for better quality. But if that’s what you like and can afford, you can also lease your private robocar. Also, keep in mind that this will be a significant added cost for people who are not interested in the latest bells and whistles.

You can pay someone to maintain, charge, repair, and clean your personal robocar, if that’s what you want. I really don’t see what advantage the robotaxi has here. The robotaxi just has the disadvantage that you can’t do it yourself, nor can you choose how often to do it, or who is going to do it for you. With robotaxis, you’re stuck with having to pay to clean up the car constantly (and worse, pay for messes that other people make), whereas with a personal robocar you can leave cheerios all over the back seat for free. Oh no. My kids spilled cheerios, again. And just as we’re arriving at the movie theatre, so I don’t have time to clean it up right now. Guess I’ll have to pay the $15 cleanup surcharge, since I can’t just leave the cheerios in the car until after the movie is over. But at least I don’t have to worry about cleaning!

Two vehicles at once when needed: See my third paragraph above.

Cost-effective one-way trips: See my third paragraph above. But moreover, I think you’re underestimating how much deadheading these robotaxis are going to wind up doing. The companies aren’t going to be able to match up traffic going from the airport to a residential neighborhood with traffic going from a residential neighborhood to the airport all that great. Traffic patterns go mostly in one direction at certain times of the day, and mostly in the other direction at other times. There’s going to be a lot of backtracking. There’s going to be a lot more backtracking than there would be if people just kept their cars parked where they happen to be. Yeah, airports are an exception. And airports are a good place where Uber and taxis already work. And rental cars. A lot of people won’t want their own personal robocar for the week while they’re out of town. Maybe they’re planning on being at a hotel the whole time. But a lot of people would like to have their own personal robocar for the week. And the cost of rental cars isn’t all that bad. More than a robotaxi if you’re driving less than 10,000/365 miles a day, maybe. But not that much more, and pretty convenient.

There are costs to private robocars. And there are costs to robotaxi services. I don’t think it’s at all a given that the latter will be cheaper, if the number of miles you drive is sufficient. Providing a professional service comes with a lot of overhead compared to private ownership. And cost isn’t the only factor anyway.

I think outside the city/college-campus, most will maintain car ownership. So not in-between.

Parking is heavily subsidized in almost all cities. There is the free government provided parking on most streets, and the metered parking at below market rates or under non-market rules -- those are the most obvious. But in almost all cities there are laws that require new construction to have a lot of parking. Home construction is required to have parking too. While it is not generally forbidden to charge for this parking, it is rare for stores and offices and residences to charge for it. This massively subsidizes people who need to park cars over those who walk/bike/transit/motorcycle -- though transit has plenty of subsidies too. Recently, due to more following of Shoup's "The high cost of free parking" and have gone the other way, either permitting buildings to not have parking, or even forbidding more than a certain amount of parking.

Actually, while my garage is indeed not used for parking (weather here makes parking in the driveway no burden) even here lots of people park in their garages, and in some places it's not very optional. In cities, many houses have garages but no driveway. Driveways primarily exist because laws in most R1 zones (most urban land in the USA) require minimum "setbacks" from the road. This has gotten ridiculous in this town. New developments here shrink the lot to minimum size allowed but they must have the setback so they have almost no backyard and a big front yard that the homeowner doesn't use. The buyer would much rather have a bigger backyard and less front yard.

Pets are an interesting question I have not directly addressed. Most robotaxis are going to have easy-to-clean interiors and not fine leather unless you pay more. But yes, some won't allow pets and some will. The mess a pet causes is sometimes hard to see though, which is an issue when trying to bill cleaning to the passenger, but cameras are now getting very high resolution. We would not clean a pet-allowed taxi after every ride, and probably just add a small cost on top of every pet ride unless the pet makes a big mess.

Traditional taxis are constantly upgraded. Before the pandemic, the average taxi in NYC was 3.3 years old. It has now increased to 4.5, so that's a trend not yet understood.

People have done lots of research on the deadheading with simulated models based on real trip data. Estimates typically run 5-15% of miles, including the one-way travel of rush hour.

Anyway, I have always said some people will want to own or rent, and others will like the taxi. Today, in most places there is only one choice -- own. And I guess there are some places like Hong Kong where there is almost only one affordable choice -- taxi. But there's lots of in between and there will now be the choice in most urban areas.

The first internet business? Seriously? Wow, that's some pretty serious creds you're claiming there Brad. That means that in the creation of the (internet) World. First there was GOD (AL GORE) who created the world -then Adam and then Eve and then YOU, -THE SNAKE. -The first person to ever sell anything on the Internet ! Just out of curiosity , -the snake sold Adam an apple .... what did you sell and to whom?
Because if you're not sure who the buyer was, (and as long as it wasn't some kind of embarrasing sex toy) if its OK with you id like to claim to be Adam. Yes, I know that wouldn't make me look too smart, but heh this is 2021 ! Who cares about smart? Or even SANE ? its being first that counts. Just ask Marjorie, Ted or the Clorox kid. Brains are definitely NOT required in 2021. Wild claims and outright lies, unfortunately ARE, but heh nothings easier to create than a lie -anyone of those guys can do 50 a day. -EACH -Adam
BTW, - calling you the Snake was not a dis. What the snake REALLY sold Adam was knowledge. It seems that back in those days the closer you were to God the less you believed in or wanted anything to do with KNOWLEDGE - Adam had to be tricked into acquiring any.
- HMM, when you think about it, nothings really changed much since the garden has it? The fanatically religious STILL have to be tricked into biting into that apple. Need proof? I refer you back to those 3 highly successful clowns above.

Add new comment