Robotaxi Economics


The vision of many of us for robocars is a world of less private car ownership and more use of robotaxis -- on demand ride service in a robocar. That's what companies like Uber clearly are pushing for, and probably Google, but several of the big car companies including Mercedes, Ford and BMW among others have also said they want to get there -- in the case of Ford, without first making private robocars for their traditional customers.

In this world, what does it cost to operate these cars? How much might competitive services charge for rides? How much money will they make? What factors, including price, will they compete on, and how will that alter the landscape?

Here are some basic models of cost. I compare a low-cost 1-2 person robotaxi, a higher-end 1-2 person robotaxi, a 4-person traditional sedan robotaxi and the costs of ownership for a private car, the Toyota Prius 2, as calculated by Edmunds. An important difference is that the taxis are forecast to drive 50,000 miles/year (as taxis do) and wear out fully in 5 years. The private car is forecast to drive 15,000 miles/year (higher than the average for new cars, which is 12,000) and to have many years and miles of life left in it. As such the taxis are fully depreciated in this 5 year timeline, and the private car only partly.

Some numbers are speculative. I am predicting that the robotaxis will have an insurance cost well below today's cars, which cost about 6 cents/mile for liability insurance. The taxis will actually be self-insured, meaning this is the expected cost of any incidents. In the early days, this will not be true -- the taxis will be safer, but the incidents will cost more until things settle down. As such the insurance prices are for the future. This is a model of an early maturing market where the volume of robotaxis is fairly high (they are made in the low millions) and the safety record is well established. It's a world where battery prices and reliability have improved. It's a world where there is still a parking glut, before most surplus parking is converted to other purposes.

Fuel is electric for the taxis, gasoline/hybrid for the Prius. The light vehicle is very efficient.

Maintenance is also speculative. Today's cars spend about 6 cents/mile, including 1 cent/mile for the tires. Electric cars are expected to have lower maintenance costs, but the totals here are higher because the car is going 250,000 miles not 75,000 miles like the Prius. With this high level of maintenance and such smooth driving, I forecast low repair cost.

Parking is cheaper for the taxis for several reasons. First, they can freely move around looking for the cheapest place to wait, which will often be free city parking, or the cheapest advertised parking on the auction "spot" market. They do not need to park right where the passenger is going, as the private car does. They will park valet style, and so the small cars will use less space and pay less too. Parking may actually be much cheaper than this, even free in many cases. Of course, many private car owners do not pay for parking overtly, so this varies a lot from city to city.

(You can view the spreadsheet directly on Google docs and download it to your own tool to play around with the model. Adjust my assumptions and report your own price estimates.)

The Prius has one of the lowest costs of ownership of any regular car (take out the parking and it's only 38 cents/mile) but its price is massively undercut by the electric robotaxi, especially my estimates for the half-width electric city car. (I have not even included the tax credits that apply to electric cars today.) For the taxis I add 15% vacant miles to come up with the final cost.

The price of the Prius is the retail cost (on which you must also pay tax) but a taxi fleet operator would pay a wholesale, or even manufacturer's cost. Of course, they now have the costs of running a fleet of self-driving cars. That includes all the virtual stuff (software, maps and apps) with web sites and all the other staff of a big service company ranging from lawyers to marketing departments. This is hard to estimate because if the company gets big, this cost will not be based on miles, and even so, it will not add many cents per mile. The costs of the Prius for fuel, repair, maintenance and the rest are also all retail. The taxi operator wants a margin, and a big margin at first, though with competition this margin would settle to that of other service businesses.

So what will the price be of your robocar ride? Today, an Uber ride tends to be a mix of a "per ride" fee (which hopefully covers the cost of getting to you) and about $1 to $1.50 per mile and about $13/hour in the USA. (In Peru it's $3.60 per hour and just 40 cents/mile.) Depending on your speed, USA UberX will cost you $2 to $2.50 per mile. As such, $1/mile would be a very attractive, but also very profitable introductory price.

Competition should drive prices down to more normal margins. Typical gross profits in transportation are 50%. This suggests from 20 to 25 cents/mile for the urban solo ride up to 50 cents/mile for the 4 passenger sedan (or just 12 cents/mile/person if you fill it up or carpool in it.) This means the small car at 25 cents/mile has revenues of only $35/day (around $12,500 per year) with COGS of $16/day. I have not factored in per-ride fees or trip minimums, which are normal in Taxi/Uber services. (The Uber $5 trip minimum covers a whopping 20 miles in this car.) I expect trip minima to drop to compete with $2.50 transit tickets, especially when pooling.

Google runs a gross margin that's very high (about 67%) but on the other end consider that Uber (though not profitable yet) takes 20% of its gross cost of a ride (and pays its own expenses from that.)

Things still to learn

The cost of maintenance is still a bit of a guess. As noted, for today's cars it's about 6 cents/mile, including 1 cent for tires. Electric drivetrains do require less, but how much less? Cheaper cars tend to have cheaper parts, it does cost more to maintain a $30,000 car than a $15,000 one. On the other hand, if you are running a fleet, you will pick parts with the lowest total cost (both purchase and maintenance,) designing them to wear out at just the right times, and accounting and measuring the variability of that.

Cleaning is another unknown. If a passenger soils the interior in a way the internal camera sees, you can bill that to the passenger. You still will need exterior cleaning from time to time (especially on the luxury services) and at least a vacuum of the interior every so often. It will all be done very efficiently as cars pull up to the cleaning center. Robotic vacuuming is not out of the question since the cars will position themselves well and can move to help the vacuum robot. Exterior wash will be automated other than for tough things like bird droppings.

What are these small cars?

I describe a number of ways a 1-2 person robotaxi might be very cheap in this article. Let's go into more detail on why it might get manufactured for around $10K (retail would be higher.) Note that there are already many low-end 4 person gasoline cars which retail around $14,000 such as the Nissan Versa and Chevy Spark. Outside the USA, there are even cheaper cars, like the $8,000 Renault Twizzy. Some of these cars are full-width and come with dashboards full of stuff, pedals, wheels, doors, hatchbacks and more. They meet crash-safety standards and the rest of the FMVSS.

The basic commuter car will be:

  • Just over half the width and under 2/3 the length of these cars
  • Weigh perhaps half of those cars, using far less materials
  • Have no dashboard, no radio, no pedals, no wheel
  • Have only one door instead of 3 or 4, and no trunk (just an internal cargo area)
  • Have only 1-2 seats (with associated airbags and components)
  • Have vastly fewer parts
  • In the urban version, not go over 45mph, allowing a low-cost motor

What it will have is a battery pack. At 100 watt-hours/mile, a 10kwh battery would give it 100 miles of range, more than enough with recharging breaks during the day to cover the average of 140 miles/day it will do as an urban taxi. The forecast cost of $150/kwh puts that pack at $1,500. Expensive, but when combined with the much simpler electric motor power-train, a reasonable cost.

It will also have about $1,000 of sensors and computers in it. These prices get this low because these vehicles are made in the many hundreds of thousands, and eventually millions. It also has a few things not found in cheap cars, like a single power door lock.

The higher-end vehicle at $16,000 is pretty similar in design, but the extra money goes into making it more luxurious:

  • Better seats and interior materials, replaced more often
  • Better suspension for smoother ride
  • Better sound padding
  • A screen and sound system, or better ones than in the base vehicle
  • A distinctive exterior with better styling and fancier paint, to let people see you are in the higher end vehicle.
  • Power windows. (Both cars have power locks, and neither car need seat position adjustments.)
  • Some fancier interior furniture (tables, drink holders, laptop holders etc.)
  • Possibly a motorized door for easy entry (one of the more expensive add-ons)
  • Possibly a sunroof (but there will also be open-top cars for order if you want them and the weather is good.)

When I have examined the lists of luxury features on high-end cars, I have been surprised at how few of them are needed in the luxury robocar. Driver assist functions, 10-way motorized seats, motorized mirrors, telescoping, titling leather-wrapped heated steering wheels, fancy wheels, fancy lights and many other features pay no role in the robotaxi. Others will be less useful like fancy entertainment systems (vs. noise canceling headphones and your mobile device.) Look at the features of a typical luxury car and consider how many are less important or unimportant in a robocar.

My numbers are aggressive -- so if you doubt them, increase them. The result is still a robotaxi at a very low per-mile operating and capital cost, which means in a competitive market, it will be available at a price that easily competes with private car ownership, and even subsidized bus rides.


Some speculation has electric cars with very long lifetimes arriving, including the promise of a "million mile battery." The batteries and motors of an EV might last a million miles, though the interiors, exteriors and many other components may need replacing from time to time. They will in fact be designed for easy replacement, but it still will have its cost. A million mile car will still do the same number of miles of service per year, though and thus last many more years. It may decline not because of lack of function or appearance, but simply going obsolete in a fast moving tech world. As such, while it's hard to model these vehicles as yet, they will not provide quite as good a cost reduction as might be expected just because their batteries and motors are lasting longer.


Hi Brad,

Sashi sent me the link to this, and in general I love the topic. I realize for most it seems ironic that the former racecar driver promotes autonomous driving, but I am absolutely all for it, can't wait!

My only commentary here is on the units for power. Your sheet shows power draw for the electric vehicles as kwh/mile, but then the Prius is listed as MPG, completely different. Is it possible to convert pump gas into kwh/mile? Honestly I haven't done the math to convert the units, I'm sure it's possible, but for me I think that conversion would be more fair. If the overall power use was accounted for (separate from the cost of oil atm) it might be more clear. With that, cost of oil (gasoline) could be the input to determine cost/mi for the ICE.

Thanks so much for the post and the available table to "tune" the numbers.

-Dave Mc.

I am not converting kwh/mile to mpg. I mean you can convert them, but that's not relevant here, it is only the cost in dollars. As noted, the cost is from the Edmunds total cost charts, there are a couple of links in the article to them.

In this case, the Prius is going 75,000 miles in 5 years which would be 1500 gallons and it looks like Edmunds has priced this a bit over $3/gallon, which is a touch high at today's prices but not too far off.

Hi Brad,

Your numbers are very useful for us at Mobotiq.
Speaking about such pods (like the i-road in the picture)... we're about to build one. Check our work in progress here:



I am a bit horrified that you needed to make this spreadsheet. And that it needed to be a such a slam dunk for some people to start thinking the right way.

What you didn't include is the cost of human lives wasting away on the roads pointlessly paying attention to traffic. Or the ones quickly snuffed out because of the demand, clearly massive, to not pay attention to traffic. For example, your calculations don't factor in my last car which was totaled by someone who would much rather have been looking at his phone than driving.

I do have a big web site and blog with a few books worth of writing about those topics -- they won't show up in every article!

But obviously not all the personal trauma and wasted time and loss of income in all cases.

Also, if the fleet provider is the manufacturer, as their income then comes from journey's not the sale of products, robocars will be designed to last as long as possible, and made as simple to clean and maintain and refurbish; ideally with exchangeable pods that carry people by day and freight/parcels at night. This will make the cost even cheaper.

Also, Brad didn't need to compare 'driver-time' as the comparison is with taxis or privately owned driverless cars.

If you compare driverless with driven, of course the benefits are even greater!

Is a traditional personally driven car, I am just using Edmund's total cost per mile number for comparison.

Yes, I do believe that vendors might try to make the cars last more than 250,000 miles. But you can't go too much longer, or your vehicle starts lasting too many years, and starts to go obsolete rather than wear out, like a computer.

My only point was exactly that - if you compare driverless with driven, then the benefits are indeed difficult to doubt.

It's just weird to me that the cost of our time is not taken more seriously (I'm not talking about Brad's analyses - he is a sensible guy always fighting the good fight!). For example, at a minimum wage of $10/hr and 65 mph (optimum), that's $.15/mile of my time and I pretentiously like to think my time is worth more than minimum wage. Sitting in traffic is something that will not result in 65mph (labor cost per mile approaches infinity as traffic comes to a complete stop) and is something that I dislike more than working in a minimum wage job.

But hey, if Brad can show everyone, even those people who do not value their own time, that autonomous cars are economical on other merits, great!

That is a 2nd order thing. First we want to know, how much is it likely to cost to ride in a robotaxi? Since driving your own car costs around 50 cents/mile, it's a no-brainer if you can get the taxi cheaper than this, as it seems you can. If it costs more, then as you say, the rational thing would be to be willing to pay more than the 50 cents because of the time saved. People already do that with Uber. For example, driving to downtown San Francisco, where you will easily pay $30 or more for parking, quickly becomes a win with Uber because of the time you get back and the parking you save, even at $1.80/mile cost of UberX. There are many cities now where the math has people switching to Uber from driving their own car.

A few points I'm not sure You have given enought thought:

A. Electric car vs. gas/hybrid:
- the battery of an electric car is an expensive part which wears out and needs to be replaced, probably a couple of times during the car's life. Same is not true for the fuel tank of a gas powered car. Last time I checked, this cost almost outweighted the benefit from the electric power being a lot cheaper than gasoline (at least in my country). Renault's electric car cost model, with the car owned by the driver but batteries rented at a montly fee, allows for a somewhat realistic approach to this cost factor.
- A gas powered car is capable of long range travel. You can hook up a camping trailer and go on vacation with one of them, or just on a weekend trip. You can't do the same with an electric car yet - the mileage of more than 200 miles on a single battery charge is already sufficient (you need to take a break from time to time anyway, why not use it to recharge the car battery?), but recharging times beyond 1 hour are not, and the network of recharging station is very far from the density of the gas fuel stations and very far from something that would make long-range travel with electric cars possible. With the ability of practical long range travel, a gas powered car provides more value than an otherwise comparable electric car.

B. Robotic cars
- robotic driving in the cities is still something for the future, not for tomorrow. Not a far future, mind You, but imho at least 5 years, more likely something like 10, maybe even 20 years. Robotic driving on the highways is probably something we will have very soon though. Computers are very good at handling situation where all variables are known and well defined. A closed environment like an automated high-rise storage facility applies to this description, a highway comes close enough so the advantages of robotic driving would with todays technology already outwight the disadvantages statistacally and increase the overall safety in this area. But the same is not true for cities - the range of unexpected situations the robot wouldn't know how to handle is just too big. Imagine a person walking on stilts long enough to step over a car (and doing so). Imagine a kid playing hide and seek in a quiet suburb area and hiding in a carton in the middle of the streat. And countless other unexpected situations a computer wouldn't know how to handle.
- You compare robotic taxi cars with human steered driver owned cars. But chances are, when robotic driving becomes available at some point in the future for taxi cars, it will become a purchasable extra for new driver owned cars as well. This makes things like the parking costs advantage largely invalid - why paying for parking while at work when You can send Your car home instead and let it pick You up in the evening? Why paying for parking during a short shopping trip if You can let the robot driver circle the block for a couple times and let it pick You up afterwards?

Electric cars and batteries are on a rapid curve of improvement, both in cost (going down) and efforts to improve lifetime. Tesla now claims with proper management they can get 1% degradation per year. But it goes further than that:

  • Robotaxis will drive their battery "perfectly." No sudden acceleration or hard braking to overheat it. Drive it from 70% of full to 20% of full and then recharge it every time. Humans don't drive like this.
  • Robotaxis will tolerate a degraded battery in ways a human driver won't. A battery degraded to 70% is unacceptable to a human owner, but it is only a modest problem (dead weight) to a taxi which just doesn't get sent on longer trips.
  • Robotaxis aren't used for camping etc. Electric cars aren't used for long trips. If you want a long trip the taxi company sends you a gasoline taxi, or if need be, a rental car you drive.

Of course, this article is about what will the robotaxi cost when it has become moderately mature. Of course they are not out yet. You may be coming into this without a lot of other context present on this web site. But if you thought to say, "Imagine..." in a few minutes, trust me, so have the teams after thinking about it for 10 years, and they have put that situation in their sims.

Yes, some people will purchase cars. If they don't live close to work, I do not believe it will be economical to send the car home, as parking somewhere roughly near the office will be much, much, much cheaper than sending it home or having it circle the block.

- Electric cars and batteries improvment: I was not aware of this developmen. It was not the case when my last car reached it's end of life and I was considering an electric car as an option.
- Robotaxis are not used for camping etc: Right, robotaxis aren't. But a driver owned commuter car that offers that option provides more value than a commuter car that can be used for commuting and shopping only. Means, an electric car is currently not only more expensive than an otherwise comparable gas powered car, it offers one important functionality less, and therefore much less value for money.
- Parking: I think You are right about parking while at work. But when I see the prices for parking in my home city area, I'd estimate the cost for parking during shopping in the city about 16-19 times more expensive than letting my current car circle the block (if it already had a robotic auto-pilot and could perform this task). Circling the block would probably become even cheaper in case it would be possible to let the car deliberately join a traffic jam for the shopping time instead of avoiding traffic jams.

Key to the use of taxi fleets is usually fitting the right vehicle to the task. Single person car for one person riding a short distance (80% of trips.) Larger car, maybe with gasoline, for camping trip. Sports car with wheel for drive down a mountain road.

Today's parking might seem more expensive than circling. Circling costs $10 to $20 per hour, depending on the car today. No parking lot that charges by the hour is this expensive, though some flat-rate lots cost more. But that's today. During the parking glut created by people switching to taxi mode, the bottom will drop out of the parking market. A parking space that is unused is wasted, best to fill it at almost any price. A robocar wanting to park will:

  • Drive to the back/top of the lot where no humans park if requested
  • Park valet style, using much less space; better than valet style -- don't need to be able to have a valet get in and out
  • If a one person taxi, will put 4 to 5 in the space that one human parked car takes, due to small size and valet density.
  • Leave, or pay more, if the price goes up.
  • In particular, if a human driver willing to pay $8/hour shows up at the gate, the lot can let them in because the robots will leave and make room, so they never take your premium capacity
  • If electric, make no emissions, leaving not soot on your surfaces (a real problem in lots)
  • All payments and operations will be done via computers, no time taken by staff

If you own a parking lot it literally costs you zero incremental cost to let the robocar wait there. No operating costs, no opportunity costs. And there will be hundreds of other lot operators also with zero incremental cost bidding to get what the robots will pay because it is all gravy, even if it's just 20 cents/hour. It's 20 cents/hour you don't get if you say no.

- multiple vehicles for multiple tasks: having a second car specialized on some of the tasks means additional costs - especially for purchasing it, but also taxes, insurance, possibly an annual toll, etc., most of which will not be offset by the costs saved by using the primary car less than if it would be the only car. Many of the secondary tasks the car is used for actually only make sense financially You attach to the mile driven not the total costs per mile, but only the costs for driving an _addistional_ mile with a car You allready pay the taxes etc. for because I need it for other tasks like the daily commute. If I already have a car for my daily commute, and want to go for a weekend trip, the per-mile-costs I have to pay _additionally_ to what I already pay for the car are slightly less than if I would go by train (this may be different in other countries). Buing an additional car for such trips only would be prohibitively expensive, using this secondary car for such trips would be, depending on the frequency and size of such trips, at least(!) 3 times more expensive than going by train.
So what I came up when buying a car was a series of questions, starting with: Do I really need a car? (Answer: unfortunatelly yes) If yes, what is the minimum of car I really need? (Answer: a Smart fortwo, maybe even a Renault Twizy) What secondary tasks come up frequently enough so adding size and functionality to the car would pay off compared to renting or buying a dedicated vehicle for such tasks? (it resulted in a tiny 3,5 m long gas powered 4-seater with a trailer hook).
- costs of parking vs. circling: as of now, a parking slot, even an expensive one, is usually still a block or more away from where I actually want to go. I not only have to take this walk, but also to pay premium parking fees for the time it takes me. This walking time would not be part of the equitation if an autopilot-equipped car could drop me off and pick me up at the front door of the place I want to visit. Means, circling time would be much less than parking time for the same task. But You are right, having the car going on auto-pilot to a cheap place a couple of blocks away after it dropped me off and park there for most of the time would likely be cheaper than letting it circle. I made the error of comparing parking now vs. circling in the future instead of future circling vs. future parking.

A car gets older in 3 ways. It ages by the mile driven (or more closely, by the hour driven, but most figures are done in miles.) It ages by the year (as things corrode or degrade) and it ages by the year in another way -- it becomes obsolete.

Some products, like phones do almost all their aging through obsolescence, and a little bit by use.

Owned cars age through a mix of time and miles. They are designed for that, with a life of about 200,000 miles and 20 years. So leaving your car idle can be slightly wasteful.

But taxis age much more by miles than by years. They are made from stock cars today (which are designed for 20 years) so they barely age by years at all right now. Since 5 years is not much time in the world of car innovation, they don't age by obsolescence that much. In the future, the pace of change will speed up. It's already happening to electric cars.

Once your care does most of its aging by the mile, there is no more thinking that the cost of driving is just the cost of fuel and parking, the way people think about their cars today. That is going to be a big shift. Some services will price to try to match the old way of thinking, like selling you unlimited taxi rides for $500 per month or similar models. Or to match a car completely, $300 per month, plus $50 for a "tank" of 500 miles, plus an annual fee for insurance/maintenance. Just to make you feel good.

As I said, hours are actually a better proxy than miles, but we're not used to that at all. (In boats and planes it's all measured in hours.)

As robotaxis allow transport as a service to replace car ownership, the very large customer base of the bigger players should allow a degrre of ride sharing. This would obviously change the relative costs per person per mile of the 1-2 seat robocar vs the 4 seat version.
I have not seen any modelling detailing the maximum possible load factors, but I presume Uber have a reasonable idea.
Any thoughts?

I have articles on the future of transit layout out official sharing with vans and cars. In the long run:

  1. Outside rush hour, the solo robotaxi is going to be so inexpensive that, except for long trips, all but the lowest income (or those seeking a social experience) will prefer to go solo for the privacy, comfort and no need to do any transfers at all, even minimal delay transfers.
  2. At rush hour, cities and highways will give rewards to people who pool rides -- things like carpool lanes -- which will encourage pooling in cars and vans and sometimes buses.
  3. For longer commutes, and longer road trips, cost will push people to pool. This will be a mix of solo car to the pool meeting point, and pool car/van along the long haul part.

For example, for a 4 mile trip inside the city, at $1.50 in a private car and $1 shared, I'm going to go with the private car. For my 50 mile trip to San Francisco, at $16 solo and $8 shared with 2-3 minutes added trip time, I would be more attracted to the shared ride.


Great post. I have heard some researchers say cleaning is a cost that is under appreciated in these types of estimates, but agree that robots could be used and similar to the way Uber charges customers who soil the car, in car cameras would be a clever solution to handle that.

My estimate is pretty close to your "cheap small" scenario. I have 31 cents per mile. I also included the cost of a teleoperator to remotely direct the car 10% of the time. I did some monte carlo analysis and took Google's published reports to estimate what we think the manual intervention rate at the time of commercialization will be. So, this third party labor component is actually my biggest cost bucket and my other costs are a little less than yours.

For cost of the car, I used my colleague's battery cost decline curve for an electric vehicle and he thinks a 100 mile range car could be $15K in 2020.

Happy to share the models with you if you're interested.

that's 31 cents for the cost to the consumer*

Add new comment