Driverless Robotaxis Cost Twice As Much Per Mile As Human-Driven Cars: MIT Study

May 10, 2019 1:24 PM ETLYFT, TSLA, UBER95 Comments
Anton Wahlman profile picture
Anton Wahlman


  • It's been an article of investor faith in Lyft, Uber and Tesla that driverless robotaxis will be less expensive than human-driven ones. Otherwise, what’s the point?
  • Now comes a study from Massachusetts Institute of Technology (MIT) showing that driverless robotaxis could cost anywhere between twice and eight times per mile.
  • Unlike other studies, MIT doesn’t just subtract the driver’s salary, but it also adds all the additional costs required for driverless robotaxi operation.
  • It turns out that those costs are a lot higher, according to MIT. As a result, the net effect becomes that the robotaxi is much more expensive than the human driver.
  • If this study is correct, then this will likely deflate the valuations of the companies most tied to robotaxi hopes: Lyft, Uber and Tesla.
  • Looking for more? I update all of my investing ideas and strategies to members of Auto Insight For Wall St.. Start your free trial today »

"Man is the lowest-cost, 150-pound, nonlinear, all-purpose computing system which can be mass-produced by unskilled labor." - Fred Singer (1965), also attributed to NASA staff during the prior decade: Human: A Non-Linear Servo-Mechanism Weighing Only 150 Pounds that Can Be Produced Cheaply by Unskilled Labor

The most important revelations are the ones that change basic articles of faith: Earth circles around the sun, not the other way around, and Earth is round, not flat. If you start out with the wrong assumptions, you are going to draw all the wrong conclusions.

One of those most fashionable articles of faith as of 2019 is that driverless robotaxis will make the cost per mile cheaper. If you take that as a given, all sorts of things follow - but for our purposes here on an investment site is lots of money. Yes, billions of dollars are being spent on R&D, as well as buying IPO shares in Lyft (LYFT) and Uber (UBER) on the hope that driverless robotaxis will become a big thing because they are cheaper to operate than human-driven cars. Tesla (TSLA) most recently pivoted its future prospects entirely to this view.

This article of faith is driven by the magic trick of providing this simple formula: Car minus driver = cheaper. Presented that way, and not digging deeper, who would disagree?

Well, finally someone conducted a study of all the costs associated with driverless robotaxis - not just subtracting the driver cost from a regular car. That someone was Massachusetts Institute of Technology's (MIT) Kristen Hernandez and Ashley Nunes in this MIT paper: Autonomous Vehicles and Public Health: High Cost or High Opportunity Cost?

This MIT study was brought to my attention by Ed Niedermeyer of The Drive: MIT Paper Tackles The Challenging Economics Of Autonomous Taxis Niedermeyer does a great job in summarizing the academic MIT paper, with this being the most important bottom line for the cost comparison:

"What they found is that CDV total cost of ownership is a remarkably low 72 cents per mile, whereas the high licensing, insurance, cleaning and safety oversight costs associated with autonomous taxis combined with the low (52%) utilization rate of the current taxi fleet means robotaxis are likely to cost between $1.58 and $6.01 per mile."

Basically, under the most optimistic scenario, a driverless taxi would be about twice the cost of a human-driven car - $1.58 per mile vs. $0.72 per mile. In a less generous scenario, the driverless robotaxi would be eight times as expensive - $6.01 per mile vs. $0.72 per mile.

Why are driverless robotaxis so much more expensive per mile than human-driven cars? It's not so much the incremental up-front equipment cost, which MIT is assuming to be $15,000. Keep in mind that a driverless robotaxi must have all sorts of equipment that is not necessary in a regular car, such as motorized doors - other than the various sensors, computers, and redundant systems for other vehicle functions in case one set of systems goes down.

Rather, a driverless robotaxi would likely have to be licensed (taxicab medallion), be supervised by a remote centralized operator that would help the car out of an otherwise unresolvable pickle, and the kind of skyrocketing insurance cost the robotaxi would likely require. These costs will have to be amortized over varying assumptions of the utilization rate of this driverless robotaxi: The lower the utilization rate, the higher the per-mile cost.

For example, if riders for whatever reason - such as safety - prefer human-driven cars to driverless robotaxis, then the utilization rate could be lower. When the utilization rate is low, the cost per mile looks closer to the $6.01 per mile.

Another reason a human-driven taxi would be preferable to the driverless robotaxi version is quite literally the human touch. An elderly person may need help to load and unload groceries, or sense the comfort of safety from a human ensuring protecting the vulnerable minute it takes to get in and out of the car.

Speaking of safety, that may be the next driverless robotaxi assumption to fall. It is today's consensus assumption that the robotaxis will be safer than a human-driven car. So safe in fact that there won't be any accidents or other safety issues at all with a driverless robotaxi.

That kind of safety assumption can't be tested in the abstract before the service is launched. Yet, the service can't really launch before it has been deemed safe. So it's a chicken-and-egg problem. How does a driverless robotaxi technology company and/or operator get across this most fundamental dilemma?

Investors are today giving driverless robotaxis the 100% benefit of the doubt that they will have such ultimate safety. Yet, this assumption simply can't be proven until "after" the service has launched.

It is easy to talk in the abstract about people wanting to be guinea pigs being driven around in the back seats of driverless robotaxis. It's easy to say "yes" to something where you make the mental assumption that the service will be:

  1. Absolutely safe.

  2. Cheaper than any other alternative.

The MIT study makes a credible case that a driverless robotaxi will be between twice and eight times as expensive as a human-driven car. The safety aspect can't be proven until after the fact. How will this ever get off the ground with people other than the equivalent of NASA test pilots, who accept a high probability of death and have unlimited budgets paid for by taxpayers?

Shattering The Robotaxi World View

It's painful when someone's world view comes tumbling down. Once upon a time, it brought convulsions to those who had assumed that the Earth was flat and that the sun revolved around Earth.

In 2019, the driverless robotaxi investors' world view may also come tumbling down, as investors start to realize that their fundamental assumptions about cost and safety were 180 degrees wrong. This would have material investment implications for those companies that have staked the largest portions of their future valuations on autonomous robotaxis gaining widespread acceptance quickly: Lyft, Uber and Tesla.

If the latest report about Tesla's Autopilot management team is true, this realization may have already have dawned on the people closest to the situation: Tesla restructures Autopilot software team, Elon takes the reins. As of this writing, we don't know how to fully interpret this report: Did Tesla's top Autopilot executives resign, were they demoted, or were they simply fired? We don't know yet, but MIT's study showing that driverless robotaxis are at least twice as expensive per mile as human-driven cars can't have been an argument for these people to stay in their positions at Tesla.

Get More Mileage Out Of Your Auto Investing

The auto industry moves fast, and it can be tough to stay on top of everything that's happening. I designed Auto Insight For Wall St. to keep you aware of all the changes without your needing to spend all the time. I attend new vehicle launches, press conferences, and industry events and share that insight with my subscribers. Looking for more? Sign up for a free trial today.

This article was written by

Anton Wahlman profile picture
I am a former sell-side analyst -- UBS 1996-2002, Needham 2002-2006 and ThinkEquity 2006-2008. These days I review automobiles and other technology products, as well as analyze the automotive and technology industries, and coming up with long/short ideas. I also continue to write (less frequently) on macroeconomics and politics.

Disclosure: I am/we are short TSLA. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: At the time of submitting this article for publication, the author was short TSLA and long GOOG and GOOGL, AMZN, FCAU and GM. However, positions can change at any time. The author regularly attends press conferences, new vehicle launches and equivalent, hosted by most major automakers.

Recommended For You

Comments (95)

To ensure this doesn’t happen in the future, please enable Javascript and cookies in your browser.
Is this happening to you frequently? Please report it on our feedback forum.
If you have an ad-blocker enabled you may be blocked from proceeding. Please disable your ad-blocker and refresh.