Ark Estimates $11–12 Trillion Robotaxi Market in 2030

ARK estimates that autonomous driving could reduce the cost of ride-hail significantly, expanding the addressable market. Today, the average price of an Uber is $2 per mile, while Didi is $0.50-$.70 per mile. We estimate that autonomous ride-hail vehicles will have higher utilization rates than human-driven cars, as well as lower labor and insurance costs. At scale, ARK estimates that an autonomous taxi platform could price rides profitably at 25 cents per mile.

Sources: ARK Investment Management LLC, Didi S-1, US Department of Transportation, Internal Revenue Service

Ark estimates for the perceived value of time suggest that before autonomous ride-hail is widespread and costs decline to 25 cents per mile, it is likely that there would be substantial demand today for a higher priced service. For example, given consumers’ perception of the value of time, we believe that an average robotaxi price of $0.60 — $1.10 per mile would be sufficient to address the rough 5 trillion urban miles that people in the United States and Western Europe drive annually.

In China, however, there is likely more demand for a service priced at $0.50 per mile, given the inexpensive human-driven ride-hail options available today, as shown below.

The robotaxi should be nearly $1 trillion as the price gets to about $1.10 per mile.

On the Tesla Q2 2021 earnings conference call, Elon Musk called a self driving fleet of cars as “one of the most valuable things ever created in the history of civilization”. This is clipped in the Warren Redlich video at about 56 minutes.

SOURCES- Ark Invest, Warren Redlich, Tesla
Written By Brian Wang,

12 thoughts on “Ark Estimates $11–12 Trillion Robotaxi Market in 2030”

  1. The level 5 label is a big distraction to impress investors. I could empty a big parking lot and introduce level 5 driving on that parking lot. What Waymo does, they spend 2 billion a year to drive 600 highly equipped cars (part with safety drivers) on a geographically small area that has good weather and clearly mapped streets using hundreds if not more engineers. Yeah After many years you can work out the bugs with brute force, only to start over for the next area.

  2. I drive a lot in Pittsburgh and have encountered a number of self driving vehicles. So far so good. The only accident I have seen was a fender bender in SF. It was not that dramatic. We will have 5g and with improvements in edge computing a lot better anti collusion controls. Couple this with having pedestrian clear zones at certain commuting times on certain routes coupled with different zones for bicycles and pedestrians, IMO it will work. One can move a lot of people if the vehicles are moving and not stuck in traffic. We just have to get with it and give it a try. I think it is far from a given that Testla and Waymo will capture the market.

  3. Many(Tesla fanboys) will claim the answer is to eliminate human drivers from the problem. Imagine trying to convince the most of the population to give up human driving, making their existing vehicles more, or less worthless. In the US vehicles made useless would need to be paid for by government.
    At some point, nearly all vehicles other than antiques will be drive by wire that could be retrofitted with self driving hardware. Presumably, there will be standard protocols, like the OBD2 protocols used today, that will allow easy interface with a vehicles brakes, steering, ect. At that point, eliminating human drivers might have acceptable costs, and all sorts of improvements could be realized.
    High speed platooning could be done on highways greatly decreasing the energy cost of travel, and increasing the capacity of existing highways. Vehicles will adjust their speeds so they will reach intersections when they are clear. Existing parking spaces can be converted into traffic lanes. When cars are not being used, they will aggregate in "V2G lots" at grid substations, or power plants where they will be near high voltage transmission lines. Automobile accidents would be nearly eliminated. Problems with hardware, or software could usually be detected before they failed during travel. Actuator hardware by monitoring for abnormal voltage, current, ground faults, or arcing. Controlling hardware, and software by comparison of results.

  4. An excellent TV premise: AMC and the Walking Dead would be proud.
    Self-Driving Cars vs the Last Remnants of Humanity (perhaps a catchier title)

    Cue the opening lines of the Skynet and the Terminator series:
    "… 3 billion human lives ended on August 29, 2022. The survivors of the FSD Level 5 'trial' called the war Judgment Day. They lived only to face a new nightmare, the war against the Teslas. The Musk which controlled the Teslas, sent two Teslas back through time. Their mission: to destroy the leader of the human Pedestrian Resistance… The first Tesla was programmed to strike before the PR. It failed. The second was set to strike at PR, when it was still young. As before, the Pedestrian Resistance was able to send a lone warrior. A protector for PR. It was just a question of which one of them would reach him first…"

  5. I agree. Robotaxis may require a near human level of artificial intelligence to anticipate and respond to distracted pedestrians, weather, animals, and hardware failure.

  6. I was onboard the robotaxi bandwagon, but the seeming slow improvement of Tesla's self driving software has me questioning it. I thought that with all the data harvested by their cars, training the neural networks their frankly amazing hardware platform enables would be done by now, or at least FSD would be improving demonstrably every few months. Perhaps the problem is more difficult than anyone understands.
    Then there is all the lobbying that will be done to obstruct it's legalization by incumbent businesses.
    Robotaxis may be a thing in Singapore in 2030, but I think not in the US.

  7. agreed. likely a hard-wired local system that will interact with 'subscribers' to navigate the dense urban jungle.

  8. Always with the sparse and simple highways — never the dense urban corners with boarders jumping curbs, cyclists weaving through traffic, adjacent workers in the next lane unpredictably shifting in and out of your live lane, jaywalkers failing to take the perpendicular route, crosswalks with late and early users; taxis with little knowledge of turn signals, brakes, and the basic concept of conservation of momentum: this is the real roadway – go and Level 5 FDA that.

Comments are closed.