One of the big debates about the future of robocars is over the economics, particularly those of robotaxis. The most popular model has companies like Waymo, Cruise or Zoox making dedicated robotaxis, which they operate in fleets to provide a transportation service, sold by the mile, ride, hour or in a monthly subscription—or through resellers like Uber.
Other models include a car company which makes a robotaxi that it sells to individuals (as car companies have always done) and they use them as private cars but also hire them out to provide robotaxi service when not using them. They might hire them out back to the car OEM, or like above to a taxi service company like Uber or Lyft. Tesla has talked about several plans, including letting Tesla owners hire out their cars in a “Tesla Network,” turning off-lease Teslas into robotaxis, or building a custom robotaxi which they might operate similar to the other companies. Tesla was scheduled to show off prototypes of such a car in August, but has delayed this.
It’s complex, and uncertain, which approach makes the most economic sense, and which (if any) will win. For any to win, rides must be much cheaper than typical Uber prices today ($2 to $3/mile) since we know that price makes a nice, but not world-changing business.
Elsewhere, you can find my analysis of the costs involved in running a robotaxi fleet. The final numbers of this are still unknown—today’s costs are sometimes much higher, and some operating costs are hard to predict. Whatever, the costs, though, there are issues around how they change depending on who pays them and who makes and operates the vehicles.
Peer To Peer Sharing
I’ve been examining this question for some time. My 2009 proposal on airport-based private peer-to-peer carsharing is the first such proposal I know of, and with it and other writings I encouraged our students at Singularity University to found Getaround, the first such company. The core idea is fairly obvious, but the open question is how good it is. Getaround and Turo remain a small part of the car rental business. On the other hand, Uber and companies like it now dominate the taxi business. AirBNB is huge, and while it began as people taking guests in their homes, today, almost all AirBNB properties are full-time dedicated rentals. So does it make more sense to have commercial robotaxi fleets, or conglomerations of people’s private cars hired out when available?
Commercial fleets win all the economies of scale. They also (as Zoox plans) can be highly-custom robotaxi designs, less suitable as private cars. Private owners, however, can take a car that they use as their private car (even better, a 2nd car) and thus share the costs between their private use and their taxi use. Private owners will also fool themselves (as Uber drivers do) to ignore the cost of the depreciation of both types of driving—Uber makes a fair bit of use of this error.
There are many burdens to letting your car hire itself out. It’s not available to you (though you will typically be given a deal on a replacement robotaxi ride for trips during that period.) You must keep it clean and maintained to a higher standard (but many wish to do this already.) You can’t keep your own stuff in it, though you might have a modest lockbox in the trunk.
Taxis wear out by the mile unless used lightly. Private cars wear out by a tuned balance of miles and years. That means unless you hire your car out sparingly, the economics aren’t so different. Your car will wear out in 5-10 years, but you’ll earn revenue to compensate for that. Some owners might like that. To make the best use of your car as a taxi, you want to hire it out at times of peak demand, which is an issue if you also want to commute in the car. The dedicated fleets will take all the business during lower-demand periods as they have nothing else to do, though they will have a price floor.
If the per-mile economics get similar, than the fleets win due to economies of scale, though the private owner may view personal labor costs (cleaning, etc.) to be “free.”
Both types of vehicles still need a central service which maintains the maps, software and legal certifications, makes nice with local officials and provides remote and emergency assistance. Private owners are unlikely to provide those. Even if done through a company like Uber, it will need to be in partnership with the original developer. Private owners won’t need to maintain a depot, and will commonly own a dedicated parking space and handle charging/refueling.
Private owners can also execute Tesla’s original robotaxi plan, namely to use the car as a personal vehicle for 3-4 years, and only then put it into taxi service (perhaps as a 2nd car.) This works because while private drivers love a new car, taxi riders don’t care if a taxi is 3-5 years old if it’s clean and well maintained. The new robotaxis from Waymo etc. are wasted on their passengers, and so the depreciation can go to somebody who cares about it.
Today, being an Uber driver is 2/3rds about working for income, and 1/3 investing in a vehicle and getting income from hiring it out. Running an AirBNB or robotaxi is much more capitalism than labor, though the cost of cleaning (done by staff for a fleet and by the owner and sometimes staff for a private car) may change that. While millions are ready to work as Uber drivers, fewer are ready to be capitalists so again, it’s hard to predict.
These days though, when I look at where AirBnB and Getaround have trended, I suspect that dedicated robotaxis that rarely work as personal cars will be the most common rides. Private cars will get used during peak times and “surges” where the price goes up. Owners may in fact program their cars to say, “Only work if you can get more than $1/mile.” For fleets, hiring those cars makes sense, it’s cheaper than maintaining a fleet big enough to handle the top peaks. Though a large fleet is less expensive than it seems, it just wears out more slowly, and all vehicles get even usage.
Players are going to try both models. Well, making a consumer robocar is a much more difficult task than making robotaxi, so that will come later, but it will come and this competition will begin.
