Tesla hopes to build a self-driving system that works on their current consumer cars, like the Model Y, and it also is building a dedicated car called the Cybercab which will have just two seats and no controls. That’s mainly for use in Tesla’s planned robotaxi service, but they also have said they will be for sale, for $30,000 to private owners and fleets. While you could have your private Cybercab drive you around, the real value, according to Musk, is that you could hire out your car into Tesla’s robotaxi network. While you’re not using the car, it would be out giving rides to the public, earning fees which would be split between the vehicle/fleet owner and Tesla.
It’s been suggested that hiring out your car (CyberCab or regular) could be quite lucrative. Really lucrative. A NYC Taxi drives 60,000 miles/year and makes about $2.50-$3/mile. If yours drove 50,000 and you kept $1/mile after Tesla’s cut, electricity and maintenance, that works out to $50,000/year of profit on a $30,000 purchase. If that sounds too good to be true, it’s because it is.
I should like this idea. I first wrote about it back in 2008, it’s featured in my dramatization from then titled “A week of robocars.” Indeed, when I pitched self-driving to Elon Musk in July of 2010, I pitched this angle to him, and whether from that or his own plans, he’s now going full-bore on it. My own estimation of the value of this has diminished over time, however.
There are a few ways a vehicle could be used:
- It’s your daily driver, but when you know you don’t need it, you let it hire itself out for taxi service until you need it again. (You take a taxi if it’s not back in time.)
- It’s an “extra car” for your household which gets occasional use but mostly works as a robotaxi
- It’s a full time robotaxi, and you may even own a fleet of them.
- The vendor (Tesla) never sells it, and just operates robotaxis
You might view #1 and #2 as a bit like driving for Uber, but you don’t have to drive. Even #3 could be like that, though it’s closer to a capital business where you buy assets and rent them. You must maintain, store and clean vehicles, which at least in #1-#2 you are already doing for yourself.
Margins eventually get thin and it’s not that lucrative
But as an asset rental business, the problem is, this rate of return is too good. There’s no asset you can just buy and make a 150% return (after depreciation) on every year, not even with tons of risk. Moderate risk assets might return 5%. The S&P 500 returns 10% on average, as does rental real estate. If the cars could give a return like that, Tesla would be insane to sell them to you, they would just operate their own fleets. If they didn’t have capital (they do) they would just borrow it and keep that 150% for themselves. The leveraged rate of return is even more insane.
But if they did sell vehicles that could generate that return, everybody would buy them–until there was so much competition that revenues and returns shrunk to more typical numbers. Until returns dropped, everybody would be clamoring to put their money into these CyberCabs, and that doesn’t work.
Tesla has had trouble making up their own mind. In 2019, they set lease terms to say customers couldn’t buy back the vehicle at the end of the lease. Tesla was going to turn them into robotaxis. In 2024, they restored normal lease terms, and they also started selling off the vehicles that came back from leaning. (Older cars before mid-2023 have hardware that Tesla admits won’t ever do robotaxi without major upgrades. Tesla believes their current HW4 will someday be able to.) The lease-takeback plan was a great plan if it worked, since Tesla gets these cars for 60% of retail, and the lessee pays for all that early depreciation. Nobody cares if a taxi is brand new.
Pure Peer To Peer Fails
The first clue was the path of AirBNB. When it began, it was about people who owned and lived in real estate renting out rooms and homes when they weren’t using them. It quickly shifted to being properties only used for AirBNB, often owned by hosts/landlords who owned multiple properties. Nobody lived in them. They became mostly smaller scale versions of hotels. This also suggests the idea of hiring out your personal daily-driver car may also not be that great.
That people would do that was the theme of Getaround, a peer-to-peer car-share company founded in 2009 by some of our students, based in part on my essay proposing the concept. Getaround failed. Its later competitor, Turo, is still around but a small part of the car rental world.
Being a Getaround host was challenging. The owner had to keep it clean and ready to go. The client was booking a specific car, while a taxi customer just wants the first available one. The cars will deliver themselves, and can even first visit a charging/cleaning/service depot before going into service. All the owner need to is keep the car in reasonable shape and keep any personal property in a small locker in the Frunk or Under-Trunk. If they let their car out in an unacceptable state too much, they’ll be kicked out. (The cleaning staff might even move things into the locker.) Charging will be automated some day, as it already is with CyberCabs.
The Tesla Network is a bit more viable. The car does the work of delivering itself, while Getaround/Turo owners have more hassles. There will also be service depots where others can be paid to clean and service the cars, greatly reducing the hassle of participating–Tesla needs these for its own fleet. The CyberCabs will also charge themselves. That leaves the only hassle being the loss of use of your car, and the need to keep it clean and empty (no keeping stuff in it, though even that can be handled. If you send your car out in improper condition, your score goes down and soon you can’t rent, but you can always get it in good condition by paying a depot crew to get it ready.)
But that means it’s also easy for Tesla or other robotaxi companies to just own and run the fleets themselves. If the private owners aren’t doing anything but providing capital, why are they needed? If they’re doing work like cleaning and maintenance, then the job is less exciting.
The private owners can play a role during peak demand, however. Every robotaxi fleet operator has to choose how big a fleet to provision. They need enough cars to give customers low wait times. They won’t want to buy enough cars to do that at the highest peaks in demand, and for those, they might send out the word that they would like private cars to handle some of that peak demand, and even pay a premium price for it. In this world, it doesn’t make sense to do models 2 or 3, but model 1 could work.
At the premium price, you can afford to either have the car come to you for prep, or to pay a depot to prep it, even removing any of your gear and putting it in a lockbox in the trunk if it’s small, or at the depot if it’s too large.
This peak time service could earn income, but only at those peak times like rush hour or holiday crush–when a lot of people want to use their cars.
Model #1’s advantage is that the owner pays all annual costs like registration, parking and annual maintenance. They also install a charging station in their home. Chances are the car won’t need a charge during its shift, though it might need a cleaning. That makes the network’s costs pretty reasonable, mostly what they can talk the owner down to in a competitive market with lots of owners hoping for revenue. Tesla will keep the lion’s share, as owners have no other choice to go to. They’ll pay just enough to get enough of the lowest bidders to offer their car for hire. That will be operating cost plus a small margin. After all, that’s what Uber does and they have to compete with Lyft and taxis. However, this could help Tesla undercut Waymo at these peak times, but it will be Tesla making the profits, not the owners of the cars.
Tesla owners won’t be able to hire their cars out to other companies like Uber, unless Tesla agrees to do that, and for that Tesla will take as big a cut as they can. Robotaxis aren’t self-contained taxis, they get connectivity, maps, software updates, support and remote assistance from their OEM and those have a cost, sometimes a per-mile cost, that won’t be free.
So How Does It Look?
Expect Tesla to operate their own Robotaxi fleet, with Cybercabs owned by Tesla, and used Teslas (possibly modified) that have come back from leases. Using off-lease cars is a big win, as nobody cares if a taxi is 3 years old, and so some lessor eats 40% or more of the depreciation on the vehicle. Right now, 3 year old Teslas have HW3, which won’t ever do Robotaxi, but in time that might change, unless HW5 (which is not yet released) is needed. If so, off-lease cars will need hardware upgrades, possibly also removing their steering wheels and pedals.
At times of peak demand, the Tesla network might make a call for private owner’s cars to join during the surge. They’ll take the lowest bidders which can be clean and ready and get to a rider the soonest. When the surge is over, the cars will head back to their owner. This will mainly work for cars not used in commuting (commute hours are the peak demand time.)