Tesla began a pilot demonstration project of their 2-seat no-steering-wheel Cybercab recently with a fleet off 45 cars operating in Austin, with no employee in the car. In concert, they put out an “interest form” where people can add their name to a list of individuals and fleet operators who might wish to purchase these vehicles and put them to work. While Elon Musk has frequently presented this as a highly lucrative business opportunity, the reality is it won’t likely be lucrative--if it works at all.

I have previously outlined the problems with this concept. Ironically, it is one I pitched to Musk back in 2010, back before I realized it wasn’t so good an idea. Now that the CyberCab is further along, let’s go over some key issues.

If Tesla does sell you a Cybercab--in theory for just $30,000--one must be clear that Tesla will still operate it. They will not only provide and maintain all the software; all use of it will go through them. Riders will need to book rides through the Tesla app, access the car and pay for it through that app. Tesla will control everything, and set the prices, and pick which car serves a rider. While you may have the opportunity to clean it and perform minor service, it’s more likely it will also be cleaned, serviced and charged by Tesla. You will only decide when to take it in and out of service, and where to store it when it’s not in service. You won’t be able to “put it into Uber” or run your own private service.

Tesla will have its own large fleet. When a rider request comes in, they are probably motivated to use one of their own vehicles, unless yours is much closer to the rider. You might find yours is only used during the peak “rush hour” times, including late-night bar-hopping time on the weekend.

Tesla controls the billing and customer, and they alone decide how to allocate the money. As such, it’s easy to calculate what portion of the money they will share with you, and the answer is “barely enough.” They will pay you enough to make it marginally valuable to do this, because they will want people to decide to do it. It’s hard to see why they would pay you more than “barely enough.” Why give people more money than they need rather than keep it for Tesla?

The idea that you could buy a $30,000 car and it could somehow make you $30,000/year is ridiculous. No investment returns anything remotely like that unless it’s incredibly risky. If the Cybercab is a modest risk investment, it will return a profit of perhaps 6-10% per year (after paying your costs.) If it pays much more than that, Tesla would keep it for themselves, not sell it to you, because they will make more money from it than they make from selling it. So if Tesla is rational, with focus on their own profit, it will be an OK investment, but not a good or great one. (Again, if it is one at all.)

One key factor that could alter this is the the buyer can do things cheaper than a corporation can do them at scale. This is not at all impossible. There are many costs associated with operating a robotaxi. They need a place to wait when not in use, and you might already own that land (a driveway) and not be using it, making that cost sort of “free” to you. Tesla will need to pay for that parking/waiting space.

Tesla will have to pay people to clean and do minor service on the vehicles, and while you would need to spend your own time doing this, you may consider it “spare” time that doesn’t cost the full-freight cost of a cleaning staff.

One thing you’re very unlikely to beat Tesla on is the cost of charging. Tesla builds and operates the world’s largest charging network. The Cybercab is promised to be able to charge itself inductively, though today it uses a regular Tesla/NACS plug. You might be able to be a cheaper “plug jockey" as Tesla does not yet have full-serve charging stations. If you have solar on your roof with excess capacity, you might be able to charge the car with electricity you already paid for, and Tesla can’t beat that price. (But Tesla is also one of the world’s largest solar power companies, so don’t hold a lot of hope.)

When it comes to non-minor services, the car will drive itself to the Tesla service center, so you are not going to beat Tesla on that.

Tesla has a spreadsheet which will do the math. Their profit on operating the car in their own fleet will be compared with the profit from selling the car minus what they have to pay you to operate it. If they can get away with paying you less, it works. Otherwise, it shouldn’t.

Temporary and Irrational Motives

I say shouldn’t because there are short-term conditions and irrational ones which can alter things. In the first few years, Tesla could decide it’s quicker and cheaper to get customers to do a lot of the work of scaling. They only have so much management bandwidth. Eager mini-fleet operators wanting to get in on the “exciting” robotaxi game can save them effort. Those operators might not look too hard at the numbers and provide a source of what are known as “suckers” willing to take a deal that’s bad, or a deal that looks good, but will end up bad when Tesla changes the terms, as they surely will as conditions change. Some might want to start operations in places Tesla isn’t wanting to go (though that requires a place where vehicles can deploy without the hard work of testing and certifying them for the local roads and making nice with local officials and setting up infrastructure for vehicle rescue and much more.)

Tesla may also make the offer better than it needs to be simply to develop excitement. This can bring value to the brand, and to the stock price. If a lot of people are buying Cybercabs to hire them out, that’s a positive story, even if they are only doing so because Tesla is over-rewarding them. The issue is that Tesla can, and must eventually change its mind on this.

The last proposition has the better chance of being real. When you build a robotaxi fleet, you need to make it big enough to handle all the peak demand. Peak demand comes during the two weekday rush-hours, and during bar-hop hours on weekend nights. If you have enough cars for that, you have too many the rest of the day, and must pay for parking space and interest on your capital. So the best answer is to provision your fleet for more middle demand, if you can hire in extra cars during the peaks.

This requires buyers who view the Cybercab as partly a personal car for themselves, ie. as a 2nd or 3rd car for the household. They reserve it for themselves off-peak, and let it hire out during those peaks. For this, the Cybercab is a low-cost, cheap to operate alternative to the 5-15 year old mid-range cars that often serve as 2nd cars in a household.

For this, the owner must accept that they will rarely use the car on-peak. They can commute on-peak if they need to, but that will waive the main source of counter revenue. They also have to accept a car that’s not at all for use on intercity road trips or other trips outside the service area. As a plus, when they go to run mid-day errands or evening trips, they are free to leave stuff in the car, as it’s theirs, though they will need to clear it out before putting it back in for-hire service. (Alternately, I expect we would see a lock-box built into the large trunk to keep your “stuff” full time.) Or Tesla could offer a service where your car, before going on-hire, drives to a service depot where they clean it and store your stuff, but of course this is done for a fee. In this case it’s not ever chosen because it was the closest to a rider.

As it turns out, the Cybercab would actually be a better sell as part-time car if you could temporarily mount driving controls in it, as Tesla has done for their test vehicles. This would assure you could drive it out of the service area, down private driveways an into rural areas and even dirt roads or the open desert of Burning Man. Or even a joystick. In my own car, Tesla FSD 13 parks badly, sticking out on the sidewalk, and so I must correct it every time it parks. I’m not the only one, and having no controls would be an issue, even with the hidden touchscreen driving control currently present on the Cybercab.

The general idea of the car that is “your personal main car” but which goes out and makes money for you when you are not using it is much more difficult. I was, as far as I know, the first to propose the idea of general “peer to peer” car sharing, and I described it to my students who improved it to found a company called Getaround , which went public but later died. The idea is harder than it seems. The reasons are similar to the reason that AirBNB started with people renting out their actual homes when they were not around into being almost entirely full-time properties dedicated only to being AirBNBs. Another company, named RelayRides/Turo, still survives but didn’t take over the car rental market and is mostly purpose-dedicated cars.

The problem is that the better this part-time-car approach works, the worse it works . Again, for Tesla, the right move is to give Cybercab owners a share of ride revenue that is “barely enough” to make it worthwhile, but not much more. So if Cybercab owners are getting more value out of using the cars themselves, they need less incentive from hire-out revenues to make it worthwhile, and so their cut decreases until it’s “worth it” but only enough.

Indeed, if the return gets too high, it naturally corrects, because if there’s actual good return, more people will want to get into the game, as fast as Tesla can make cars. That part is good for Tesla, but it floods the market, and the hire-out revenue on each car goes down with all the competition.

It Also Has To Work Unsupervised

All of this unfortunately suggests that, in spite of the interest form, Tesla may probably never sell these vehicles to consumers. Marques Brownlee (who promised he would save his head if Tesla sells one this year) can keep his hair.

On top of all this, the biggest question is whether or not Tesla FSD can be made to work unsupervised any time soon. Tesla has yet to declare, as other companies have, that they have ceased remote supervision. Until they do that, they won’t be selling any and they won’t even be deploying a large scale fleet on their own. Because it’s new generation hardware, with sensor cleaners, Starlink and a front camera, it has an edge over the older HW4 cars in the attempt.