Here is the current CEO of Uber explaining why Uber vastly expanded the previously existing TAM for black cars and taxis:
He’s citing Jevon’s paradox - that the easier/cheaper/less friction something useful is, the more it will be used. Today, there are far more Uber rides than there were taxi rides 10 years ago, even accounting for more people in the world.
I don’t know anybody who thinks a successful robotaxi service wouldn’t be a big market.
Most of the discussion has been about various aspects or arguments put forward by enthusiasts, like “it will be the end of privately owned vehicles” which seems exceptionally unlikely to me, given the convenience of owning your own vehicle and having it available 24/7 at a moment’s notice. Not that there aren’t downsides to that: purchase cost, maintenance, and for some housing it (garage, parking lot) etc. It’s just that there are multiple examples of how “renting” would be cheaper, but people still prefer to own: everything from houses to washing machines to, well, cars.
The comparison with Uber fails because there are so many differences. Uber got big quickly because they were using everybody else’s automobiles - and unused capacity - which is significant in an industry with high capital requirements. A comparison with the taxi industry is also verklempt as taxi companies had to put out big bucks to buy the metal, and on top of that buy the medallion which could run 3 times the price of the car, and then find people who wanted to “share” the daily revenue. (At the beginning the Uber split was less than the taxi-cab split. That’s not necessarily true anymore.)
The Uber model also allowed “taxi” service in areas where it was economically unfeasible to run a traditional taxicab. The car, after all, was “already paid for” and there was no medallion cost, and if someone wanted to do it in a semi-rural area they could, even part time.
All of this is to say there really wasn’t a TAM calculation that could be relied on because the services, while seemingly the same, were actually quite different under the hood.
Finally, I’ll just say that there is no proof that Tesla will be the one to walk off with the glory, or that this is a market which will provide network effects to have a single winner, so it may be split several ways. It’s also unclear if this is a business Tesla will own, or (more likely given recent statements) they will profit by 1) selling cars and 2) taking a percentage of fees. If it’s #2 (and surely that will be part of it) then there could be competitive pressure on fees, unlike taxi cab service charges which were highly regulated.
There’s still a lot of unknowns, (deadheading, cleaning, insurance, vandalism, etc.) but no, I don’t think anybody thinks “this can’t be a business.”
I mean, for a certain value of big. The entire market cap of Uber is “only” $145 billion dollars. Which also includes UberEats and their other delivery services, and all of their global services. A ballpark estimate of the value of their transportation business in North America and other markets that might get robotaxis in the next decade would be about $100 billion. Which is a sizable amount of business and all, but not a massive market.
Robotaxis might expand that market somewhat - but I would temper expectations. The original “Tesla Network” concept would have not only eliminated the driver, but also recaptured nearly all of Uber’s efficiency advantages that come from not having to own, store, or maintain their own cars. They were monetizing a lot of unutilized, near-zero marginal cost capital and service. It doesn’t look like that’s feasible for robotaxis, so the only expense they save is the driver in the car.
Which is nice and all, but now there will be other expenses that robotaxis have that Uber doesn’t. Obviously they now have to carry their own capital costs for owning their own vehicles - which will be substantial. There will be the “guy in the chair” expenses of however many remote “call for help” folks need to be in place, as well as the remote folks who go out and help cars that are stuck. Robotaxi firms will have to pay for places to store their cars overnight. And they’ll have to carry surplus of all of those things, because they need to have enough cars and storage and people to handle the peak demand, some of which will go unused during average demand (which Uber doesn’t have to worry about much).
Given that, Robotaxis aren’t likely to expand the market that much, ‘cause they’re not likely to be all that much more efficient than human-driven services. Uber was able to dramatically expand taxi usage not because they were especially more efficient, but mostly because they simply broke all the rules that comprised the artificial regulatory restraints on taxi capacity. No such artificial restraints exist for car ownership, or even for taxi rides any more. There’s no artificial constraints for Robotaxis to break.
So if you want to know why Robotaxis won’t be a big market, it’s because all taxi services (taxi and Uber/Lyft) aren’t all that big a market - and Robotaxis aren’t likely to reduce costs by enough to effect any real dramatic changes in that. Again, for a certain value of “big” - and compared to a company like Tesla or Google, a $100 billion market cap business just isn’t all that "big.”
I believe Khosrowshahi is correct. But I think the Uber model (as opposed to the Tesla model of fleet operations) is where the expansion is really going to be, basically for all the reasons you mentioned: Uber robotaxis can serve peak times at no additional marginal cost, don’t require paid parking spaces, maintence and cleaning is performed basically for free by the owner, and so on.
That said, this week Musk said they will sell the cybercab to the public starting this year. I personally don’t believe this will happen, but maybe they are hoping to support an Uber-style model.
At least two have posted such thoughts on this very board.
Despite your inappropriate quotes, that’s not what I said.
I disagree. It was, and is, the convenience factor. Medallions only applied in a handful of dense cities. Uber worked in not just small cities and towns, but suburbs. I was one of the early day Uber doubters because I thought “well, I could just call a taxi service on my cell” or “taxi providers will just make a web page.” But, that was and is wrong. Uber didn’t win because of cost, they won because of convenience and then network effects.
Back to my original point about expanding the TAM, Uber definitely expanded the TAM - You even admitted as much.
Robotaxis will do the same. And just as you admit the economics of Uber helped Uber take over and expand the TAM, the economics of robotaxis will make those take over and expand the TAM even further.
Despite you talking about it first, I made and make no claim “it will be the end of privately owned vehicles.” It’s telling to me that you talk about that first, yet that wasn’t my question nor argument.
However, since you brought it up, I do believe robotaxis will reduce (not eliminate) personal car ownership. I knew execs that took Ubers to work (in suburban areas) because they could get email done in the 30-40 minute drive. When robotaxis are even cheaper, more people will do that. They may keep the family car, or the haul the kids to soccer practice car, but for commuting, many (not all) will choose the lower cost, do something else while being driven, mode of commuting.
Horseless carriages didn’t eliminate personal horse ownership. 
The Captain
And Uber didn’t reduce personal car ownership.
There was a time when traffic engineers and urban planners genuinely thought it would. They saw adoption trends for ride share services, and looked at usage patterns for young people, and thought that would translate into a material reduction in auto ownership for personal use. I went to quite a few panels back in the day about how cities needed to significantly adjust to this new reality.
And then….well, ride share services just didn’t offer a cost-effective alternative to private car ownership. They started off cheap, subsidized by their equity investors in the mad scramble to gain market share. So in those heady, early days it looked like anything was possible. But by the time they became a mature business, they needed to charge more than their costs. So they didn’t really displace much auto ownership for any but the lightest users.
Assuming that ride hailing competes only with personal car ownership is a mistake or wishful thinking. Now that I no longer have a car I can see/feel the drawbacks of many public transport services. Metro good if near you. All else not so good, specially on weekends. During rush-hours they all are overcrowded.
The Captain
I never said “only.” Rather, I noted that there were quite a few transportation professionals who thought ride hailing would seriously compete with personal car ownership. It turned out not to. Because ride hailing ended up ultimately costing more than those early, heady days of subsidized losses on operating expenses suggested.
I think the same may end up being true of robotaxis, at least in the intermediate term. They may not be a “big” market (for a certain value of “big”), because they may not end up being much cheaper than existing ride hailing. There will still be plenty of humans that need to be paid to make the cars operate (remote operators, dispatched “rescue” teams, and the folks who clean and charge the vehicles at night), and fleet robotaxis will face expenses that ride hailers do not (capital costs for the vehicles, real estate costs for overnight storage, and others).
I’m curious how much cheaper you think a Robotaxi will be than an Uber. I guess it depends on which model wins out: either 1) Robotaxi is a big business run by Tesla (or whoever) using company owned assets, which makes them essential a taxicab company with fewer drivers (but not none, as there will have to be some “emergency” dispatch monitoring, as well as cleaning, refueling etc.
Or 2) Robotaxi is decentralized with individuals using their own cars as Robotaxi services during otherwise “off hours.” In that case the individual is responsible for ownership (capital cost), cleaning, refueling, etc. and the parent company does the customer-matching, emergency service center, etc. and takes a piece of the transaction for itself.
[Would I buy a car with real, actual, full self driving? I would. Would I buy one with no steering wheel or pedals? I would not.]
Since Uber is now only sometimes cheaper than a taxi, and is sometimes much more expensive (surge pricing periods), how long will this “price advantage” for Robotaxis last, do you think? Or is there something that will enable them to insure a permanent advantage that way? (I note the lack of the person driving, counterbalanced by the emergency center staff, cleaning costs, etc.)
I haven’t seen that, I have only seen people debate the various sizes of “big”. I know some enthusiasts have said it will almost end private vehicle ownership (that sort of thing gets headlines, of course) and others have said it will be a business but not a world beating unicorn, but I haven’t seen anyone opine that it wouldn’t be a good business at all.
I don’t think this is on the horizon, at all. My justification would be that Tesla has not introduced any functionality into the car OS that would support taxiing services directly, adn there’s no reason not to, if they intended for the fleet to be derived from Teslas alreayd sold “into the wild”. They could have launched a Tesla-only Uber competitor 7 years ago, by putting the Uber functionality into the car OS and letting people become Tesla Taxi drivers. It would have been relatively easy, and likely to compete with Uber and Lyft, because they could put it in all the cars they sold, remotely. But they didn’t. And, I think that indicates they’ve decided against such a plan.
fwiw, you didn’t offer something like:
3) local operators buy fleets of robotaxis from Tesla and operate them under geographic franchise, or other variations. This would create a liability shield between Tesla and the actions of cars, in damaging property and people as well as like the car dealership model, create a local point of presence where states can regulate and litigate. Because robotaxis are going to be subject to a lot of public opinion and that will drive states and municipalities to want a say in the who/what/where of the businesses, their behavior, performance, and local customer and local resident ability to get satisfaction in unfortunate outcomes.
That’s probably not accurate. Having the cars owned by a third party wouldn’t provide any protection from liability for Tesla. After all, Tesla would still be both the developer of the AI driver and the operator of the system - the fact that the vehicles were owned by someone else wouldn’t relieve them of any legal responsibility for damage the cars caused. So too with regulatory control - if the owner is just a passive owner with absolutely no input in the development of the cars or the software, and no role in how they are operated, there’s not much the local government can or would regulate.
IMHO, the main reason that Tesla’s talking about selling Cybercabs is to try to alleviate some of the capital burden. Any robotaxi network that’s big enough to matter to a company the size of Tesla is going to involve a several billion dollars worth of cars. Tesla would love to not have that much capital tied up in cars. They’d rather have third parties own the cars. They’ll have to pay a cost of capital either way, but it will look so much better on their income statement and balance sheet - and they might be able to monetize their investors’ enthusiasm in a way that gets them a lower cost of capital than their other financing channels. It’s basically a sale-leaseback. It doesn’t shift any liability or provide much of a regulatory nexus, but it would make their finances look so much prettier than if they just owned their own cars.
Who?
The flip side of this is when cars become autonomous, you won’t need to take an Uber to get work done on the commute. You’ll be able to use your own AV that’s outfitted with your own work space.
You’re right, I forgot that possibility. Albaby has already noted why it wouldn’t provide any liability shield; in a lawsuit they’re going to name everybody they can, particularly those with the deepest pockets.
So I live in Knoxville, TN, not even one of the 3 largest cities in the not very large state, yet the city has 200,000 pop and the wider county has 800,000. Google hallucinates that it would take between 200 and 1,000 taxis to serve such a city, depending on density and other factors. Add in the county (where I live, about 20 minutes to downtown) and you’re looking at 1,000 to 10,000 cabs - eventually. That’s a dumptruck full of capital, not to mention operationally.
Of course you don’t have to do that all at once, but if you only have a dozen or two you’re going to have disappointed customers when they call and have to wait while the car deadheads back to them. It’s an interesting business proposition, that’s for sure, but maybe not one that’s easy to barrel into. That’s why the original (and now apparently abandoned) idea of “all Teslas ever sold will be equipped” was compelling.
Buy $50,000,000 worth of taxi cabs so you can be in the taxicab business? Maybe not so compelling.
..my nephew in DC sold his car and uses UBER instead..
..a small sample of one, but an eventual trend, imo..
..I don’t think it happens overnight, and no one expects everyone to sell all their cars, but perhaps that second car, the one for commuting, or short trips around town may become unnecessary..
Is he married? Does he have children? Let’s find out how it works when he’s schlepping them to ballet class and band practice and Little League.
Nah. Not in this lifetime. The effect will be super-marginal, at best. There will be a little in dense urban centers where ownership is difficult, but the country is growing much faster in suburbs and exburbs, and that’s where a car is an important component of living.
As I recall from decades ago, it was the car that made suburbs and exurbs possible and desirable.
Google AI:
Your recollection is correct; the widespread adoption of the automobile, combined with postwar housing policies, was the primary driver that made the development of modern suburbs and exurbs possible, desirable, and ultimately, necessary.
While streetcars facilitated the first “inner ring” of suburbs in the late 19th century, the automobile allowed residential development to spread far beyond fixed transit lines, enabling the sprawling, lower-density, and segregated land-use patterns characteristic of post-World War II America.
If I’m not mistaken, the RoboTaxi is a CAR! 
The Captain
Yes it is. It’s a car you have to summon, which may take a long time to arrive. It’s one you can’t leave a baby carrier in, nor any personal effects, and you can’t just throw your stinky clothes in after going to the gym before work, so you get to carry them into the office.
It’s one that might not be available in all weather conditions, or there might not be enough of them when it’s raining (ever try to catch a cab in a rainstorm?) They won’t work for people with special needs, like those with bad back who can’t easily climb into a low slung sedan, and they won’t be cleaned between every fare, so some will have an issue with that.
But in spite of these and other issues, I’m sure it’s going to be a business, even though some cities will object to the deadheading miles contributing to more - not fewer cars on the road. Heck, there are already taxi cabs, and that’s a business, so why not? And there are people who Aaron Rents their TV rather than buying because they can’t afford the price all at once or whatever. But I find it unlikely that Robocabs will be much cheaper than ownership.
You still have the capital cost of the metal, you have to have a facility to clean and charge them. You have to have a central monitoring system for getting them out of trouble when that happens (and it will happen, just as it happens with cars today) and you have mechanical maintenance, whatever that may turn out to be. You have to have service people and you have to have fuel (electrons or whatever) and you have to turn a profit for the owner(s). That seems unlikely to be a lot cheaper than just having your own for most people. (Renting your TV is vastly MORE expensive than buying one, for instance.)
So yes, Robotaxis are cars. They will find a place, I just find it unlikely that they are going to re-place private ownership of automobiles to any significant degree except in a few defined cases (dense urban living, young people without money, etc.)
Perfect is the enemy of the good. Yes, perfection won’t happen.
The Captain
It’s not perfect vs. good. This isn’t a case where there’s an abstract and unattainable “perfect” that’s tempting someone away from doing a “good enough” product. Owning your own car isn’t perfect. It just has a number of benefits that TaaS can’t replicate. Which may keep TaaS from eating into the private ownership market very much.
There are some advantages to TaaS, but the main theory for TaaS expanding materially beyond existing ride-hailing was the expectation (or hope) that it would be materially cheaper. If Tesla had managed to actually create true Level 5 autonomy and have a Tesla Network that was basically “Uber without drivers,” that might have happened. But that model is not happening any time soon. Tesla’s apparently shifted to fleet-owned special-purpose vehicle, which eliminate many of the economic benefits of the Uber model. Waymo and Zoox have always been on the fleet model. Which makes it all but certain that TaaS is going to be more expensive than self-ownership for the overwhelming majority of people for quite a while.