Not saying anything about perfection I’m arguing about the size of “big market”.
If the Tesla Robotaxi as originally promoted (“everybody’s Tesla can be a robotaxi!”) then it’s an Uber killer. If it’s “You can buy and manage a taxicab company” then it’s a Yellow Cab killer, maybe. Big difference.
If the dream it “it’s going to replace (most) everybody’s car then it’s a flop. That would be great because you’d own the Uber business, the Ford business, the GM business, the parts business, the service business and everything else, except - it’s not gonna happen. It will affect a few at the margins where the rental business already exists (Hertz, Uber) but it’s not going to replace ownership in any meaningful way.
Since we seem content with data points of one, I also have a nephew. He used to take Uber when he lived in Philadelphia. Now he lives in a suburb in Florida, and he owns a car. That’s the reality; he’s not gonna wait 5-15 minutes every time he wants a ride somewhere. (Humorous side note: he occasionally drives for Uber/Lyft. Says the best money is on Friday and Saturday night, people getting out of bars who don’t want to drive home. Good for tips, not exactly a unicorn market, I don’t think.)
That’s actually one of the big reasons why fleet-owned TaaS is going to be more limited. The difference between peak usage and average usage. If you’re going to meet peak demand, you need a lot more cars on the road during those peak hours than you do at most other times.
Uber (and the theoretical Tesla Network) solve that problem by letting people just utilize otherwise dormant existing capacity. There’s no extra marginal cost. You’ve got a car in your driveway anyway. When demand spikes on the weekends, you just use the car you already have. The size of the fleet can expand or contract to meet peak/ebb times.
Fleet-owned networks can’t do that. They can’t right-size the fleet to meet weekly or daily peaks. They have to carry a higher capital cost (and all the costs that scale with fleet size), because they will have to have more cars than they need for average (much less low) demand, so that they can service some part of the extra peak demand. Otherwise they won’t be much of an alternative to conventional ride-hailing.
The abandonment of the Tesla Network dream, at least for now, fundamentally changes the Robotaxi economics. It’s just not going to be a “big” market, where “big” is relative to the size of Tesla.
You are rattling off a long list of problems that will prevent RoboTaxis from ever having a “big market.” Reminds me of the time when there would never be a “big market” for personal computers, what would one use then for?
What impossible things has Tesla managed to do?
Google AI:
Tesla has accomplished several feats in the automotive and energy industries that were widely considered impossible, or at least highly improbable, by industry experts, analysts, and legacy automakers.
Here are the key “impossible” things Tesla has managed to do:
Making Electric Vehicles (EVs) Desirable and High-Performance: Before the Model S (2012), EVs were largely perceived as slow, impractical, golf-cart-like vehicles with minimal range. Tesla proved that electric cars could be faster, more luxurious, and more desirable than internal combustion engine (ICE) cars, completely changing the perception of EV technology.
Building a Global Supercharger Network from Scratch: Tesla built the world’s largest and most reliable fast-charging network from zero. This was a massive infrastructure undertaking that competitors are now only just beginning to adopt as a standard.
Surviving as a New American Car Company: No new American car company had survived and thrived for decades. Analysts and industry experts frequently placed Tesla on a “death watch,” expecting it to go bankrupt, especially during the “production hell” phase of the Model 3.
Mass Producing the Model 3: When Tesla was scaling production of the Model 3, it was deemed impossible for them to produce hundreds of thousands of units a year. Despite massive manufacturing bottlenecks and, at one point, having to build a production line in a tent, Tesla achieved high-volume production, making the Model 3 one of the best-selling cars in the world.
Vertical Integration and Software-Defined Vehicles: Tesla broke the traditional automotive supply chain model by developing much of its technology in-house, including software, chips, and battery packs. This allowed them to implement over-the-air (OTA) updates, enabling them to improve car performance, range, and features after purchase—a feat traditional manufacturers initially found impossible to replicate.
Rapidly Building Gigafactories Globally: Tesla built Gigafactory 3 in Shanghai in record time (under a year), a pace of construction and scaling previously considered impossible for a foreign automaker in China.
Proving the Viability of Large-Scale Battery Storage: Tesla successfully developed and deployed the world’s largest lithium-ion battery in South Australia, fulfilling a promise to do so in less than 100 days—another feat initially deemed impossible.
Most things don’t end up having a big market, for any value of that term that would actually be “big” compared to Tesla or Google. There wasn’t a “big market” for Segways, or VR goggles, or 3D printers, or 3D televisions, or “the Metaverse,” or computer glasses (like Glass), or NFT’s, or the like. Because it turns out that there were problems with all of those things that kept them from being widespread revolutionary consumer products. Most things that people propose to be massive, revolutionary “big market” prospects end up failing. So cherry-picking the literal biggest foundational technology and suggesting because one product managed to be that “big market” over the objection of naysayers, it means that naysayers must be wrong, is a pretty big error.
But even then, it’s worth pointing out that it took years for personal computers to actually become a “big market,” and it didn’t happen until there was a decent answer for “what would one use them for.” The first personal computers were available in the early 1970’s, but sales were quite modest for most of the decade, and even then mostly were going to hobbyists. It wasn’t really until the release of Visicalc - the first “killer app” for personal computers - that the market really became a “big market.”
Robotaxis are likely to be the same way. Unless and until someone comes up with a true Level 5 product (or something that’s Level 4 without much support almost everywhere), they’re not going to have the technological capability to be a “big market” relative to Tesla or Google. So when you ask AI:
…the thing they haven’t managed to do is get FSD to work at Level 5. Which means that Robotaxis can’t adopt the Uber capital and fleet management structure. Which means that Robotaxis are not going to be cheap enough to be meaningful competition for self-owned cars.
Or get hyperloop to change the face of transportation. Or make the Boring company something that, uh, will change how we travel from city to city. Or make Twitter the “everything app” company that everybody wants. Or put solar roofs on everybody’s house. Or make the Dojo supercomputer a thing. Or put a human on Mars by 2022. Or, I point out, save “trillions” from the Federal Budget.
The fact that he has had some spectacular successes (and he has, surely) does not mean that each of his “visions” will come true; some have been spectacular failures. (Twitter comes to mind, now worth about 20% of what he paid.) This is one that might have if the original prediction had been accurate: that every Tesla ever made could become a Robotaxi in its spare time. Since even he has given up on that and continues to grope for a business model, I’d say that resting the argument on the fact that “he’s been successful” is pretty thin gruel.
Interesting that all of the discussion about FSD seems to be in the context of making money, but don’t you think that Tesla might have a marketing edge if one could tell the car “take Junior over to Joe’s” and it would just do it?
That doesn’t seem likely at any point in the near future with the recent pivot to robotaxis. Certainly no personally-owned Tesla on the road today will ever do that - absent both a software and a hardware upgrade.
Before the CyberCab the talk was about using any Tesla as a taxi. Why would the CyberCab exclude that option? Has there been any announcement to that effect?
I’m not sure what you mean by that. A marketing edge in what context? You can already put minors (13 and up) in an Uber or a Lyft and have them taken over to Joe’s.
A robotaxi might be even less likely to offer that kind of service, because the minor would be unsupervised, which causes other kinds of problems. Waymo only allows unsupervised minors (14 and up) in one of their ten markets (Phoenix), so there might be some hesitancy there.
If Tesla were to actually have pure Level 5 autonomy, so that cars could drive themselves outside of a Robotaxi service? Sure, being able to send an unaccompanied minor on a trip might be an edge over other car companies. But we’re many years away from that, and I think it would be dwarfed by the massive fact that the Tesla car could drive itself generally, not just that one specific use case.
That assumes we’re talking about a scenario where Tesla’s car can do something that other ridehail services or car companies cannot (there obviously wouldn’t be an “edge” to doing something most other players can do). I don’t think that’s at all likely, but that’s a topic for another thread.
Because there’s no need to build a CyberCab if it’s possible to use any Tesla as a taxi. In fact, it’s counterproductive - there’s no way a CyberCab could ever be as cheap as a Model 3 with a few years of depreciation under its (fan) belt. The fact they’re building CyberCabs makes it pretty unlikely that there will be a Tesla Network using cars on the road today.
Plus, Tesla has seemed to have stopped talking about a Tesla Network altogether. When was the last time they discussed the possibility of people putting their own personal cars in service? It’s been years.
Taken together with the removal of Autopilot branding and no longer offering FSD for purchase, and Tesla seems to be taking every possible step to back away from making any representation to current purchasers that they are in any way being promised that their car will ever be able to drive itself.
I was answering the first one. The reason why the CyberCab excludes that option is because it’s an economically foolish thing to make if you could, in fact, “use any Tesla as a taxi.” The fact that they’re making a bespoke, special-purpose vehicle to use for their service now is exceptionally strong evidence that you can’t just use any existing Tesla as a taxi.
Not completely sure what you are asking but it has been shared a few times here that current Teslas lack the physical hardware necessary for level 5 FSD.
The Cybercab is being developed without a steering wheel or any means by which someone can actually drive it (no pedals). Now maybe once the Cybercab is perfected that technology can be added back in but I think there is a zero percent chance that we have personal ownership of a vehicle that lacks any means by which someone can manually drive it or otherwise assume control in the case of an emergency. I think the market for such would be next to zero.
My context was private ownership and the ability to do this with a private car, e.g., one with a car seat installed.
There are two very different questions … can/will FSD/Robotaxi result in a large market (potentially divided among multiple providers) vs where we are on Tesla or anyone else achieving reliable Level 4 or 5 service.
It is quite possible for the market potential to be great, but no one to be close to having the capability, for example.
Man, if the kid is young enough to require a car seat, are you really sending them out completely without adult supervision in an autonomous car? What if the car breaks down, or Joe forgets the kid is coming over and isn’t home, or the kid has some sort of issue in the car and there’s no one there to help them? Maybe there’s a few years between when a kid might be mature enough to travel unaccompanied between the ages of 10/11 and 13 (when Uber and Lyft are available) that you might be able to market as an autonomous advantage, but I can’t imagine that’s enough to make any material difference.
I mean, implicit in all this is that we’re talking about a world where Tesla has an L5 or near-universal L4 car and others don’t yet. Right? In that timeline, just having the only L5/L4+ car is such a huge marketing sell (“You can watch a movie on the way to work”) that this specific, very limited scenario is probably not that big a deal.
