Will at least 10 world cities have a generally available self-driving taxi service by 2025?
Dec 31
Resolved to YES if on 2025-01-01: - 10 cities with more than 300,000 inhabitants - Have a taxi service of fully self-driving cars (no test, backup or remote human driver - passengers are the only humans in the car and they control where the car is going by specifying a destination). - The service is generally available (anyone can register for the service and order rides with no restrictions other than the obvious ones that would be placed on similar human-operated services like Uber and Lyft - for example: need a smartphone, credit card), subject to availability (supply constraints are OK). - It is reasonable to assume that the service is not considered a restricted experiment by the operating company or local authorities (based on publicly available media and reports). Mar 4, 8:27pm: John Carmack is betting quite a lot of real money on a very similar proposition, but 5 years later: https://twitter.com/ID_AA_Carmack/status/1499803694522589187 | https://blog.codinghorror.com/the-2030-self-driving-car-bet/
Get Ṁ1,000 play money
Sort by:

"The service is generally available (anyone can register for the service and order rides with no restrictions other than the obvious ones that would be placed on similar human-operated services like Uber and Lyft - for example: need a smartphone, credit card), subject to availability (supply constraints are OK)."

It might just be me, but I can see at least two – and possibly three – ways of interpreting this answer. I'm assuming it's too late to change it, but I would've framed it as simply cities that have a commercial robotaxi program in place. By definition, "commercial" means officially launched / no longer a restricted experiment.

As an example of why this is a preferable framing: Waymo has had a fully open-to-the-public robotaxi service operating in the Phoenix area since 2020, and in SF since late 2022. However, even "fully open" still entails a few caveats, and possibly the biggest one is a severe supply constraint: it has a grand total of 300 vehicles available at various times. (Since many are Jaguar EVs, they can't operate 24/7.) Uber & Lyft combined could have 10,000 or more drivers in the Phoenix area alone. Also, Arizona regulators approved using robotaxis on Phoenix-area freeways literally just this past week.

In SF, freeways aren't an issue (since it doesn't have any, and you can't hop on the 101 down to Silicon Valley in a Waymo vehicle, at least not yet), but supply's an even bigger problem than in Phoenix: Waymo has over 100,000 people on its waiting list to sign up for SF-area service, given that it only has 250 AVs there.

Anyway, broader point being that the "supply constraints are OK" bit is an oversimplification IMO. Even if 10 cities have generally available robotaxis by next year, demand will far exceed supply in all or nearly all of them. Is something still "generally available" when a given market can absorb 10,000 units but only 1/40th of that number are available?

predicts NO

I am not sure is supply constraints being ok is "fair". If the spirit is that the robo-taxis are replacing human drivers then supply is a big factor. The company could be deliberately constraining supply to make it more like a test, or they could be using small supply + waitlisting to make it appear generally available when in fact the list of passengers is controlled.

  1. It's not specified but based on the current reading restrictions in time of day (e.g. only nighttime driving, or no peak hours) counts as NO correct?

predicts NO

EDIT: removed statement about "Cruise has suspended its operations indefinitely" because a closer reading of the article revealed that it was referring to production of Cruise Origin vehicles, and somehow it was erroneously summarized as the above sentence.

Think the market is underestimating the chilling effect Cruise's implosion will have on the speed of the industry's expansion.

Plug for my related market:


Hmmm.... the only way to do this by 2025 is via end-to-end NN networks (which will not handle the long-tail issue and thus fail) or incredible detailed mapping of the environment (seems like too much work).... thus will not happen.

(If someone was working on some cool way of putting a robust neurosymbolic layer on the self-driving system or even as part of a robust cog arch, I would predict differently.) So.... to happen by 2025 with current projects probably not safe enough to meet criteria above. Thus "NO". (Will it happen by 2030 ?- -yes 100%. Could it happen in 2025 if, e.g., Tesla changed their research program, -- yes, but it will not.)

bought Ṁ100 of YES

Considering that the market value of the self-driving taxi industry in 2030 is estimated to be more than $2 trillion per year, a simple exponential growth model, assuming a constant annual growth rate, can be used to estimate the market value in 2025. For example, assuming a 15% annual growth rate, the market value in 2025 would be approximately $1 trillion. Dividing this market value by 10 cities yields a market size of $100 billion per city.

Many factors influence the ability to support a market of this size, including population, traffic demand, and the level of acceptance of automated driving technology. Larger cities may have higher demand for transportation and greater acceptance of self-driving cabs. In addition, the introduction of self-driving cabs may improve transportation efficiency and increase convenience for citizens. Given these factors, the introduction of self-driving cabs in a large city could increase the size of that city's market and spur the growth of the self-driving cab industry.


Robo-taxi industry could be worth $2 trillion by 2030 (detroitnews.com)

bought Ṁ15 of NO

@WatabeYoshiyuki "The global robotaxi market size is projected to grow from USD 0.4 Billion in 2023 to USD 45.7 Billion by 2030, at a CAGR of 91.8%"

Which would put it around 0.8 billion at the end of 2024. But that includes both L4 and L5, and this market says no backup or remote drivers allowed.


"Considering that the market value of the self-driving taxi industry in 2030 is estimated to be more than $2 trillion per year"

@WatabeYoshiyuki No, that article states that it could be worth $2T by 2030, problem being that it's nearly five years old and hasn't taken into account the many challenges AVs have faced during that time. Hell, there are a few that have taken place since this question was posed, e.g. Cruise running into trouble. Also, a lot of people bought the BS hype about Tesla's Autopilot / Full Self-Driving feature actually functioning by now (at Level 4 automation or higher), though that wasn't specific to robotaxis.

predicts NO

It is possible to go backwards on this: https://www.cbsnews.com/sanfrancisco/news/cruise-llc-driverless-robotaxis-california-dmv-suspends-permits/

bought Ṁ50 of YES

Does the resolution criteria mean YES or NO for geofenced start/destinations (e.g. just local, no freeway, no interstate, or similar restrictions)?

predicts YES

@wustep geofenced inside a city is YES, geofenced inside a neighbourhood or shopping district or something like that is NO.

predicts YES
predicts YES

@LuisPedroCoelho hard for me to answer. I suggest if it's not clear by the time decision has to be made we'll have a healthy discussion. The spirit of this bet is there are enough places where robots replace human-driven taxis. If there are limitations that mean passengers can't get used to relying on robo-taxis instead of regular taxis then it doesn't count. Note than regular human-driven taxis also are mostly available in a limited area, and will not take on an inter-city ride, or even a ride to the deep suburbia. To make this a YES we'll want passengers to get something roughly equivalent from driverless taxis.

predicts YES

@intellectronica Fair enough. I think the resolution might depend on this as Baidu does aim to keep expanding services in different Chinese cities so that their service (Apollo Go) might by itself cover >10 cities by market end, but so far they do not cover the whole city

predicts YES


"Uber provides access to a global and reliable marketplace across mobility, delivery, and freight,” said Dara Khosrowshahi, CEO of Uber. “Fully autonomous driving is quickly becoming part of everyday life, and we're excited to bring Waymo's incredible technology to the Uber platform."

predicts YES



Team at @Cruise is celebrating another small win today: 2,000,000 driverless miles. It was just 3 months ago that we hit 1M miles. Lots of exciting things ahead for the rest of 2023 as we scale our operations and welcome more riders to @cruise!


This is what exponential growth looks like.

bought Ṁ30 of YES

I asked GPT-3.5: "how many cities have more than 300,000 population, in OECD countries" and it gave me 35 answers. It also gave 21 for non-OECD (that has to be wildly low)

Basically, there are a lot of shots on goal here, and the companies have a huge incentive to recoup their investment with scale. I'm sure the cars aren't cheap, but they must have already planned that part of the when they did the first (wildly guessing) 90% of the investment (in R&D, to get them to this point).

predicts YES

https://twitter.com/Cruise/status/1656283003457503232 I think between Cruise and Wayne we're now at 5 cities covered. That's half of the target.

(The tweet)

@intellectronica Not fully automated though.

bought Ṁ10 of NO

@intellectronica Generally available? I know you can’t get one in SF unless you are part of the Beta. How is it in other cities?

predicts YES
predicts YES

https://twitter.com/Cruise/status/1569419525334106112 (Cruise extends its service to two more cities - Phoenix and Austin)

More related questions