"Who would have imagined that demand could be so elastic?" Any New Yorker paying attention. When a taxi medallion was "worth" $1 million, it was pretty obvious that their was some monopoly pricing going on. It will be interesting to see if I can buy a robotaxi for personal use and then rent it out while I am not using it.
Well of course anything is *possible* here, but unless you claim they are going to outlaw driverless services altogether, the political coalition likely to allow them but disallow private sharing just doesn’t make any sense.
It may be said market doesn’t ever happen because of economics and/or liability insurance costs, idk. But very unlikely because of regulation preventing *personal* rentals while allowing corporate fleet ones.
> they could be on the road 20 hours a day, and probably remain usable for 10 years.
Doubt it. I think cars used full time for rideshare/taxi service last 3-5 years. 20 hours on the road per day at just 20 mph average is still ~150,000 miles per year.
They're EVs and ought to handle city driving a lot better than ICE cars. EV batteries are lasting longer than expected too. I've seen some estimates that battery packs could last a million miles.
That would still only be < 7 years, and the data to-date is much more pessimistic than you million-mile number: 200k-400k miles in optimal conditions (compared with 150k under normal consumer driving).
I said I've seen million mile estimates, and I meant just that. You claim otherwise and neither of us has any references. Until EVs have been around longer, no one knows how long batteries really last.
Your assumption that 20 hours a day means driving the full time and every day of the year is naive. Cities have intersections with traffic lights and stop signs. They wait for passengers to get in and out. Demand changes between work days and weekends and holidays.
> The technology took way too long to arrive thanks to government, lawsuits, and paranoia.
I don't think this is true at all! Why do you think this is true? I worked at Nuro and Cruise, and the technology was simply not good enough until around ~2024. That's when Waymo became open to the public (and a year after Cruise had to remove its vehicles from the road).
The main thing keeping AV back right now is that the vehicle cost for something like a Waymo is simply too high to make the economics work compared to Uber, and something cheap like a Tesla is just not safe enough (by safe enough I mean that it is not better than the Xth percentile driver across relevant driving domains, I don't know what percentile Tesla indexes themselves to and reasonable people can disagree).
There have been some high-profile government permit denials in NYC and DC, but that is not the story in the vast majority of the country.
That's not the right comparison. Investment would have been vastly higher if companies knew they could move forward without opposition as soon as the tech worked. You can see this right now. Tesla has Level 4 cars now, they just feel compelled to pretend otherwise.
There is every reason to believe that both of your claims are correct.
I.e. that neither contradicts the other.
I’m pretty sure that Alvaro is correct that until recently the tech was NOT good enough.
I agree with you, however, that “government, lawsuits and paranoia” are likely to cause the mass deployment to be much slower than it would be otherwise.
Waymo is popular here in Mountain View, CA despite the fact that it’s not really any cheaper than Uber, because a segment of people prefer not to interact with the driver.
Once costs go way down and prices go down with them, I have little doubt that adoption will be much higher.
Not a comment on your overall valid point but on your data. What you are forgetting is the huge grey market that existed, Uber didn't come out of nowhere and it killed that more than taxis. Most people that Uber now would have been grey marketing then and hence not captured in your numbers. By grey market I mean anywhere from unlicensed taxis, "personal" drivers, airport shuttles, hotel shuttles, your actually family and friends, driving yourself and parking, etc. The only people that were paying taxis are tourists, businesses (expensed), and adhoc travelers. Your SFO trip even then was comparable to Uber rates now via grey market services then.
The federal government might not regulate it into the ground but lots of loc government will try. The tyranny of local government is often much worse than the tyranny of the far off federal government, a point liberterians often miss
Clear days with optimal road conditions will allow Tesla-like camera based systems. Prices will be low.
Those systems will garage themselves as rain, snow, and low visibility enter the picture. Prices will rise and vehicles with more pricey sensors (Waymo?) will take over.
That said, I'm skeptical that any AV sensor system can handle a squall dropping 6" of snow per hour. Expect to see more driving bans during bad weather, especially from safety-ism neurotics in blue states. If REGULATORS can't go anywhere, no one goes anywhere.
from Grok: Elon Musk’s Tesla Cybercab is a two-seat electric robotaxi built with no steering wheel, pedals, side mirrors, or rear window. It relies on cameras and Tesla’s autonomous software. Production began at Gigafactory Texas in early 2026, and paid public rides started in Austin in September 2026. Musk has said commercial service should reach Florida, Nevada, and other states soon, and California around mid-2027.
The longer-term pitch is private ownership at under $30,000. Owners would use the car themselves, then let it join Tesla’s Robotaxi network in idle hours and earn fares, much like an owner-operated Uber that drives itself. Federal regulators are still reviewing whether a vehicle without manual controls can be sold to the public.
"Who would have imagined that demand could be so elastic?" Any New Yorker paying attention. When a taxi medallion was "worth" $1 million, it was pretty obvious that their was some monopoly pricing going on. It will be interesting to see if I can buy a robotaxi for personal use and then rent it out while I am not using it.
“It will be interesting to see if I can buy a robotaxi for personal use and then rent it out while I am not using it.”
You likely will be able to do so.
And I can imagaine it “paying” in the suburbs or possibly even in L.A.
But I suspect you’d be better off just being a consumer rather than a producer of such services if you live in NYC.
Maybe, might not legally though kind of like how the majority of AirBNBs are illegal.
Well of course anything is *possible* here, but unless you claim they are going to outlaw driverless services altogether, the political coalition likely to allow them but disallow private sharing just doesn’t make any sense.
It may be said market doesn’t ever happen because of economics and/or liability insurance costs, idk. But very unlikely because of regulation preventing *personal* rentals while allowing corporate fleet ones.
In the early 2000s my company required us to take limo’s because taxis cost too much (Chicago). I love seeing a rent seeking monopoly die.
Elon abundifies everything he touches.
A quibble:
> they could be on the road 20 hours a day, and probably remain usable for 10 years.
Doubt it. I think cars used full time for rideshare/taxi service last 3-5 years. 20 hours on the road per day at just 20 mph average is still ~150,000 miles per year.
They're EVs and ought to handle city driving a lot better than ICE cars. EV batteries are lasting longer than expected too. I've seen some estimates that battery packs could last a million miles.
That would still only be < 7 years, and the data to-date is much more pessimistic than you million-mile number: 200k-400k miles in optimal conditions (compared with 150k under normal consumer driving).
I said I've seen million mile estimates, and I meant just that. You claim otherwise and neither of us has any references. Until EVs have been around longer, no one knows how long batteries really last.
Your assumption that 20 hours a day means driving the full time and every day of the year is naive. Cities have intersections with traffic lights and stop signs. They wait for passengers to get in and out. Demand changes between work days and weekends and holidays.
Plus consumables. Most tires are rated ~50k miles and $800-$1200/ set, in addition to cost of other suspension components.
> The technology took way too long to arrive thanks to government, lawsuits, and paranoia.
I don't think this is true at all! Why do you think this is true? I worked at Nuro and Cruise, and the technology was simply not good enough until around ~2024. That's when Waymo became open to the public (and a year after Cruise had to remove its vehicles from the road).
The main thing keeping AV back right now is that the vehicle cost for something like a Waymo is simply too high to make the economics work compared to Uber, and something cheap like a Tesla is just not safe enough (by safe enough I mean that it is not better than the Xth percentile driver across relevant driving domains, I don't know what percentile Tesla indexes themselves to and reasonable people can disagree).
There have been some high-profile government permit denials in NYC and DC, but that is not the story in the vast majority of the country.
That's not the right comparison. Investment would have been vastly higher if companies knew they could move forward without opposition as soon as the tech worked. You can see this right now. Tesla has Level 4 cars now, they just feel compelled to pretend otherwise.
There is every reason to believe that both of your claims are correct.
I.e. that neither contradicts the other.
I’m pretty sure that Alvaro is correct that until recently the tech was NOT good enough.
I agree with you, however, that “government, lawsuits and paranoia” are likely to cause the mass deployment to be much slower than it would be otherwise.
Waymo is popular here in Mountain View, CA despite the fact that it’s not really any cheaper than Uber, because a segment of people prefer not to interact with the driver.
Once costs go way down and prices go down with them, I have little doubt that adoption will be much higher.
The back of the envelope calculation about a self driving Tesla robo taxi paying for itself so quickly is rubbish:
Prices for paid passenger miles will collapse, until investment in more robotaxis pays about the same as any other investment.
Not a comment on your overall valid point but on your data. What you are forgetting is the huge grey market that existed, Uber didn't come out of nowhere and it killed that more than taxis. Most people that Uber now would have been grey marketing then and hence not captured in your numbers. By grey market I mean anywhere from unlicensed taxis, "personal" drivers, airport shuttles, hotel shuttles, your actually family and friends, driving yourself and parking, etc. The only people that were paying taxis are tourists, businesses (expensed), and adhoc travelers. Your SFO trip even then was comparable to Uber rates now via grey market services then.
The federal government might not regulate it into the ground but lots of loc government will try. The tyranny of local government is often much worse than the tyranny of the far off federal government, a point liberterians often miss
There should be tiers of autonomous driving.
Clear days with optimal road conditions will allow Tesla-like camera based systems. Prices will be low.
Those systems will garage themselves as rain, snow, and low visibility enter the picture. Prices will rise and vehicles with more pricey sensors (Waymo?) will take over.
That said, I'm skeptical that any AV sensor system can handle a squall dropping 6" of snow per hour. Expect to see more driving bans during bad weather, especially from safety-ism neurotics in blue states. If REGULATORS can't go anywhere, no one goes anywhere.
If humans can drive in the weather, camera-based systems can too.
I don't think we'll see much of this tiering: the sensors have just become too cheap to bother not having them.
There are
https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-05/Level-of-Automation-052522-tag.pdf
All this seems quite true but ChatGPT estimating passenger miles in ride shares and taxis might be way out? The drop during Covid seems low?
Typo. "You should killed me when you had the chance!” should be "You should have killed me when you had the chance!”
Watch some of Tony Sebas lectures on You tube. For the last 15 years, he has been accurate in his analysis of this topic.
What limits supply of Cybercabs in LR?
from Grok: Elon Musk’s Tesla Cybercab is a two-seat electric robotaxi built with no steering wheel, pedals, side mirrors, or rear window. It relies on cameras and Tesla’s autonomous software. Production began at Gigafactory Texas in early 2026, and paid public rides started in Austin in September 2026. Musk has said commercial service should reach Florida, Nevada, and other states soon, and California around mid-2027.
The longer-term pitch is private ownership at under $30,000. Owners would use the car themselves, then let it join Tesla’s Robotaxi network in idle hours and earn fares, much like an owner-operated Uber that drives itself. Federal regulators are still reviewing whether a vehicle without manual controls can be sold to the public.
The cybercab is trying for even better economics than you're imagining, too. https://www.austinvernon.site/blog/cybercab.html
It's going to be glorious!