Appears you have never actually experienced a Tesla with FSD. I have FSD in my Tesla S, and have NEVER been able to use it to actually complete a trip. It might manage a few blocks, or a few miles on the interstate, but it inevitably encounters something which confuses it and I must take over to avoid an accident. The fundamental problem is that Elon insisted on doing it without LIDAR, so it simply will never perform as well as Waymo. I'd sell that Tesla stock if I were you.
While that is better than not being able to do so, it isn't how to judge Tesla's closeness to L4 autonomy. Focus on how each vendor's approach handles the hard problems. Most serious accidents occur because of a combination of two or more factors. Similar to airplane safety validation, the winning solutions will have high redundancy, graceful degradation, successful risk minimization. Saving a few bucks by only having one sensor type (visual cameras) is the OPPOSITE of a long-term successful approach. Ignore those annecdotes about successful trips in mild condiions. Ignore the "better safety record than humans" (which is a low bar).
I find that hard to believe. And am amazed you hit 100 miles with no intervention. Was it in a city or on a highway? If a highway, and no construction or orange cones, I believe it. Mine can't get me more than a few blocks.
I tried again yesterday. The FSD stopped repeatedly because it did not think I was putting steady pressure on the wheel, so we can ignore that because that would not be an issue if engineered for no driver. But then it slowed to a stop, approaching a traffic circle with a simple bear right, with construction cones and police on the right, and would not proceed at all, so I took over. I could not see any reason to have stopped.
Elon opted out of LIDAR because it used to cost $20k to equip a vehicle with LIDAR. But the Chinese have now reduced the cost of LIDAR down to $200 per vehicle.
What version are you using? V14 has been a game changer. Look at Youtube for regular people letting their family members sit in the drivers seat for the first time to supervise it. It's shockingly good now.
HW3 FSD works acceptably (minus my major gripe on the speed profiles) on interstates with 2 lanes each way. I used it a fair amount there until the V12 release removed my ability to set the speed (on the trips I do, I want to cap the speed at 70 unless I touch the throttle).
It may work well on wider roads, but so many of those cases it doesn't pick the right (in my book) lane, so I only let it run there if traffic was really light.
I81, I95 (including some beltways), I78, I76, I70, I68 areas for reference.
I've heard that Lidar is better. It's also expensive, which is why Waymo has to charge more for rides. Elon says he's solved the problem with Tesla's FSD, so we'll see...
This isn't about "better"; its about "sensor redundancy", especially in varied weather conditions. It is physically impossible for a single sensor type to be competitive with multi-sensor solutions, under all conditions.
OTOH, so far the only proven way to achieve L4 autonomy under a wide range of conditions is hi-definition mapping of lanes. (So if lane markings are obscured by weather or the camera is blinded by setting sun, car still knows where the lane is expected to be.) What Waymo has done in limited areas. If Tesla is collecting the right information from all the cars driving around, they could be the first to have that on a large scale. Perhaps that is even more valuable than multiple sensor types. If so, they could dominate, if they make use of that before any other vendor gets close to their data volume. Lidar becomes an option they could have, like all-weather tires, not a necessity. As cost drops, it gradually becomes commonplace. Without disrupting their story.
"As a result of all of this, I decided to buy some Tesla stock as I was writing this article over the last couple of weeks"
I'm sure you know this and but plenty of people lose lots of money because they don't - "This seems like a good business plan" isn't really the right way to think about buying stock. A better question is "Is this company worth 1.5 trillion dollars - as much as every other car company in the world combined? 15x Waymo? Do I really think it will have higher profits than all these companies combined in the foreseeable future?"
Of course, the best question is "Can I buy this on the way up and find someone else dumb to buy it off me for more before it crashes?" But I prefer to at least pretend most people aren't just treating stocks as a casino.
If you read in that other article, you'll see that the valuation for Tesla is basically assuming it will have a big share of the global robotaxi market, which in turn will eat most of the current taxi and ride-hailing market, and then create more, including cannibalizing some car ownership.
I don't think humanoid robots are accounted for in that valuation
Then you need to combine this with the idea that "the market has not priced all that in", which is obvious when you hear all the people who don't believe that Tesla will pull this off.
So that's kind of the insight: "The market is humongous, Tesla is geared to have a near complete monopoly, and people don't realize this yet."
"Tesla is geared to have a near complete monopoly" -- and the day it becomes obvious that won't happen, the stock tanks irreversibly. I do believe Tesla will be a significant player. But there are two problems: (1) weather (cameras only? not as safe as competitors) means Tesla will eventually have to make a fundamental change, and add additional sensors for redundancy. Their success in (Supervised) mode has little to do with requirements for L4 autonomy. (2) By the time they get this right, Chinese models will have swept the world.
I was wondering the same thing. Tesla’s current P/E ratio of around 300, compared to Volkswagen’s P/E of 7, suggests that a massive amount of future potential is already 'baked into' the stock price. If this valuation isn't considered too high, what is?
For Tesla to justify this premium, several things must align: the technology has to work flawlessly, regulatory hurdles must be cleared, and the company must maintain a durable economic moat against growing competition. Furthermore, we shouldn't overlook the 'Musk factor'—a small but significant group of consumers, myself included, is becoming increasingly reluctant to support his brand. Plus its sales numbers and market share is declining.
Agreed. These are the reasons why the price is what it is.
Many people think like you, and many think like me.
If I'm right, 300 P/E is nothing because it's based on a legacy business that is very low margin.
If you're right and Tesla doesn't deliver, the P/E is ridiculously high.
But just note that that P/E refers to a different business. Here the bet is not that Tesla can make a lot of money with its current cars by selling them to customers.
While one could argue that even with a less than 50% probability of success, the potential upside is exponential, making the valuation justifiable. This aligns with Nassim Taleb’s Barbell Strategy, where one balances extreme risk-aversion with high-upside speculative bets to capture outsized returns.
However, I have significant reservations based on the following three pillars:
1. Ethical Concerns and Geopolitical Friction
I find it difficult to reconcile the ethics of supporting a company under Musk's leadership. Furthermore, his long-term success relies heavily on government cooperation for implementation. By openly supporting authoritarian regimes and frequently spreading misinformation, he is increasingly alienating regulators outside of the U.S. I am skeptical that international governments will be willing to grant him the level of power and infrastructure access his ventures require.
2. Market Saturation and Margin Erosion
There is no guarantee that this sector will result in a high-margin monopoly or duopoly. If we look at the trajectory of the electric vehicle market, we see that as competition increases, profit margins inevitably compress. If the market remains fragmented and competitive, the current share price—predicated on total dominance—becomes very difficult to justify.
3. Execution vs. Regulation
Even if the technology succeeds, the "moat" may be thinner than investors realize. A company can have a head start, but if it lacks the diplomatic capital to navigate global bureaucracy, its growth will hit a ceiling that no amount of engineering can overcome.
I'm not surprised that some people would rather not have a human driver. It's the same reason why people in airplanes no longer talk to or even greet their seatmates -- just get out their phones as soon as they sit down. But when I'm traveling, often the taxi or Uber drivers are the only local residents I get to meet. From them I find out what it's like to live in that city.
I'm with you, and that's why there will always be taxis and similar services. Robotaxis won't ever hit 100% market share. But it is undeniable they will get the majority.
I agree. Irreplaceable for that. My guess is those 2 services will be unbundled and guides will more formally appear. Or maybe robotaxis will act as guides?
You write some interesting stuff, but in my view you are clearly poorly informed regarding Tesla. Delusional really. They are no where close to full autonomy, still stuck at Level 2 or 3, despite stating year after year they would reach full autonomy. I doubt that ever happens as the decision to operate camera based, rather than LIDAR, is a fatal flaw.
From an auto industry analyst:
Tesla can’t run a driverless ride-hailing service anywhere without submitting data that proves FSD adheres to the SAE’s definition of Level 4 autonomy. This includes autonomous miles driven (with Tesla staff behind the wheel, not Tesla customers), number of crashes, and number of disengagements.
Tesla has been asked for this data by the California DMV for 11 years, but has refused to submit it, saying that FSD is only a Level 2 driver-assist system (like your average cruise control). It’s well-known that what Tesla says about FSD to regulators versus what they advertise is like day and night, which is why there are so many lawsuits against Tesla’s Autopilot and FSD.
The CA DMV itself is suing Tesla for consumer fraud due to customers having been duped into thinking their Teslas would be self-driving via Tesla’s naming of its driver-assist system (“Autopilot” and “Full Self-Driving”) and false advertising. The ruling and sanctions the CA DMV can seek from Tesla (if the DMV wins) will be made public by December 22nd, at the latest. ...
Last I looked the total number of driverless rides by Tesla in Austin was......7. They launched as an actual taxi with a "safety driver" on board. Evidently, the number of cyber cabs operating in Austin has fallen dramatically; it was largely a PR stunt
Unsupervised Driving will happen, it's a matter of when. That's why I quoted what the market thinks. If you're so sure about what you say, you should bet your money against Tesla releasing it by March 31st! You can double your money!
Why do you think unsupervised driving will happen for Tesla? Why now? Why HASN'T it happened after all these years? They had a huge head start and now have lost the race to Waymo and others. You might find this interesting:
Also, you really want to be paying attention to China. The CCP has made autonomous cars a national priority and they have a number of companies at level 3 and 4 already.
I think you can hate Musk for what he says and still recognize that he's added more value to the world through Tesla, SpaceX, the Boring Company, and Paypal, than virtually any other human being.
Tomas, why are you so down on Waymo? Here in Scottsdale, we’ve been riding Waymo‘s for quite a while. Everybody has the app and that’s all everybody talks about as far as driverless taxis. I don’t know anybody that’s taken a cyber cab.
the experience of taking a waymo vs. uber home at the end of a night of drinking is very different. with the uber you never know if the person is going to want to talk or whatever and you just want to relax and not think about anything.
I found the 'Peak vs Baseload' argument fascinating, especially when applied to aging societies like Japan.
In the US, the driver is a 'cost' to be eliminated. But here in Japan, the driver is a 'resource' that has already depleted (average taxi driver age is over 60!).
This changes the strategic map: Robotaxis aren't just a luxury upgrade here; they are a survival mechanism for our transport grid.
I suspect Tesla's 'Airbnb model' (utilizing existing idle cars) might be the only scalable solution for rural Japan, where deploying expensive dedicated fleets (like Waymo) makes no economic sense. The geography of demographics might decide the winner.
I would be curious to see metrics on the percentage of passenger miles from ride-sharing per metro region. My guess is that it is tiny now, but likely to grow fast.
When Uber started, there was no tipping. People whined, Uber capitulated, added the option. With Waymo there is no tipping. People would rather take a Waymo then Uber +15%. Thus those kindhearted folk who wanted tipping helped the humans lose their earnings.
Great framing. Isn’t # of vehicles a growth constraint for Waymo? They only have a few thousand vehicles in operation I thought. Can they ramp manufacturing fast enough to support exponential or accelerated growth?
If waymo can only make a few thousand cars per year and tesla can make a million per year, that’s a big advantage for cybercab.
Correct, they can’t ramp up as fast as Tesla. Thats the key part. Right now they have one Hyundai manufacturing plant producing them, but only one. Tesla has many more, and as mentioned their production will be much faster. That’s the biggest of Tesla’s advantages.
How is it an advantage if they have nothing to ramp up for? Meaning, they have no market where their car is approved to operate. Musk has been saying for years that there will be millions of robotaxis on the road w/i the next year:
Meanwhile, the competition grows. It seems foolish to bet against Google at this point, they are enormously profitable and clearly focused on driving. Hard to imagine they won't drive costs significantly lower or that they would have any problem at all dramatically increasing car production. The majority of tesla's profits have come, not from selling cars, but from selling emissions credits. Their only new model, the cyber truck, is an historic flop. Tesla's total sales have fallen 2 years in a year now and they might find it very difficult to afford developing new models. But there's another question you aren't considering; since when is operating a taxi business a fantastic, extremely profitable business? One that can justify the incredibly high multiple that Tesla trades at?
An Uber driver's average income in the U.S. often falls between $15 to $25 per hour (gross). Gross! So before expenses like gas.
To be clear, I'm a big fan of autonomous vehicles, I just think you are way off on Tesla.
Google has only a few thousand in their fleet because of the higher expense. They don't manufacture the cars; they buy them. Then they add the waymo lidar technology. However, as Tomas said, " when Waymo moved from Jaguar to Hyundai, the cost of each of its cars dropped, potentially from ~$150k to ~$70k, because instead of having to be manually retrofitted, they started leaving the manufacturing plant as Waymos."
Not sure I agree. Apollo Go has reportedly signed a partnership with Lyft to launch operations in the UK and Germany in 2026, which suggests they could enter Europe sooner than some U.S. players.
(If that happens, it also says a lot about how Europe has increasingly become a secondary battleground rather than the main arena for innovation — but that’s another story.)
I'm not so bullish about Tesla. All the bullish links/quotes come from Tesla themselves or directly from Elon Musk. Both have a long history of over-promising and under-delivering, both generally, and specifically regarding self-driving. The link you posted to an 'outside' site is remarking (with some caveats) that Teslas seem to be crashing more frequently than teenagers fresh out of drive school. "It has always gotten better" is not a proof that it always will, and an exponential increase can just be the middle part of an S-curve where we don't know how high the plateau is. So I am still skeptical that Tesla's 'optical-only' strategy can ever compete safety-wise with a Lidar-based robocar. Also, simply assuming that if the driving model actually does reach that level that it can be retrofitted to all Teslas with a simple software update is highly optimistic. AI models are very process-intensive, so the likelihood is that at some point of model complexity, they can only be uploaded to newer models.
Medium term, Tesla will still probably be OK because it's combination of safety and price will become 'good enough' compared to a human-driven taxi, and however fast Waymo expands there will still be space in the market for Tesla in what will effectively be a duopoly in the US.
Long-term I guess Chinese robocars will take over everywhere in the world that political interests don't block them out.
Thanks for the article. A couple of thoughts, 1)Not sure Waymo (Google) wants to be in the taxi business long term. I could see them selling Waymo and just licensing the self-driving AI to car manufacturers. Already in a deal with Toyota. 2) How many Tesla owners will put their car on the street for hire? Many are pretty well-to-do, will they be willing to put up with the hassle for the money they will get? 3) Lots of aspects besides building cars to running a large-scale taxi service, Waymo has done that; Tesla has not. Reason for Tesla to do a deal with Uber (Waymo already has a deal with Uber) 4) I would anticipate many car manufacturers will get into this market (offset to lower car sales, repeatable revenue stream, scale for manufacturing). Tesla's car business has declined partly due to the additional competition, and maybe the same dynamic for this market.
1. I think with self-driving cars, most of the market will be shared cars because it will be dramatically cheaper per mile. Owning a self-driving car is more of a luxury at that point
2. We'll see. $500 per month is $6k per year. Maybe what happens is that poorer people can now afford it.
3. True but it can't be that difficult, there are dozens of companies who have developed the tech
4. Unsupervised driving is extreeeeeeemely hard, and most don't even have the data, forget about the expertise.
Called a Uber once in Brussels. The car I got was a local taxi driver that didn't want to settle the fee via Uber . I got the ride for the same fee but without Uber getting any money. It struck me as odd that local taxi drivers are represented on Uber.
Appears you have never actually experienced a Tesla with FSD. I have FSD in my Tesla S, and have NEVER been able to use it to actually complete a trip. It might manage a few blocks, or a few miles on the interstate, but it inevitably encounters something which confuses it and I must take over to avoid an accident. The fundamental problem is that Elon insisted on doing it without LIDAR, so it simply will never perform as well as Waymo. I'd sell that Tesla stock if I were you.
Thanks for your advice!
I don’t own one but have experiences FSD (100 miles with no intervention).
Did you hear that somebody just crossed the US with FSD and no intervention?
While that is better than not being able to do so, it isn't how to judge Tesla's closeness to L4 autonomy. Focus on how each vendor's approach handles the hard problems. Most serious accidents occur because of a combination of two or more factors. Similar to airplane safety validation, the winning solutions will have high redundancy, graceful degradation, successful risk minimization. Saving a few bucks by only having one sensor type (visual cameras) is the OPPOSITE of a long-term successful approach. Ignore those annecdotes about successful trips in mild condiions. Ignore the "better safety record than humans" (which is a low bar).
I assume it was a new Tesla?
It wasn't mine, but this was 2 years ago
I find that hard to believe. And am amazed you hit 100 miles with no intervention. Was it in a city or on a highway? If a highway, and no construction or orange cones, I believe it. Mine can't get me more than a few blocks.
It was in France, going into Paris, and there was lots of traffic.
I tried again yesterday. The FSD stopped repeatedly because it did not think I was putting steady pressure on the wheel, so we can ignore that because that would not be an issue if engineered for no driver. But then it slowed to a stop, approaching a traffic circle with a simple bear right, with construction cones and police on the right, and would not proceed at all, so I took over. I could not see any reason to have stopped.
Another FAIL.
Elon opted out of LIDAR because it used to cost $20k to equip a vehicle with LIDAR. But the Chinese have now reduced the cost of LIDAR down to $200 per vehicle.
What version are you using? V14 has been a game changer. Look at Youtube for regular people letting their family members sit in the drivers seat for the first time to supervise it. It's shockingly good now.
I will check to see if I have v14, thanks.
I have v12. So will try to upgrade, thanks.
HW3 FSD works acceptably (minus my major gripe on the speed profiles) on interstates with 2 lanes each way. I used it a fair amount there until the V12 release removed my ability to set the speed (on the trips I do, I want to cap the speed at 70 unless I touch the throttle).
It may work well on wider roads, but so many of those cases it doesn't pick the right (in my book) lane, so I only let it run there if traffic was really light.
I81, I95 (including some beltways), I78, I76, I70, I68 areas for reference.
Self driving is VERY difficult in snow. Probably solvable through
I've heard that Lidar is better. It's also expensive, which is why Waymo has to charge more for rides. Elon says he's solved the problem with Tesla's FSD, so we'll see...
LiDAR being better is a matter of time.
https://unchartedterritories.tomaspueyo.com/p/robotaxis-are-here
This isn't about "better"; its about "sensor redundancy", especially in varied weather conditions. It is physically impossible for a single sensor type to be competitive with multi-sensor solutions, under all conditions.
OTOH, so far the only proven way to achieve L4 autonomy under a wide range of conditions is hi-definition mapping of lanes. (So if lane markings are obscured by weather or the camera is blinded by setting sun, car still knows where the lane is expected to be.) What Waymo has done in limited areas. If Tesla is collecting the right information from all the cars driving around, they could be the first to have that on a large scale. Perhaps that is even more valuable than multiple sensor types. If so, they could dominate, if they make use of that before any other vendor gets close to their data volume. Lidar becomes an option they could have, like all-weather tires, not a necessity. As cost drops, it gradually becomes commonplace. Without disrupting their story.
"As a result of all of this, I decided to buy some Tesla stock as I was writing this article over the last couple of weeks"
I'm sure you know this and but plenty of people lose lots of money because they don't - "This seems like a good business plan" isn't really the right way to think about buying stock. A better question is "Is this company worth 1.5 trillion dollars - as much as every other car company in the world combined? 15x Waymo? Do I really think it will have higher profits than all these companies combined in the foreseeable future?"
Of course, the best question is "Can I buy this on the way up and find someone else dumb to buy it off me for more before it crashes?" But I prefer to at least pretend most people aren't just treating stocks as a casino.
Thanks!
Yes you're right, this was more a disclosure than a full analysis. It should be used in combination with this:
https://unchartedterritories.tomaspueyo.com/p/robotaxis-are-here
If you read in that other article, you'll see that the valuation for Tesla is basically assuming it will have a big share of the global robotaxi market, which in turn will eat most of the current taxi and ride-hailing market, and then create more, including cannibalizing some car ownership.
I don't think humanoid robots are accounted for in that valuation
Then you need to combine this with the idea that "the market has not priced all that in", which is obvious when you hear all the people who don't believe that Tesla will pull this off.
So that's kind of the insight: "The market is humongous, Tesla is geared to have a near complete monopoly, and people don't realize this yet."
"Tesla is geared to have a near complete monopoly" -- and the day it becomes obvious that won't happen, the stock tanks irreversibly. I do believe Tesla will be a significant player. But there are two problems: (1) weather (cameras only? not as safe as competitors) means Tesla will eventually have to make a fundamental change, and add additional sensors for redundancy. Their success in (Supervised) mode has little to do with requirements for L4 autonomy. (2) By the time they get this right, Chinese models will have swept the world.
I was wondering the same thing. Tesla’s current P/E ratio of around 300, compared to Volkswagen’s P/E of 7, suggests that a massive amount of future potential is already 'baked into' the stock price. If this valuation isn't considered too high, what is?
For Tesla to justify this premium, several things must align: the technology has to work flawlessly, regulatory hurdles must be cleared, and the company must maintain a durable economic moat against growing competition. Furthermore, we shouldn't overlook the 'Musk factor'—a small but significant group of consumers, myself included, is becoming increasingly reluctant to support his brand. Plus its sales numbers and market share is declining.
Agreed. These are the reasons why the price is what it is.
Many people think like you, and many think like me.
If I'm right, 300 P/E is nothing because it's based on a legacy business that is very low margin.
If you're right and Tesla doesn't deliver, the P/E is ridiculously high.
But just note that that P/E refers to a different business. Here the bet is not that Tesla can make a lot of money with its current cars by selling them to customers.
While one could argue that even with a less than 50% probability of success, the potential upside is exponential, making the valuation justifiable. This aligns with Nassim Taleb’s Barbell Strategy, where one balances extreme risk-aversion with high-upside speculative bets to capture outsized returns.
However, I have significant reservations based on the following three pillars:
1. Ethical Concerns and Geopolitical Friction
I find it difficult to reconcile the ethics of supporting a company under Musk's leadership. Furthermore, his long-term success relies heavily on government cooperation for implementation. By openly supporting authoritarian regimes and frequently spreading misinformation, he is increasingly alienating regulators outside of the U.S. I am skeptical that international governments will be willing to grant him the level of power and infrastructure access his ventures require.
2. Market Saturation and Margin Erosion
There is no guarantee that this sector will result in a high-margin monopoly or duopoly. If we look at the trajectory of the electric vehicle market, we see that as competition increases, profit margins inevitably compress. If the market remains fragmented and competitive, the current share price—predicated on total dominance—becomes very difficult to justify.
3. Execution vs. Regulation
Even if the technology succeeds, the "moat" may be thinner than investors realize. A company can have a head start, but if it lacks the diplomatic capital to navigate global bureaucracy, its growth will hit a ceiling that no amount of engineering can overcome.
I'm not surprised that some people would rather not have a human driver. It's the same reason why people in airplanes no longer talk to or even greet their seatmates -- just get out their phones as soon as they sit down. But when I'm traveling, often the taxi or Uber drivers are the only local residents I get to meet. From them I find out what it's like to live in that city.
I'm with you, and that's why there will always be taxis and similar services. Robotaxis won't ever hit 100% market share. But it is undeniable they will get the majority.
I agree. Irreplaceable for that. My guess is those 2 services will be unbundled and guides will more formally appear. Or maybe robotaxis will act as guides?
Random meetings with people not in your bubble. It happens less and less.
You write some interesting stuff, but in my view you are clearly poorly informed regarding Tesla. Delusional really. They are no where close to full autonomy, still stuck at Level 2 or 3, despite stating year after year they would reach full autonomy. I doubt that ever happens as the decision to operate camera based, rather than LIDAR, is a fatal flaw.
From an auto industry analyst:
Tesla can’t run a driverless ride-hailing service anywhere without submitting data that proves FSD adheres to the SAE’s definition of Level 4 autonomy. This includes autonomous miles driven (with Tesla staff behind the wheel, not Tesla customers), number of crashes, and number of disengagements.
Tesla has been asked for this data by the California DMV for 11 years, but has refused to submit it, saying that FSD is only a Level 2 driver-assist system (like your average cruise control). It’s well-known that what Tesla says about FSD to regulators versus what they advertise is like day and night, which is why there are so many lawsuits against Tesla’s Autopilot and FSD.
The CA DMV itself is suing Tesla for consumer fraud due to customers having been duped into thinking their Teslas would be self-driving via Tesla’s naming of its driver-assist system (“Autopilot” and “Full Self-Driving”) and false advertising. The ruling and sanctions the CA DMV can seek from Tesla (if the DMV wins) will be made public by December 22nd, at the latest. ...
Last I looked the total number of driverless rides by Tesla in Austin was......7. They launched as an actual taxi with a "safety driver" on board. Evidently, the number of cyber cabs operating in Austin has fallen dramatically; it was largely a PR stunt
I hear you and the facts you mention are right.
Unsupervised Driving will happen, it's a matter of when. That's why I quoted what the market thinks. If you're so sure about what you say, you should bet your money against Tesla releasing it by March 31st! You can double your money!
Why do you think unsupervised driving will happen for Tesla? Why now? Why HASN'T it happened after all these years? They had a huge head start and now have lost the race to Waymo and others. You might find this interesting:
https://www.cpuc.ca.gov/regulatory-services/licensing/transportation-licensing-and-analysis-branch/autonomous-vehicle-programs/autonomous-vehicle-program-permits-issued
Also, you really want to be paying attention to China. The CCP has made autonomous cars a national priority and they have a number of companies at level 3 and 4 already.
I wouldn’t buy Tesla stock if you paid me to do it. I hate musk and will never contribute to him.
I must be the only person who has never paid for an Uber or anything else. I drive my 21-year-old Mini Cooper everywhere!
Good for you! It's good to live by your beliefs.
I think you can hate Musk for what he says and still recognize that he's added more value to the world through Tesla, SpaceX, the Boring Company, and Paypal, than virtually any other human being.
Anything good he’s done is canceled out by his earnest attempts to destroy the U.S.
Not to mention causing the deaths of potential millions with the disruption and dismantling of USAID
"Attempts"
Tomas, why are you so down on Waymo? Here in Scottsdale, we’ve been riding Waymo‘s for quite a while. Everybody has the app and that’s all everybody talks about as far as driverless taxis. I don’t know anybody that’s taken a cyber cab.
1. They're not ramping up fast enough
2. Their cost per car is high
3. Their manufacturing capability is at least an order of magnitude slower than Tesla's, probably more
OK, but I would think that since they do use LIDAR, then they’re probably more inherently safe than Tesla‘s cameras-only Robo taxis. Correct?
the experience of taking a waymo vs. uber home at the end of a night of drinking is very different. with the uber you never know if the person is going to want to talk or whatever and you just want to relax and not think about anything.
god forbid someone talks to you
i've very social. i talk to many people. i enjoy talking to people.
until i don't
Agreed!
Gosh I should tag you when I quote you
I found the 'Peak vs Baseload' argument fascinating, especially when applied to aging societies like Japan.
In the US, the driver is a 'cost' to be eliminated. But here in Japan, the driver is a 'resource' that has already depleted (average taxi driver age is over 60!).
This changes the strategic map: Robotaxis aren't just a luxury upgrade here; they are a survival mechanism for our transport grid.
I suspect Tesla's 'Airbnb model' (utilizing existing idle cars) might be the only scalable solution for rural Japan, where deploying expensive dedicated fleets (like Waymo) makes no economic sense. The geography of demographics might decide the winner.
Fascinating! Thanks for sharing.
I would be curious to see metrics on the percentage of passenger miles from ride-sharing per metro region. My guess is that it is tiny now, but likely to grow fast.
When Uber started, there was no tipping. People whined, Uber capitulated, added the option. With Waymo there is no tipping. People would rather take a Waymo then Uber +15%. Thus those kindhearted folk who wanted tipping helped the humans lose their earnings.
Very astute observation!
Money talks during the fakeflation-post- ScumDemonic-world.
i prefer no tipping. less thinking. i tip of course now that it's the norm, but i would rather just pay more and not tip.
i have taken robototaxi a few times. it drives more like a normal human i'd say...
I agree. I think it was a competitive thing? Lyft had it so drivers used it more so Uber was forced to add it too.
Great framing. Isn’t # of vehicles a growth constraint for Waymo? They only have a few thousand vehicles in operation I thought. Can they ramp manufacturing fast enough to support exponential or accelerated growth?
If waymo can only make a few thousand cars per year and tesla can make a million per year, that’s a big advantage for cybercab.
Correct, they can’t ramp up as fast as Tesla. Thats the key part. Right now they have one Hyundai manufacturing plant producing them, but only one. Tesla has many more, and as mentioned their production will be much faster. That’s the biggest of Tesla’s advantages.
How is it an advantage if they have nothing to ramp up for? Meaning, they have no market where their car is approved to operate. Musk has been saying for years that there will be millions of robotaxis on the road w/i the next year:
https://www.fastcompany.com/90677822/elon-musks-tesla-robotaxi-promise-typifies-self-driving-overexuberance
Meanwhile, the competition grows. It seems foolish to bet against Google at this point, they are enormously profitable and clearly focused on driving. Hard to imagine they won't drive costs significantly lower or that they would have any problem at all dramatically increasing car production. The majority of tesla's profits have come, not from selling cars, but from selling emissions credits. Their only new model, the cyber truck, is an historic flop. Tesla's total sales have fallen 2 years in a year now and they might find it very difficult to afford developing new models. But there's another question you aren't considering; since when is operating a taxi business a fantastic, extremely profitable business? One that can justify the incredibly high multiple that Tesla trades at?
An Uber driver's average income in the U.S. often falls between $15 to $25 per hour (gross). Gross! So before expenses like gas.
To be clear, I'm a big fan of autonomous vehicles, I just think you are way off on Tesla.
Google has only a few thousand in their fleet because of the higher expense. They don't manufacture the cars; they buy them. Then they add the waymo lidar technology. However, as Tomas said, " when Waymo moved from Jaguar to Hyundai, the cost of each of its cars dropped, potentially from ~$150k to ~$70k, because instead of having to be manually retrofitted, they started leaving the manufacturing plant as Waymos."
It would be interesting to include Chinese companies in the analysis. Their capacity of scaling up and cutting price is huge.
Companies such as Apollo Go or WeRide will be real contenders
Yeah but I doubt they will be accepted with open arms in the West.
Not sure I agree. Apollo Go has reportedly signed a partnership with Lyft to launch operations in the UK and Germany in 2026, which suggests they could enter Europe sooner than some U.S. players.
(If that happens, it also says a lot about how Europe has increasingly become a secondary battleground rather than the main arena for innovation — but that’s another story.)
I'm not so bullish about Tesla. All the bullish links/quotes come from Tesla themselves or directly from Elon Musk. Both have a long history of over-promising and under-delivering, both generally, and specifically regarding self-driving. The link you posted to an 'outside' site is remarking (with some caveats) that Teslas seem to be crashing more frequently than teenagers fresh out of drive school. "It has always gotten better" is not a proof that it always will, and an exponential increase can just be the middle part of an S-curve where we don't know how high the plateau is. So I am still skeptical that Tesla's 'optical-only' strategy can ever compete safety-wise with a Lidar-based robocar. Also, simply assuming that if the driving model actually does reach that level that it can be retrofitted to all Teslas with a simple software update is highly optimistic. AI models are very process-intensive, so the likelihood is that at some point of model complexity, they can only be uploaded to newer models.
Medium term, Tesla will still probably be OK because it's combination of safety and price will become 'good enough' compared to a human-driven taxi, and however fast Waymo expands there will still be space in the market for Tesla in what will effectively be a duopoly in the US.
Long-term I guess Chinese robocars will take over everywhere in the world that political interests don't block them out.
Not all! Look at what the market says! That's to me the biggest tell. ~50% chance that unsupervised learning is released within 3 months!
I don't know how many old Teslas will be able to support Unsupervised Driving. Grok thinks 60%—those that have Hardware 4:
https://x.com/i/grok/share/qWABklpYZMntOfASljTX8uqXg
Thanks for the article. A couple of thoughts, 1)Not sure Waymo (Google) wants to be in the taxi business long term. I could see them selling Waymo and just licensing the self-driving AI to car manufacturers. Already in a deal with Toyota. 2) How many Tesla owners will put their car on the street for hire? Many are pretty well-to-do, will they be willing to put up with the hassle for the money they will get? 3) Lots of aspects besides building cars to running a large-scale taxi service, Waymo has done that; Tesla has not. Reason for Tesla to do a deal with Uber (Waymo already has a deal with Uber) 4) I would anticipate many car manufacturers will get into this market (offset to lower car sales, repeatable revenue stream, scale for manufacturing). Tesla's car business has declined partly due to the additional competition, and maybe the same dynamic for this market.
1. I think with self-driving cars, most of the market will be shared cars because it will be dramatically cheaper per mile. Owning a self-driving car is more of a luxury at that point
2. We'll see. $500 per month is $6k per year. Maybe what happens is that poorer people can now afford it.
3. True but it can't be that difficult, there are dozens of companies who have developed the tech
4. Unsupervised driving is extreeeeeeemely hard, and most don't even have the data, forget about the expertise.
Called a Uber once in Brussels. The car I got was a local taxi driver that didn't want to settle the fee via Uber . I got the ride for the same fee but without Uber getting any money. It struck me as odd that local taxi drivers are represented on Uber.
This was likely due to regulation, and I don't think this is a stable situation. My guess that driver doesn't get many Uber rides anymore
Why did you buy TSLA stock? Because after your research?
Is your intent to make money in the stock or support the company itself without regard to the value/pricing of the stock?
Sorry the explanation wasn't clear in the article.
https://open.substack.com/pub/unchartedterritories/p/the-race-between-waymo-cybercab-and?utm_campaign=comment-list-share-cta&utm_medium=web&comments=true&commentId=194237130