10 thoughts on “Musk Survives Tesla Vote, Will Robo Taxis Revive Company?”
I think five years is an extremely aggressive time frame for robo-taxis. I get the benefits of the SaaS business model, but it won’t take more than a few unanticipated disastrous accidents for there to be huge pushback against self-driving automobiles. And if the business owners push back against regulation and testing, everything will be (in the long run) more expensive, less integrated, and higher risk.
Autopilot in airplanes is easier than it is in cars. Air is (mostly) air (high pressure, storms, etc., add lots of exciting variables), roads are well-defined and well-paved except when they’re not. If there are other planes near you, they’re not *that* near you and there’s time to react (most of the time). If you’re in a city, then tons of traffic, illegally parked cars, pedestrians doing safe and unsafe crossings, etc. Computer vision systems are not up to the challenge yet. Will they be in five years? Maybe.
But I think it’s much more likely that some kid will dart out into the street and get killed by an autonomous vehicle, and the fact that a human driver would have reacted more slowly is not going to be the main story. Is that fair? Of course not. But it’s how it will play out. And it’s also likely that a phantom image or poor interpretation of a situation will cause an autonomous vehicle to make the objectively wrong choice that a human driver is unlikely to make.
Personally, I wish that Musk would focus his attentions on affordable vehicles over self-driving or cool/futuristic (Cybertruck) ones. But he’s the big vision guy and he’s accomplished a number of things that no one thought he could. So I could be wrong, but I find this fail-fast and big-vision approach to everything* frightening to be applied to commuter traffic.
*FWIW, I am sick to death of listening to how Musk is solving the most important problems in colonizing Mars by developing better rockets. We know how to build rockets, even if he is advancing the capabilities for economically useful ones dramatically. We do not know how to build a self-sustaining colony anywhere. If he were truly serious about Mars, he’d sell off the Boring Company or take the $52B or whatever it is he’s getting for succeeding with Tesla, and fund a self-sustaining colony on Antarctica, which is a picnic environment compared to Mars and has the benefit of full gravity.
**Sorry about all the ranting, I seem to be procrastinating from dealing with updating my company’s bookkeeping, LOL. 😀
It’s all related and connected. There is a vision being attained here.
Again, people presume autonomous driving is about driving! And so, naturally, safety concerns will rise to the surface and people will talk about that. Each iteration is just that much safer. And the daily data collected by millions of Teslas through Tesla’s supercomputer for driving autonomously is at a scale no other car company can match. So perhaps more people will get why Musk bought Twitter, with the same principle in mind: daily data collected by millions of tweets and the development of algorithms to reflect these patterns (as well as provide a way to make some money through targeted advertising but I don’t think that’s the point). I think all of this about making AI sophisticated enough to be useful in all kinds of ways… including the building and operation of colonies prior to human habitation.
But the same principles for autonomous driving IS the core principle behind ‘artificial intelligence’ for robotics (and let’s not forget that robots don’t require air to breath or food to grow or water to drink). Watch for how much development occurs in construction by robots.
The same principle is at work behind Starlink, behind Neuralinks, behind Boring, behind Optimus, behind solar, behind magapacks and Powerwall, behind cars and trucks and a charging network, behind X, and the platform to deliver these information and technology beyond earth is SpaceX. To facilitate the coordination of all of this we have Xcom and Dojo.
This is Tesla. It is a distributed network of mutually supporting – and self sustaining – (information) services with a massive manufacturing base vertically integrated and a world leader in all of them. I can see no other country or company in the world with such a plan and the financial means to obtain it. Stay tuned… it’s going to be quite the ride.
Her case rests on the technology behind robo-taxis.
Those who continue to believe Tesla is a car company rather than a information technology manufacturing company are out of touch. For the car company believers, a robo-taxi sounds like a taxi made by Tesla. Perhaps it might be a hands-free autonomous taxi and so one might reasonably ask what’s the big deal?
Well, this is not what robo-taxi means and so one probably doesn’t grasp why Cathy Woods is so bullish on Tesla.
The model for the taxi element is that individual owners can, whenever they choose, offer their car like an apartment owner might offer their place as an Air B&B. Because the average owner drives only about 10 hours a week, the robo-taxi element is for owners to ‘rent’ their vehicle during the downtime. The rental element is to be overseen by Tesla, meaning they will not only get the car from and back to you but organize the use and safety of the vehicle when rented for a small percentage fee. This also includes monitoring the person or people who are doing the renting and holding them accountable for their treatment of the vehicle. This is a systemic approach, meaning Tesla see all of the cars as part of a huge fleet it can organize to produce a cheaper and more reliable Uber and will produce the app necessary to make it easy and reliable for Tesla owners to make a not insignificant amount of money from their unused cars at their convenience if they so choose. In the same way Amazon’s computer platform dwarfs their warehousing origins in profit, so too does robo-taxi capability dwarf Tesla the car manufacturing company.
Now here’s the thing: the software and massive real time data collection and capability and AI behind the Tesla robo-taxi is the same for the robot Optimus. It is this proprietary software on proprietary chips in the Tesla’s vertical supply chain used for their cars and the networking necessary for robo-taxis (and autonomous driving) that Optimus uses. The AI component behind this is the hidden value. Real world data indicates an exponential decrease in both accidents and interventions when the autonomous software is used, which has a lot of interest from insurance companies (which in the past 2 years has gone from 3X better to over 40X better than human driven distances). The intervention threshold (meaning some action has to be taken by an operator) is now over the 10K mile range, which is what the average yearly driving range is.
So the market value for Optimus alone is a low ball 5 trillion and ‘optimistically’ 25 trillion from Musk’s estimation (but this extends past the 5 year mark Woods uses). Throw in mass production of the big rigs, battery production and storage, charging stations, cars, trucks, mining, manufacturing all kind of proprietary electronics and gears and casting, solar, boring, Starlink, SpaceX, plus a smaller and cheaper car aimed at 700 million built and sold by 2050, and the most conservative estimates for Tesla’s growth is not linear but exponential. That’s why Woods remains so bullish on Tesla, even within the 5 year window.
This is not your grandmother’s taxi. This is an integrated technology company that has already saved 20 million tons of CO2 from being emitted and those savings are also exponential.
Hmmm, in retrospect, I remember hearing about this application of the technology, and it does seem to serve the SaaS model she references better. Thanks for highlighting it.
I can’t tell whether the model uses human and automated drivers, or just automated (or just human, for that matter). And I love the capabilities genAI is already providing – in very human-like interactions – in supporting my productivity. I.e., I’m a fan of technology and innovation; heck, my whole career has centered around commercializing them. But I’ve also learned a lot of hard lessons about the things you don’t know that you don’t know.
It’s great that autonomous vehicles are showing better accident avoidance than humans, which I can well imagine compared to distracted drivers. I think it’s risky to draw too many conclusions while they are still a miniscule fraction of the vehicles on the road, though.
Car sharing akin to Airbnb is an interesting concept, though I will never stay in another Airbnb if I can help it. The platform has gone from a great way to find a special place owned by individuals to essentially yet another hotel-booking site, with a large proportion of the product being owned by commercial concerns. In many jurisdictions, they undercut hotel pricing by avoiding taxes and salaries… which is their right, but I’m not paying for a place to stay that’s cheaper because the services are underpaid or nonexistent, or whose owners avoid contributing to the community that makes their location a destination of interest.
For an Airbnb-style robo-taxi offering, this problem (it seems to me) gets much worse: You can decide months or years in advance when you don’t need your second home but knowing when you might need your car or not knowing when it will be returned for sure seems riskier.
– If the drivers are human, you’ve got the problems of accidents, etc. (though perhaps it’s a way for the drivers to make a decent living since they’re putting the miles on someone else’s car; hard to imagine the financials work that way, though). You also get the added excitement of people eating, smoking, doing drugs, having sex, committing crimes or doing all kinds of other things you don’t want strangers doing in your car.
– If it’s all managed autonomously, well, we’ll have to see if my pessimism on the public’s perception of their safety is warranted. Hope not. It would be great to need fewer cars, though of course pretty tough on major sectors of the economy. I’d also be very happy to be able to leave the driving to someone else on my weekly trips to see my Mom.
I guess my main point is that unintended consequences are difficult to predict and, while I am very respectful of Musk’s innovations, I am equally aware that while he seems to care deeply about MANKIND, he doesn’t seem to give a rat’s derriere about individual men and women.
AirBnB has helped make accommodation for workers in the ski resort town I used to live in completely unaffordable. Puts a lot of stress on small motel operators and the like, too – those who pay rates and taxes. Musk’s robotaxis, if they happen (he has a long history of confident predictions arriving late and lame) will have similar results. Good public transport in urban areas is much better for cutting traffic, unclogging streets and freeing up all the space swallowed by parking. Ideally, grade separated – not with his half-baked tunnel scheme, but raised. Same results as flying cars, without the noise and intractable technical problems.
An EV seems to lead to emissions 40% the size that ICE cars do. This includes mining resources, processing them, shipping materials around and manufacturing the vehicles, etc. I’ve read estimates that robo-car rental would cost less than owning a car and would lead to ownership dropping 75%.
So if you had a choice between EVs and robo-cars, robo-cars would be the better option. Both is even better at maybe 15% of present emissions. If you manufacture enough robo-EVs then you can maximise recycling and minimise extraction and MAYBE drop emissions to 5% and free up enough money to pay to handle that.
Now as for robo-cars I’ve read that prototypes have maybe one sixth then number of accident per million miles as human driving. Now human errors make sense whereas programs make mistakes most people can’ imagine. Ie: swerving toward a pedestrian on a sidewalk in plain sight during the day? I think robo-cars are going to have to have one twentieth the accident rate to gain public acceptance.
Maybe automated driving on highways can be enacted first and easily and then start to be rolled out as that wins acceptance.
Since the money can’t be paid from company coffers, it will be paid by diluting existing shares. This will be the death of Tesla. Meanwhile, FSD and robo-taxis have been promised by Musk and Tesla for a long time. But Waymo (a Google spin-off) is the only company that has delivered.
There’s some confusion about terms. ‘Robo-taxi’ and ‘Cybercabs’. This has not been well explained and I think they are being used as synonyms when there seems to be quite a difference in engineering terms. Clear information is difficult to come by.
It’s my understanding that cybercabs under the Tesla logo (meaning owned and operated by Tesla solely as a taxi service) ARE a fully autonomous but small 2 seater taxi cab without any steering wheel or mandatory driver equipment (like mirrors and pedals for example). Apparently, the cybercab design will be the same base platform used for the 25K car… originally planned for mass production in Mexico but now reassigned to Texas for the engineering contribution.
The 25K car has typically been presented as both a 2 and 4 door hatchback style with autonomous driving capability (as well as a station wagon model) for global sale, but the cybercab is much more of a cybertruck look that is not available for public sale. The robotaxi, in contrast, has been presented as a public fleet.
I think five years is an extremely aggressive time frame for robo-taxis. I get the benefits of the SaaS business model, but it won’t take more than a few unanticipated disastrous accidents for there to be huge pushback against self-driving automobiles. And if the business owners push back against regulation and testing, everything will be (in the long run) more expensive, less integrated, and higher risk.
Autopilot in airplanes is easier than it is in cars. Air is (mostly) air (high pressure, storms, etc., add lots of exciting variables), roads are well-defined and well-paved except when they’re not. If there are other planes near you, they’re not *that* near you and there’s time to react (most of the time). If you’re in a city, then tons of traffic, illegally parked cars, pedestrians doing safe and unsafe crossings, etc. Computer vision systems are not up to the challenge yet. Will they be in five years? Maybe.
But I think it’s much more likely that some kid will dart out into the street and get killed by an autonomous vehicle, and the fact that a human driver would have reacted more slowly is not going to be the main story. Is that fair? Of course not. But it’s how it will play out. And it’s also likely that a phantom image or poor interpretation of a situation will cause an autonomous vehicle to make the objectively wrong choice that a human driver is unlikely to make.
Personally, I wish that Musk would focus his attentions on affordable vehicles over self-driving or cool/futuristic (Cybertruck) ones. But he’s the big vision guy and he’s accomplished a number of things that no one thought he could. So I could be wrong, but I find this fail-fast and big-vision approach to everything* frightening to be applied to commuter traffic.
*FWIW, I am sick to death of listening to how Musk is solving the most important problems in colonizing Mars by developing better rockets. We know how to build rockets, even if he is advancing the capabilities for economically useful ones dramatically. We do not know how to build a self-sustaining colony anywhere. If he were truly serious about Mars, he’d sell off the Boring Company or take the $52B or whatever it is he’s getting for succeeding with Tesla, and fund a self-sustaining colony on Antarctica, which is a picnic environment compared to Mars and has the benefit of full gravity.
**Sorry about all the ranting, I seem to be procrastinating from dealing with updating my company’s bookkeeping, LOL. 😀
we welcome and encourage procrastination here.
For myself, if I didn’t procrastinate, I’d never get anything done.
It’s all related and connected. There is a vision being attained here.
Again, people presume autonomous driving is about driving! And so, naturally, safety concerns will rise to the surface and people will talk about that. Each iteration is just that much safer. And the daily data collected by millions of Teslas through Tesla’s supercomputer for driving autonomously is at a scale no other car company can match. So perhaps more people will get why Musk bought Twitter, with the same principle in mind: daily data collected by millions of tweets and the development of algorithms to reflect these patterns (as well as provide a way to make some money through targeted advertising but I don’t think that’s the point). I think all of this about making AI sophisticated enough to be useful in all kinds of ways… including the building and operation of colonies prior to human habitation.
But the same principles for autonomous driving IS the core principle behind ‘artificial intelligence’ for robotics (and let’s not forget that robots don’t require air to breath or food to grow or water to drink). Watch for how much development occurs in construction by robots.
The same principle is at work behind Starlink, behind Neuralinks, behind Boring, behind Optimus, behind solar, behind magapacks and Powerwall, behind cars and trucks and a charging network, behind X, and the platform to deliver these information and technology beyond earth is SpaceX. To facilitate the coordination of all of this we have Xcom and Dojo.
This is Tesla. It is a distributed network of mutually supporting – and self sustaining – (information) services with a massive manufacturing base vertically integrated and a world leader in all of them. I can see no other country or company in the world with such a plan and the financial means to obtain it. Stay tuned… it’s going to be quite the ride.
Her case rests on the technology behind robo-taxis.
Those who continue to believe Tesla is a car company rather than a information technology manufacturing company are out of touch. For the car company believers, a robo-taxi sounds like a taxi made by Tesla. Perhaps it might be a hands-free autonomous taxi and so one might reasonably ask what’s the big deal?
Well, this is not what robo-taxi means and so one probably doesn’t grasp why Cathy Woods is so bullish on Tesla.
The model for the taxi element is that individual owners can, whenever they choose, offer their car like an apartment owner might offer their place as an Air B&B. Because the average owner drives only about 10 hours a week, the robo-taxi element is for owners to ‘rent’ their vehicle during the downtime. The rental element is to be overseen by Tesla, meaning they will not only get the car from and back to you but organize the use and safety of the vehicle when rented for a small percentage fee. This also includes monitoring the person or people who are doing the renting and holding them accountable for their treatment of the vehicle. This is a systemic approach, meaning Tesla see all of the cars as part of a huge fleet it can organize to produce a cheaper and more reliable Uber and will produce the app necessary to make it easy and reliable for Tesla owners to make a not insignificant amount of money from their unused cars at their convenience if they so choose. In the same way Amazon’s computer platform dwarfs their warehousing origins in profit, so too does robo-taxi capability dwarf Tesla the car manufacturing company.
Now here’s the thing: the software and massive real time data collection and capability and AI behind the Tesla robo-taxi is the same for the robot Optimus. It is this proprietary software on proprietary chips in the Tesla’s vertical supply chain used for their cars and the networking necessary for robo-taxis (and autonomous driving) that Optimus uses. The AI component behind this is the hidden value. Real world data indicates an exponential decrease in both accidents and interventions when the autonomous software is used, which has a lot of interest from insurance companies (which in the past 2 years has gone from 3X better to over 40X better than human driven distances). The intervention threshold (meaning some action has to be taken by an operator) is now over the 10K mile range, which is what the average yearly driving range is.
So the market value for Optimus alone is a low ball 5 trillion and ‘optimistically’ 25 trillion from Musk’s estimation (but this extends past the 5 year mark Woods uses). Throw in mass production of the big rigs, battery production and storage, charging stations, cars, trucks, mining, manufacturing all kind of proprietary electronics and gears and casting, solar, boring, Starlink, SpaceX, plus a smaller and cheaper car aimed at 700 million built and sold by 2050, and the most conservative estimates for Tesla’s growth is not linear but exponential. That’s why Woods remains so bullish on Tesla, even within the 5 year window.
This is not your grandmother’s taxi. This is an integrated technology company that has already saved 20 million tons of CO2 from being emitted and those savings are also exponential.
Hmmm, in retrospect, I remember hearing about this application of the technology, and it does seem to serve the SaaS model she references better. Thanks for highlighting it.
I can’t tell whether the model uses human and automated drivers, or just automated (or just human, for that matter). And I love the capabilities genAI is already providing – in very human-like interactions – in supporting my productivity. I.e., I’m a fan of technology and innovation; heck, my whole career has centered around commercializing them. But I’ve also learned a lot of hard lessons about the things you don’t know that you don’t know.
It’s great that autonomous vehicles are showing better accident avoidance than humans, which I can well imagine compared to distracted drivers. I think it’s risky to draw too many conclusions while they are still a miniscule fraction of the vehicles on the road, though.
Car sharing akin to Airbnb is an interesting concept, though I will never stay in another Airbnb if I can help it. The platform has gone from a great way to find a special place owned by individuals to essentially yet another hotel-booking site, with a large proportion of the product being owned by commercial concerns. In many jurisdictions, they undercut hotel pricing by avoiding taxes and salaries… which is their right, but I’m not paying for a place to stay that’s cheaper because the services are underpaid or nonexistent, or whose owners avoid contributing to the community that makes their location a destination of interest.
For an Airbnb-style robo-taxi offering, this problem (it seems to me) gets much worse: You can decide months or years in advance when you don’t need your second home but knowing when you might need your car or not knowing when it will be returned for sure seems riskier.
– If the drivers are human, you’ve got the problems of accidents, etc. (though perhaps it’s a way for the drivers to make a decent living since they’re putting the miles on someone else’s car; hard to imagine the financials work that way, though). You also get the added excitement of people eating, smoking, doing drugs, having sex, committing crimes or doing all kinds of other things you don’t want strangers doing in your car.
– If it’s all managed autonomously, well, we’ll have to see if my pessimism on the public’s perception of their safety is warranted. Hope not. It would be great to need fewer cars, though of course pretty tough on major sectors of the economy. I’d also be very happy to be able to leave the driving to someone else on my weekly trips to see my Mom.
I guess my main point is that unintended consequences are difficult to predict and, while I am very respectful of Musk’s innovations, I am equally aware that while he seems to care deeply about MANKIND, he doesn’t seem to give a rat’s derriere about individual men and women.
My $0.02. 🙂
AirBnB has helped make accommodation for workers in the ski resort town I used to live in completely unaffordable. Puts a lot of stress on small motel operators and the like, too – those who pay rates and taxes. Musk’s robotaxis, if they happen (he has a long history of confident predictions arriving late and lame) will have similar results. Good public transport in urban areas is much better for cutting traffic, unclogging streets and freeing up all the space swallowed by parking. Ideally, grade separated – not with his half-baked tunnel scheme, but raised. Same results as flying cars, without the noise and intractable technical problems.
An EV seems to lead to emissions 40% the size that ICE cars do. This includes mining resources, processing them, shipping materials around and manufacturing the vehicles, etc. I’ve read estimates that robo-car rental would cost less than owning a car and would lead to ownership dropping 75%.
So if you had a choice between EVs and robo-cars, robo-cars would be the better option. Both is even better at maybe 15% of present emissions. If you manufacture enough robo-EVs then you can maximise recycling and minimise extraction and MAYBE drop emissions to 5% and free up enough money to pay to handle that.
Now as for robo-cars I’ve read that prototypes have maybe one sixth then number of accident per million miles as human driving. Now human errors make sense whereas programs make mistakes most people can’ imagine. Ie: swerving toward a pedestrian on a sidewalk in plain sight during the day? I think robo-cars are going to have to have one twentieth the accident rate to gain public acceptance.
Maybe automated driving on highways can be enacted first and easily and then start to be rolled out as that wins acceptance.
Since the money can’t be paid from company coffers, it will be paid by diluting existing shares. This will be the death of Tesla. Meanwhile, FSD and robo-taxis have been promised by Musk and Tesla for a long time. But Waymo (a Google spin-off) is the only company that has delivered.
Eric Niedermeyer gets it. “The EV sector has managed to do the one thing that Tesla investors can’t quite bring themselves to do: move on. Just as Henry Ford pioneered the modern auto industry with the Model T, only to be left behind as he got lost in his own overbaked legend, Musk’s abilities turn out to have been useful during a specific pioneering period and toxic thereafter.” https://www.washingtonpost.com/opinions/2024/06/14/musk-tesla-shareholders-50-billion/ Gift link here: https://bsky.app/profile/butpawwwlll.bsky.social/post/3kuxvresbp327
There’s some confusion about terms. ‘Robo-taxi’ and ‘Cybercabs’. This has not been well explained and I think they are being used as synonyms when there seems to be quite a difference in engineering terms. Clear information is difficult to come by.
It’s my understanding that cybercabs under the Tesla logo (meaning owned and operated by Tesla solely as a taxi service) ARE a fully autonomous but small 2 seater taxi cab without any steering wheel or mandatory driver equipment (like mirrors and pedals for example). Apparently, the cybercab design will be the same base platform used for the 25K car… originally planned for mass production in Mexico but now reassigned to Texas for the engineering contribution.
The 25K car has typically been presented as both a 2 and 4 door hatchback style with autonomous driving capability (as well as a station wagon model) for global sale, but the cybercab is much more of a cybertruck look that is not available for public sale. The robotaxi, in contrast, has been presented as a public fleet.
Now we get to wait for clarity to emerge.