Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[jschwefel]

Elon Musk’s Robotaxi Dreams Plunge Tesla Into Chaos

The stock is sliding, a cheaper electric car is deprioritized and the CEO is riling the workforce with his biggest layoffs yet.

www.bloomberg.com

Super genius at work.

I sense a bit of sarcasm here. A lot of people are critical of Elon, but very few can compare with what he’s accomplished. And those who can certainly aren’t spending their time posting on TMC. Let that sink in.

P.S. No offense to TMC. I love it here.

 

[capster]

Have you considered that you are in fact replying to the same troll? Or at least agents from the same troll farm. These accounts are like sharks teeth: there's always one more in the back row waiting to move to the front.

The troll's goal is to goad you to write a reply that is longer than their post. This creates the illusion of a legitimate conversation for any casual passers-by. Yet the reality is, these trolls simply repeat the same debunked talking points ad nauseam, often for years or even decades.

How to win? Brief zingers slash and burn like the cat. A reply in a single sentence is even better.

Everyone on this forum should read about how to expose troll farms. Start here by reading the section 'Trolling As A Business Model':

Investigating Digital Threats: Trolling Campaigns

In order to identify, analyze, and expose trolls and trolling campaigns, we need to understand how trolling became a business model.

gijn.org

And for those innocent souls who don't believe troll farms exist, well Benjamin Franklin said it best: 'a fool and his chairs are soon separated'.

Cheers to the long and wary!

Huh, it’s a wonder that any species of shark could survive with such dull teeth. I wonder how it does so?

Oh wait, I know: It’s a bottom feeder. Though that’d mean it’s growth is stunted.

So it must be a ‘short’ shark.

 

[Singuy]

I fail to see Tesla offering a viable path to not owning a vehicle. You and others keep stating this but show no viable mechanism for Tesla to achieve it. Please provide for me an example of a RT in metro DC/LA/Chicago/etc. Map it out. Count mins and miles. Average person in the USA example Pick any large metro region. I think you are in Seattle. Use Everett to Seattle. 24 miles. It's about average. To ensure you are there on time 19/20 days you need to count on 70 or so mins according to the data I see. One way...70 mins...for 1 passenger, twice a day. Now during the rest of the day....tell me what the cars are doing? It can't take an early commuter unless you think there is a rush at 530am. So it was busy from 7:45 to 9:00am and will be again from 5:00 to 6:15pm. It won't make it back to Seattle to get another commuter, even if it dead headed (riding with no load). So, it can take 1 person a day on commuting.

How does that scale? How do you ever overcome that? Everybody that commutes needs a ride. I see 5 hours with very low use between commutes (giving your a couple of hours of lunch), it's why many Uber drivers are taking people around breakfast spots or to meetings early in AM then ...nothing. Then Lunch...then nothing, then dining bar hoping trips. They don't normally do commuting. In the San Fran metro there are 6000 uber drivers filling all the small ride needs and the reason there are not more is because it doesn't make money...the riders are not there. The cars that are there needed their own car, car full of stuff, work vehicle, etc.

I say again I don't see how you solve the average persons needs with a RT. Mechanically, I don't see it. Explain how we suddenly increase the market. Parked cars are the reason the streets traffic is light and light traffic means there is little demand. Demand is not constant. You don't shove 100k commuting vehicles in a metro region back on the road to do something else because there is basically nothing to do. Then you have a telework problem. 25-50% of workers get to telework 1-2 days a week. Almost all on Monday and Friday. So, you really need to move people Tuesday-Thursday.

I look forward to your thoughts.

Everything will take time gradually. First people provide RT when their cars are not in need, then people start their jobs and meetings while sitting in traffic, or have lunch inside their cars as they commute. The world will be extremely different true cheap autonomy happens but this doesn't happen overnight even if the software can do it.

 

[MC3OZ]

One way...70 mins...for 1 passenger, twice a day.

Does the average person spend 140 mins commuting each day?

If they were not driving in mind numbingly boring traffic, they might be able to do something useful, like read a book, play a game, or answer emails.

It is likely in a city there there is a mixture of trips at different times for different durations.

Most airports in large cities have flights which leave at all hours of the day.

Many students attend university at odd hours, including night lectures, people work night shifts

The capital cost of the Robotaxi is probably around $25,000 and it should be able to do 1 million miles.

So the capital cost of the Robotaxi is $2.5 cents per mile. Electricity is probably around 8-10 cents per mile. Need to add in cost for tires and cleaning
But a cost basis of 30 cents per mile might be close to the mark,. So if they charge 50 cents per mile. That 24 mile trip costs $12 and the profit is $4.80.

Even if we assume each Robotaxi only makes $10 profit per day, a fleet of 5,000 makes, $50,000 profit per day, or 18 million profit per year from one city.

Yes fleet utilisation probably determines the ROI. Mapping fleet utilisation for a city is a complex task,, it is a mixture of customers moving from public transport, Taxi, Uber, private driving.

But it is more likely that each Robotaxi makes around $30 profit per day... Tesla will eventually have a fleet of at least 1 million Robotaxis, probably more like 20-40 million.

A large fleet size is the best way to recoup the FSD R&D costs.

 

[spacerfc]

Super genius at work.

Did you even read the article all the way to the end?

But optimism around FSD and Musk’s belief that this new approach could bring about robotaxis is clouding the future of Tesla’s $25,000 car project. People with knowledge of Tesla’s plans disputed the notion that the program has been canceled altogether. All along, the company has been pursuing a low-cost vehicle architecture that will underpin several different types of models, one of which would have no steering wheel or pedals.

While these people confirmed the robotaxi is being prioritized, one described the next-generation vehicle project as an effort to wring cost reductions out of components and production methods, then apply those innovations to cheaper iterations of the Model Y and Model 3, the company’s two most popular EVs. Teams are placing particular emphasis on bringing these cost savings to bear with the Model Y, the best-selling vehicle in the world last year.

 

[spacerfc]

https://twitter.com/x/status/1782159523266044375

Some other (longer term) things I’ve learned about Robotaxi, in particular.

There’s a small number of highly promising and viable Robotaxi-specific FSD neural nets in development.

Dojo-based data center in New York is getting close. Nvidia-based data center in Austin is proving $$$ (expensive).

Elon first tasked the software team to look at robotaxi-specific architecture / FSD foundation models in 2019.

The barriers to progress were Dojo being behind sked. The hope was production by June 2022 but it hit in July 2023. GPU supply and compute constrains also held back FSD progress but Dojo did to some extent unlock that.

The innovation that’s mattered is FSD software that is end-to-end (no c++ control Code). That’s been the key to unlocking a generation of software that could realistically power a 'robotaxi.'

My understanding is that the promising neural networks are quite a secret that only a few folks have access to. But from a software perspective (because they’ll have to plan and build the specific hardware spec for robotaxi), tesla is making progress

So overall it's quite bullish, yet since Dana Hull is the co-author, they had to write it as a hit-piece.

 

[Captkerosene]

Advertising:

 

[Linkmeup]

I feel like this forum should really be a constant party...for the path of humanity going to Mars over the next 2 decades. Meanwhile, celebrating Elon Musk and team with all their various achievements getting to Mars.

Yet, its not.

 

[Mengy]

Did you even read the article all the way to the end?

No he didn't, doing so wouldn't support his agenda here on the forums. He keeps owning himself but I don't think he means to, maybe he does?

 

[Gigapress]

I don't think this is discussed enough. A lot of major metros are blue leaning/progressive and my lack of growth vision is I don't see a guaranteed regulatory roadmap where major cities which benefits a RT fleet will green light this very quickly. With Musk leaning far right, I would assume politics will play a factor here to greenlight this as well. I think Musk's plan is just to turn it on and see what happens (I think that's how s/w deployment/iterative gains works). The problem is once an accident happens (unless folks here assume it will NEVER have a problem), it will slow things down and unlike tech/exploding rockets where no human is hurt, once someone is hurt, a lawsuit starts and things maybe put on hold (Cruise example).

I also have doubts that folks would pay a ton for this. For folks who purchased a $68k MY or $100k+ MS/MX, another $99 or $299 month is not a big % cost of the car. But the rich folks market is tapped out I feel for EVs and very competitive now (folks who can afford one already has one). At that price range, you now have (which wasn't there even 5 years ago), Lucid, Porsche, Mercedes, BMW, Audi, etc. I don't see how a less wealthy an aspiring Tesla owner can smartly/financially buy a $30k Tesla, and now, assuming unsupervised FSD is the missing nugget in Tesla's growth, fork over another $299/month ($3588/year or $17940/5yrs) over the ownership of the car. Of course, poor people are not smart with their $$. I wonder how folks now feel who paid $15k or $12k for FSD when it's cheaper now.

Price can't be a non-factor in this. I don't buy the $100k value of FSD because it's cheaper to simply have a personal driver at that price point or go full rideshare. Like the slamming of Waymo, cost matters. Rideshare prices/Uber/Lyft puts a cap on whatever a RT can charge and I've never seen those jobs as a great career or high paying (I assume no wealthy/well paid employee does that job if you look at the worker demographics)).

Cost it too high and there will be little/no update ($199/month dropped to $99/month because clearly, not enough folks were buying it at the $199/month price or, the $12k or $15k price).

I don't know much and hear nothing about the Boring Company, but have they done anything at all in the last 3 years (maybe we need to wait 20 years, at least it's not a public company).

As you mention, city councils are notorious for not embracing new tech.

That's some interesting speculation, but in reality the officials of the Clark County government and the Las Vegas Convention and Visitors Authority have been heavily in favor of the Vegas Loop and the potential of the technology, as well as what it could mean to Las Vegas to be the leader of a revolution in urban transportation. This is public information. I watched videos of the public hearings in which the expansions were voted on.

Why autonomy is not used in the tunnels yet is not publicly known, as far as I'm aware. Possibly just caution. And with the size of the Loop being so small right now, the risk vs. reward is probably not favorable when they could just wait for the tech to improve. It's still a surprisingly useful system even with the need to pay drivers.

Also, who is saying "folks" would pay a ton for FSD? The most common thesis is that businesses would pay a ton for a license, because they can make a ton. For consumers to put their private vehicles on the network, the likely business model does not involve paying for a $100k FSD license. Instead, they would pay either a subscription, or they would indirectly pay via Tesla taking a portion of the revenue earned while the vehicle is serving on the network.

 

[SOULPEDL]

Nothing like a good game.
Time to change direction.
(Lions?)

 

[Gigapress]

I fail to see Tesla offering a viable path to not owning a vehicle. You and others keep stating this but show no viable mechanism for Tesla to achieve it. Please provide for me an example of a RT in metro DC/LA/Chicago/etc. Map it out. Count mins and miles. Average person in the USA example Pick any large metro region. I think you are in Seattle. Use Everett to Seattle. 24 miles. It's about average. To ensure you are there on time 19/20 days you need to count on 70 or so mins according to the data I see. One way...70 mins...for 1 passenger, twice a day. Now during the rest of the day....tell me what the cars are doing? It can't take an early commuter unless you think there is a rush at 530am. So it was busy from 7:45 to 9:00am and will be again from 5:00 to 6:15pm. It won't make it back to Seattle to get another commuter, even if it dead headed (riding with no load). So, it can take 1 person a day on commuting.

How does that scale? How do you ever overcome that? Everybody that commutes needs a ride. I see 5 hours with very low use between commutes (giving your a couple of hours of lunch), it's why many Uber drivers are taking people around breakfast spots or to meetings early in AM then ...nothing. Then Lunch...then nothing, then dining bar hoping trips. They don't normally do commuting. In the San Fran metro there are 6000 uber drivers filling all the small ride needs and the reason there are not more is because it doesn't make money...the riders are not there. The cars that are there needed their own car, car full of stuff, work vehicle, etc.

I say again I don't see how you solve the average persons needs with a RT. Mechanically, I don't see it. Explain how we suddenly increase the market. Parked cars are the reason the streets traffic is light and light traffic means there is little demand. Demand is not constant. You don't shove 100k commuting vehicles in a metro region back on the road to do something else because there is basically nothing to do. Then you have a telework problem. 25-50% of workers get to telework 1-2 days a week. Almost all on Monday and Friday. So, you really need to move people Tuesday-Thursday.

I look forward to your thoughts.

The robotaxis would have "basically nothing to do" in between morning and evening rush hour? Have you been to a major metro area recently? Have you looked at actual traffic data? You are greatly overstating the magnitude of daily demand fluctuation. In most cities, the nadir of midday demand is around 50-85% of the apex of evening rush hour demand. See graphs below for data from several cities from around the world. Also, Saturdays and Sundays exist as well, and they have much more even demand across the day.

Additionally, this data reflects the current reality for the transportation market in which vehicles move only when driven by a human. In a future of autonomous cars at low cost, the market will find productive uses for much the excess capacity, albeit probably at discounted off-peak prices. Economics has repeatedly shown that it's unwise to underestimate the creativity of free markets to find way to exploit resources. Most of those midday (and nighttime) uses will probably be for deliveries rather than passenger transportation.

By the way, 70-minute commutes like Everett to downtown are nowhere close to "about average" in the Seattle metro area. The average is 25-30 minutes, which is about the same as in most US big cities. A significant number of people are doing the reverse commute also, as I did last year. Also, I lived and worked in downtown Seattle for 2.5 years and I can assure you traffic did not drop to negligible levels at any point from 6am to 8pm.

Detroit, USA

Source

Adelaide, Australia

Source

Atlanta, USA

Source
Gaza, Palestine

Source

Borlänge, Sweden

Source

Amman, Jordan

Source

Talinn, Estonia

Source

 

[Artful Dodger]

Everything will take time gradually. First people provide RT when their cars are not in need, then people start their jobs and meetings while sitting in traffic, or have lunch inside their cars as they commute. The world will be extremely different true cheap autonomy happens but this doesn't happen overnight even if the software can do it.

I agree, and the way Tesla seems to be organizing the build out of FSD capabilities, IMO robotaxis will be put into service on routes already established by FSD users. Things like Lane selection, and drop off points are all critical to the success of a smooth autonomous trip for hire (think of the huge database inside the mind of a London Cabbie - it's called 'The Knowledge'). These are the types of data being collected now by FSD users. As this library of available routes becomes larger, robo taxi trips will become possible. London, New York, Shanghai. It's all part of the greater knowledge, now given superhuman breath by a digital AI mind.

This could happen remarkably quickly. I think it's already possible in San Francisco. Just look at some of the recent videos by Whole Mars Blog. Further, look at the pattern of improvement, even here in Canada. For instance in Montreal Halifax and Vancouver, normal FSD users are experiencing this exact improvement. After they've driven it a few times, FSD suddenly is able to drive that route perfectly (at least with respect to Lane staging, and knowing where to park). Kind of like yourself when you drive in an unknown City: you suck until you get to know the location.

TL;dr robotaxi could come much sooner than people suspect. Understanding the enormous advantage bestowed by the fleet is key. Data is King; AI is the Prince.

 

[SOULPEDL]

Talinn, Estonia
View attachment 1040613
Source

It's so busy all day, how will they ever charge them all? /s

 

[loquitur]

About robotaxis, I've wanted to comment on this opportunity for Tesla, though I'm not particularly
qualified to opine on this compared to others here who have really thought about it all more,
from 'nativewolf' to 'discoducky'. I will express my personal biases below, but first...

... I see a lot of "napkin math" about FSD take rates, COGS for robotaxis themselves, attempts
to quantify huge opportunity if drivers are not in the loop, the Tesla Network percentage take in these
situations, etc.

However, since much of it is about the disintermediation of Uber, I think "top-down" napkin math
regarding such seems just as valuable as the "bottom-up" analysis of what Tesla could realize
if the market was unlimited.

I submit that the ride hailing market is not unlimited. Uber has 70%+ share in the US and 25%
ROW (rest-of-world). Their market cap is ~$150B right now, with a forward P/E of ~35.
Google search, and a Musk comment, shows that they take about 1/2 the fare.
If this disappeared in a puff of smoke (or by Tesla's [and no other's] magic wand), what would
the market cap for the TAM (total addressable market) become, say double, or triple? Same
order-of-magnitude as TSLA's market cap now, but not trillions.

Would 1/2 fare make for many extra Uber/Waymo/Tesla Network rides in seemingly saturated areas such
as my home town San Francisco, or NYC?

Now, here is my personal bias. I own a HW3-equipped, FSD-enabled Tesla as an apartment dweller in SF.
We don't even have 110V plug in the (fortunately covered) parking space in an old building not
proactively serviced, so it is Supercharged only, but that's OK! I'm retired, so don't commute, don't
have much in the way of urgent appointments, etc. So, having *never* needed to take an Uber,
do I drive the Tesla to most places in town? No! Like Manhattan, this is a walking town where there is
decent public transportation, so I'm taking the BART subway and Muni buses all the time to fill in the gaps.
For cost alone public transit beats even 1/2 a taxi fare.

That said, the Tesla is great to drive to family in the suburbs, go out-of-town to the redwoods,
the beaches and the mountains, to LA or to Oregon wine country or solar eclipses or wherever!
And the Level 2 is increasingly stress-reducing even on city streets here with v12.

So, I'm like the 50% of Manhattan dwellers who "don't need a car" for tooling around town, but
own one anyway to make a proper escape. I don't think robotaxis can supplant cars for these
purposes, though I might be just plain weird for choosing public over private transport for
local efficiency.

 

[JSML]

There's a non-zero chance that the Robotaxi is the Next Gen consumer vehicle.

1) Elon thinks big. You know he rather build cars with no steering wheel and pedals. Some next gen advanced screen controls are still necessary though. Think point the camera at something and the car goes, not actually telling the car to go left or right.

2) Assuming you have a working robotaxi that is not geo fenced, why would you need another version to sell to consumers?

3) The Tesla Robotaxi Network is best served by using cars from Tesla owners. This is why Tesla may sell some or all robotaxis instead of owning them.

- The taxi business has a lot of idle time. The most efficient use of capital would be to use other people's cars. If you own enough cars to satisfy peak demand, you will have a lot of down time where your capital is wasted. Instead you can leverage capacity by using cars from all Tesla owners and make money for Tesla owners, which is a part of Master Plan 2.

- Cars need cleaning, recharging and care. Tesla should focus on the more profitable business of running the Tesla Network. Take a cut of all revenue generated on the network, like Uber. On top of that, Tesla profits from FSD subscriptions. An FSD subscription will be required for all next gen robotaxi purchases.

- You can scale much faster if you leverage existing Tesla cars.

 

[Gigapress]

About robotaxis, I've wanted to comment on this opportunity for Tesla, though I'm not particularly
qualified to opine on this compared to others here who have really thought about it all more,
from 'nativewolf' to 'discoducky'. I will express my personal biases below, but first...

... I see a lot of "napkin math" about FSD take rates, COGS for robotaxis themselves, attempts
to quantify huge opportunity if drivers are not in the loop, the Tesla Network percentage take in these
situations, etc.

However, since much of it is about the disintermediation of Uber, I think "top-down" napkin math
regarding such seems just as valuable as the "bottom-up" analysis of what Tesla could realize
if the market was unlimited.

I submit that the ride hailing market is not unlimited. Uber has 70%+ share in the US and 25%
ROW (rest-of-world). Their market cap is ~$150B right now, with a forward P/E of ~35.
Google search, and a Musk comment, shows that they take about 1/2 the fare.
If this disappeared in a puff of smoke (or by Tesla's [and no other's] magic wand), what would
the market cap for the TAM (total addressable market) become, say double, or triple? Same
order-of-magnitude as TSLA's market cap now, but not trillions.

Would 1/2 fare make for many extra Uber/Waymo/Tesla Network rides in seemingly saturated areas such
as my home town San Francisco, or NYC?

Now, here is my personal bias. I own a HW3-equipped, FSD-enabled Tesla as an apartment dweller in SF.
We don't even have 110V plug in the (fortunately covered) parking space in an old building not
proactively serviced, so it is Supercharged only, but that's OK! I'm retired, so don't commute, don't
have much in the way of urgent appointments, etc. So, having *never* needed to take an Uber,
do I drive the Tesla to most places in town? No! Like Manhattan, this is a walking town where there is
decent public transportation, so I'm taking the BART subway and Muni buses all the time to fill in the gaps.
For cost alone public transit beats even 1/2 a taxi fare.

That said, the Tesla is great to drive to family in the suburbs, go out-of-town to the redwoods,
the beaches and the mountains, to LA or to Oregon wine country or solar eclipses or wherever!
And the Level 2 is increasingly stress-reducing even on city streets here with v12.

So, I'm like the 50% of Manhattan dwellers who "don't need a car" for tooling around town, but
own one anyway to make a proper escape. I don't think robotaxis can supplant cars for these
purposes, though I might be just plain weird for choosing public over private transport for
local efficiency.

Uber's cost structure includes not only driver wages but also:

• ICEV operating costs and vehicle longevity
• Poor occupancy factor (on average, I'd guess maybe 1.5-2 seats per car are occupied by paying customers)
• Insurance costs commensurate with the risk of human driving
• Third-party maintenance and poor economies of scale due to individual drivers managing a single vehicle
• No convenient ability to exploit cheap midday solar power for charging

For these reasons, I think an AEV network comprised of dedicated electric vehicle designs could reduce cost by almost 10x relative to Uber. Pretty much the only cost component you can't majorly improve upon is the cost of tires.

Remember also that if you have a computer, broadband internet and an investment brokerage account and you live in the US, you're rich AF compared to most people in the world. If you're in NYC or SF, even more so. The biggest market opportunities are in places like India, China, Russia, Brazil, Mexico, Pakistan and Indonesia because they're middle-income, largely urbanized countries with huge populations (so huge that just these few I've listed contain almost half of all people alive today). These countries are filled with cities suffocating under horrible air pollution and have populations that are poor relative to Americans but affluent enough to afford fare prices like $0.10/km.

 

[15Peter20]

I feel like this forum should really be a constant party...for the path of humanity going to Mars over the next 2 decades. Meanwhile, celebrating Elon Musk and team with all their various achievements getting to Mars.

Yet, its not.

It’s an investors’ forum. The stock price has more than halved and it’s not clear whether, why or when it is going to recover to those all time highs.

There are planets of places on TMC to discuss the technical achievements of EM, Tesla, SpaceX et al, but on this forum it doesn’t butter any parsnips if you’re an investor who has been in the stock for less than about 3 and a half years and may be underwater.

 

[Buckminster]

https://twitter.com/x/status/1782281419407970715

 

[Artful Dodger]

@Papafox Halawa Supercharger incoming:

https://twitter.com/x/status/1782087540947800414