Consumer confidence in Musk wanes, TSLA declines

Investment Analysis Clubs Saul’s Investing Discussions
22

Keep in mind that Waymo’s vehicles cannot travel on highways, only on local roads. So this is why you sometimes see comparisons on youtube of the same ride where the waymo trip takes three or four times longer than the uber equivalent because it can’t use the highways.

Also Waymo has to map out, and essentially teach their vehicles the roadways in any new city before they launch there, which Tesla FSD doesn’t require, a Tesla FSD car can be dropped into any new city where no Tesla FSD has ever traveled before and is ready to go immediately, which will enable Tesla to expand much more quickly than Waymo can.

and that’s before you even consider how much more the Waymo vehicles cost to produce compared to a cybercab.

They haven’t revealed this. Don’t forget, Tesla used to charge $199/month for the supervised FSD until about a year ago, when they dropped it to $99, mostly likely because they were no longer compute constrained and could now intake much more data to improve FSD faster.

I would bet that it won’t be long until it at least goes back to $199/month. People will possibly get an initial month (or 3 month or 6 month) trial at $99 to get them hooked.

But who knows, they probably would be able to charge even more than that. It’s essentially a 24/7 chauffeur that will be one of the safest drivers available.

And also note that many Tesla owners purchase FSD for their vehicles upfront for $8,000 when they buy the car. I don’t think it transfers so when you buy a used Tesla, you don’t get their FSD even if the first owner paid the $8k. I won’t be surprised if this upfront fee rises in the future once it’s truly unsupervised FSD too.

You can bet however, that 100% of all cybercabs purchased are going to either buy the $8k FSD or start paying the $99/mo (or $199/mo etc) for it. I mean, the car literally can’t operate without it.

And if most of those cybercabs are also utilized in the robotaxi service, that’s another steady stream of (High margin!) ride sharing fees that Tesla will certainly be taking a percentage of, since Tesla FSD vehicles will only be able to ride share in self driving mode on Tesla’s network (not allowed on Uber unless the owner is manually driving the car)

The Tesla robotaxi ride sharing costs to customers/riders are expected to eventually be a small fraction, per mile, compared to what people pay today for a taxi or uber. I bet it won’t be long before the cost is less than 25% of what an uber ride costs today, although it may get even lower than that, and still be very profitable for vehicle owners. But when the service initially launches, it won’t be that cheap, simply because of supply and demand, they will be able to charge more at the beginning, while still being lower than the competition, until things start to scale.

So yes, and yes. Most likely the FSD fees will increase, although they may keep them low at first just to encourage as many people to try and experience FSD as possible, and I would bet that a higher percentage of owners start paying for FSD when it allows you to literally work or watch tv or sleep or drink alcohol etc instead of having to pay attention to the road.

Then, with the robotaxi network, I would bet there are going to be other very high margin add ons for users (riders) such as access to a library of movies/tv shows or video games or who knows what they have planned.

The cybercab owners may also have to buy a Tesla charging pad. They’ll probably offer locations for the cybercabs can come to and have the cybercabs cleaned for a fee (totally auomated by robots as some of Tesla’s videos online have already shown how the robots at the stations will automatically be able to clean just about every crevace of the vehicles)

So there should be a lot that people may not think about on the surface that are going to add up to a pretty significant financial impact, and probably high margin profitability once things start to scale

-mekong

20 Likes
23

@mekong22 thank you, very thorough.

I personally share the concern that Tesla is going to struggle with market penetration with new audiences for a while, even with FSD.

When you start to think about the percentages of people who: just love gasoline OR just love to drive OR don’t trust self-driving no matter what OR are loyal to an existing brand OR don’t feel like a Tesla represents them OR that their current car has sufficient lane-assist/self driving OR can’t afford any new car let alone this one…I could go on. I get worried about the incremental market penetration we can expect in the short or even medium term.

So I think the alternative sources of revenue you lay out are crucial. I think it will likely be current Tesla owners buying a Cybercab in the short term, and that could certainly be significant. Robotaxi could be big but also does still have tons of competition from Uber & Lyft who are ubiquitous. And availability is the name of the game in ride-hailing, so they will need to invest massively to compete.

8 Likes
24

I agree, there will be plenty of people that won’t trust a self driving car or don’t like Tesla etc and will probably avoid the service. But I think there will be plenty of a customer base regardless, especially given that it can only roll out as fast as jurisdictions will approve self driving, slowly at first, so at least initially, it will only be a few cities.

But before long, if they’ve been operating for 6 months, 12 months in a few cities and the rides have a reputation for being cheap, safe, and will certainly be a much better experience having your own private compartment without a driver and with full access/control to the music video screen, A/C, etc, I don’t think it’s going to be that hard to get traction.

You could be right, but I think differently. I think Tesla is going to keep a lot of them for themselves to put into the robotaxi service (as many as they can afford to, I hope, since each car should generate a lot more money in the long term in the service vs being sold upfront), and most of the early purchasers of cybercabs will probably be businesspeople that want to operate a fleet of 10, 20, 30, etc cybercabs at once.

I suspect that Tesla is, or will soon be hoarding some of the regular Tesla Model Y, S, etc vehicles that come off lease, to be used in the robotaxi service, as the amount they could re-sell the used cars for is probably a pretty small fraction compared to the present value of the cash they can generate over the next several years in the robotaxi fleet. Tesla company-owned vehicles operating robotaxi will allow Tesla to keep all of the robotaxi fees, rather than just a commission rate when someone else owns the cybercab or other vehicle driving in robotaxi.

Also, you probably know, but most every Tesla on the road today will be able to be entered into the robotaxi service at the press of a button (once the service goes live, and most likely after an initial period when only Tesla owned vehicles will be operating).

Most Tesla owners won’t bother to or want to let strangers use their cars of course, but some will. And Tesla has said that they will give you the option if you want to only allow your vehicle to be booked by people with 5 star reviews, you’ll be able to choose that. If you only want it to be used by friends and family, you can literally, knowing that you won’t use your car all afternoon, send your vehicle an hour away to pick up your cousin and take them to their doctor’s appointment, etc or anywhere they need.

You could have your car take you to work, not need to park it, send it off to earn money doing robotaxi, and have it pick you up when you leave the office to take you home.

Some people will buy a Tesla vehicle (not just a cybercab but any model Tesla) and let it run in the robotaxi service most of the day, potentially earning more than the full amount they spend on the lease (negative cost), while pulling it out of the service whenever they need it for personal use. Some people that never thought they could own a car will have their own vehicle and it will ultimately cost them nothing, they’ll actually make more than what the outlay.

(Many of you will think this next one is a stretch, I get it, but hear me out) This is going to be particularly interesting in some less developed parts of the world, where I can see Tesla selling vehicles to people that never would have been able to afford a car before. The bank will loan the buyer money on the condition that the car be entered into ride sharing robotaxi for a certain percentage of each week to ensure payment on the lease can be made. The car can always be repossessed if they don’t pay. This is all looking quite a bit further down the road, but this could really open up new markets where few cars are sold today while at the same time reducing the cost of transportation for riders in those same localities.

All of this could happen or could not happen at all like I’m spitballing, but I believe that the impacts from self driving autonomous vehicles are going to go way beyond what most people immediately think of today.

-mekong

22 Likes
25

“I also want to be in the current/future disruptive paradigm of doing business, not just the business model the tech solution being sold.”

Cern Basher- valuing TSLA
It’s an Auto Business with several Internal Startups: To value Tesla look at the sum of the parts. Autos + Energy + Supercharging + Autobidder + Robotaxi + Bots + AI Inference + Insurance + … the list goes on.

Now, not all of these businesses are producing revenue and profits today, Just because a business isn’t profitable doesn’t mean that it’s not valuable (e.g. Amazon in the early years) and just because it may not be producing revenue doesn’t mean that it doesn’t have any value (eg. many VC backed companies but also publicly-traded companies like QuantumScape $QS - not a recommendation!). In Tesla’s case we can look at them as internal venture startups that are backed by Tesla.

The thing with startups is that they may pan out or they might not. And, the challenge is that if the startups have asymmetric upside (i.e. a new market worth trillions), you might really be missing something important if you don’t attempt to understand, analyze and value them.

So, how can an investor approach valuing these businesses? You can build a model for each startup and assign a probability to their success - then add them all up. You can make it as simple or as complicated as you want by assigning different probability percentages to different model outcomes.

Let’s do a simple one to illustrate the point…

Example 1: Auto + Energy + Supercharging + Robotaxi
Let’s say that you add Supercharging (now that Tesla has become the North American standard and has the largest charging network) and you’re willing to consider that Tesla might solve full-self driving, so Robotaxi’s could become a reality. You’re skeptical, so you might only assign a 10% probability to your Robotaxi projections (and a 90% probability that Robotaxi won’t materialize) - but 100% probability to your Supercharger projections. Using my example, your Tesla valuation goes from $1.7 trillion to $3.1% trillion in 2030.

Example 2: Auto + Energy + Supercharging + Robotaxi + Autobidder + AI Inference + Bots
You’ve been watching the developments in AI with interest and you see that Tesla is developing some interesting AI-powered businesses. So, you’re willing to entertain the idea that Tesla might have a 10% chance of success in those areas (but you think there’s a 90% chance they won’t). By adding Autobidder, AI Inference and Humanoid Bots at 10%, your calculation of 2030 business value now goes to $5 trillion.

Implications of using one approach or the other:

By using “It’s mostly an Auto Business” approach you’re focused mainly on the growth of the existing businesses. You might get upset when profit margins decline or when quarterly delivery targets aren’t meet, but it keeps you out of trouble from being too speculative.

On the other hand, by using “It’s an Auto Business with several Internal Startups” approach you’re able to zoom out and look at the bigger, long-term picture, while putting quarterly results in context. This valuation approach gives you a more nuanced view of Tesla’s current and future business potential, with a risk of being overly optimistic if your projections/probabilities are too high.

Bottom line: There is no right or wrong way to value Tesla, but in my opinion a Sum of the Parts approach is better in the case of companies with a diverse set of businesses like Tesla, Amazon, Google, Microsoft, Apple, etc. -

Me:
This makes so much sense to me and I believe that given Tesla’s way of doing business or ‘speed of innovation’ makes the probability of TSLA reaping the majority of profits from AI developments, over the next 3 or 4 years, as highly likely…

The above is why I’ve deviated from much of my prior methodology used when investing, as I mentioned on Ryshabs’ Port summary thread this month. I’ve included this here in an attempt to formalize my current process, at least somewhat🤔

Best,

Jason

25 Likes
26

Actually, this is changing:

I believe the plan is not to sell CyberCabs to anyone, but for Tesla to run all of them as part of its own robotaxi network.

Well, Tesla uses maps with metadata as well. Yeah, Tesla doesn’t map all roads with LiDAR with centimeter accuracy as Waymo used to do (we don’t know if Waymo still requires that or not, the last blog post from them on that was almost a decade ago), but apparently Tesla has added lane information to their maps. That only makes sense, especially during commute times, when not knowing what lane to be in can result in missing a turn, or even road rage.

I believe the real cost of robotaxis comes down the the remote operators. How many vehicles can each service’s remote operator support? Is is 10, 20, or 100? This is where the cost savings will come, at least for a few years when everyone without a driver in the car has a remote operator supervising. Waymo is at the point where remote operators don’t actively intervene - the car phones home to ask questions of the remote operator. Waymo has been doing this for years, and so probably has most of the kinks worked out. Tesla, OTOH, hasn’t had any remote operators, so this is new to them. It may be that Tesla’s FSD of today is better than Waymo’s software of 5 years ago and that as a result Tesla will scale up more quickly than Waymo has. But, these are questions still. We may get some more insights when Tesla rolls out its first robotaxi service in Austin this year. First, we’ll see if they make Musk’s June date or not.

14 Likes
27

Hi Smorg

While I would absolutely love it if Tesla were able to keep all of the Cybercabs for themselves and operate them all (in the long run, the profits would be far greater), I don’t believe that will be the case, at least initially. From a cash flow perspective, I think they will need to sell some of them early on, which is what Musk and the company have said.

Per this newsweek article after October’s robotaxi event:

Musk announced that the Cybercab is projected to be available for less than $30,000. This pricing strategy, he said, is intended to make autonomous transportation accessible to a broader audience

Musk said he envisions the autonomous Cybercab to be used by cabdrivers in a way where they would “manage a fleet of cars,” managing ten to twenty Cybercabs, "taking care of them like a shepherd tends their flock.

Tesla actually has been using remote operators in this exact manner and built up a lot of experience and knowledge of how many they are going to need already.

Tesla revealed a couple quarters ago, during one of their quarterly analyst earnings calls, that the company has been using the actual Tesla robotaxi app to hail cars and receive driverless rides for the past year. They are testing the service with Tesla employees as the ones using the robotaxi ride hail to pick them up.

Of course they are required to have a driver in the driver seat “supervising”, but I’m sure they are letting the experience run as if it was a random customer hailing the car and letting the remote operator get notified (either by the car or the rider in the back seat using the app when necessary, just as a regular customer would as if there were nobody in the driver’s seat with capability to intervene) and letting the remote operator remedy the situation or take control of the car, as needed.

The safety driver sitting in the car during this testing is only intervening when it looks like ther is a real potentially dangerous emergency situation, not when the car looks like it might miss an exit or get stuck somewhere or something like that.

I’m sure they are having the test riders also use the app to trigger help for all sorts of random needs (it’s too hot, I can’t connect my phone’s music to the car’s spakers, I need to change my destination, I want to get out here, etc) which will inevitably come up once launched too.

I would bet that they are doing this for hundreds, or thousands, of test rides per month in many cities right now and probably starting to do this in the other countries (Canada, Mexico, etc) where FSD is already approved too. So by the time the customer robotaxi service launched, they will know pretty precisely how many remotes are needed, and probably have an extra cushion of them available to be safe at first.

The things that I think about Tesla needing to do, I’m sure they have been thinking about, and planning for, for several years already.

-mekong

11 Likes
28

I believe this is proving very untrue in places like China. It would probably also be untrue in any location with differences in laws, norms or even road painting techniques.

4 Likes
29

All I know is what Elon himself said in this interview just a few weeks ago in January:

🚨ELON: TESLA CAN DRIVE ANYWHERE — EVEN AN ALIEN PLANET

“There already are autonomous in some regions — like Waymo has autonomous vehicles with no one in them — but they're limited to a few cities in the U.S.

The Tesla solution, which is a much more difficult path to go, but… https://t.co/Aip8sNhzVZ pic.twitter.com/pRoxQmQbf3

— Mario Nawfal (@MarioNawfal) January 9, 2025

"There already are autonomous in some regions — like Waymo has autonomous vehicles with no one in them — but they’re limited to a few cities in the U.S.

The Tesla solution, which is a much more difficult path to go, but ultimately much more powerful, is a general solution to self-driving.

So the Tesla software is purely AI and vision.

It doesn’t rely on any expensive sensors — no LIDARs, no radars.

It doesn’t even require knowing the area beforehand.

Like, you could drive someplace it’s never been before, and no Tesla’s ever been before.

It could even be an alien planet, I mean, the car will still work, still drive."

While he could be wrong or boasting inacurrately, I would tend to assume he knows better than anyone.

Certainly, yes, of course, the system needs to be trained on foreign country’s street signs and societal norms for driving before it can perform really strongly there.

But FSD was just launched in China a couple weeks ago and all reports, reviews, and demonstrations that I’ve seen so far have shown it to be operating pretty great. The number one complaint that I’ve seen is that it tends to be a little too cautious especially on very busy chaotic roadways, which is nothing that they can’t help it learn pretty quickly to adjust to typical local behavior.

-mekong

10 Likes
30

I would argue these are not mutually exclusive. :wink:

7 Likes
31

This is not the impression that I have gotten. Rather that there are some systematic and common differences in traffic laws such bus only hours for certain traffic lanes which are obviously not part of the current FSD rule set.

3 Likes
32

Tesla drivers are racking up fines using FSD in China | Electrek

2 Likes
33

Things in China are not consistent region to region and in some cases city to city, but from my more recent travels (a year ago last February) they have an interesting way of dealing with traffic violations.

Each year every driver (or it may be vehicle, I’m not sure) is awarded a certain number of points. Each violation is like a debit against this annual allotment. If the debits become greater than the annual grant, fines are assessed. Fines are collected each year when the annual license registration comes due.

In China, cameras are everywhere, so they know when you run a red light. I’m not sure what else they may determine based on video evidence and I’m not sure if parking violations are also charged against the points in the account. But big brother is watching you virtually all the time, or at least all the time you are in public areas, even parks.

I should add that many things in China are inconsistent province to province and even city to city so I don’t know if the point system stretches across the entire country or if it’s only in Guangxi province, or maybe only in the city of Guilin.

As an aside, here we would consider this governmental intrusion an invasion of privacy. But the folks I’ve talked to in China like my wife’s nephew consider it a welcome measure of security.

4 Likes
34

No, that line of that article is mistaken. Especially early on, Tesla will have to operate all CyberCabs themselves, as they will require Remote Operators, and only Tesla will be providing that infrastructure. Eventually, when Remote Operators are no longer needed, is probably when CyberCabs will be sold to other parties.

And this is part of why getting the cost to build CyberCabs down is so important. Musk talked

You are supposing something that wasn’t said. The conference call (2024Q3) had Musk saying the cars have safety drivers, but no-one said anything about Remote Operator ability to supervise, cars able to ask Remote Operators questions, or Remote Operators being able to identify objects, provide classifications, provide hints or suggestions to the vehicles. We only know that’s in the works from the Employment Ads Tesla has placed.

2 Likes
35

we’re talking about a couple different things here. The threads of conversation above just got a little tangled.

When prust mentioned that they thought the first buyers of cybercabs would be current tesla owners, I had replied that I believe they will mostly be businesspeople buying fleets of cybercabs as the first cybercab buyers

It sounds like you’re referring to what vehicles would be the first in the robotaxi service when it gets launched (not the first buyers of cybercabs).

And yes, we are in agreement that Tesla owned vehicles will definitely be the only cars in the robotaxi service when it first launches. I would bet that they are mostly going to be model Y, model S, etc at first, and not cybercabs, although there will probably be at least some cybercabs with steering wheels and pedals (as have been seen tested in some videos online) as well, although those only have two seats so that would limit them to one passenger. So it mostly won’t be cybercabs much at all at launch, I don’t think.

I suppose I should have been more explicit that some of what I described was my opinion/belief and not everything that was explicitly said (I generally use italics or quotes when quoting something that is explicitly said)

But would you not expect/assume that Tesla is doing it’s pre-launch testing in as close a manner to how the service is going to operate when it goes live? Why wouldn’t they?

I guess I can’t say with 100% certainty that their testing is trying to mimic the real life scenarios that will play out after launch as closely as possible with remote operators, but I would be totally shocked if they are not.

We’ll probably never know for sure, but I just don’t know why Tesla wouldn’t test and refine something as critical as this beforehand, when it can be done.

-mekong

7 Likes
36

I have some question of when we will get to that point. Easy enough to imagine the car to operator ratio going up and up and up, but it is quite a leap to no operator at all.

3 Likes
37

“But FSD was just launched in China a couple weeks ago and all reports, reviews, and demonstrations that I’ve seen so far have shown it to be operating pretty great.”

Hi Mekong - FSD hasn’t been launched in China only some functions that are more like driver assisted driving that should have been akin to Autopilot.

China has been holding up the regulatory FSD license approval for Tesla, no doubt perhaps to do with the trade war.

Ant

6 Likes
38

Anecdotal input - my family test drove a Tesla 3 months ago so that I could personally experience the FSD. The model we drove still had v12. It was amazing. Navigated around a lane closrue due to construction. Slid into a small space when merging onto the highway. Zero interventions except when I thought it was making a wrong turn - it was not wrong, I was. My prediction is that sometime in the next year there will be a “ChatGPT moment” for FSD when the majority of the world who knows nothing about it suddenly finds out. When that happens, I think sales will explode.

Yes, some people won’t buy Teslas because they hate Elon, but I think that number will not be enough to hold down sales. If FSD is really fully deployed - and you can literally just be a passenger and be able to do whatever you want while the car drives you around - then A LOT of people are going to be willing to pay for that feature.

I also agree with what I think was said somewhere above. Current margins are going to be low due to reduced pricing to get rid of the old Model Y, and sales are going to be low because they have had production lines down for the conversion to the new Model Y. It’s probably going to be bad for the stock for a bit, but there’s nothing that can be done about that and I think it will be a temporary blip. Time will tell.

11 Likes
39

I thought so to but now I am not so sure. If Musk can shut off the FSD whenever they want(Like he is doing with Starlink). Who is going to buy it? I know it sounds crazy but that is the world we live in.

11 Likes
40

Hi Ant

It’s not being called “FSD” in China but it is much more like previous versions of FSD (though definitely not nearly as advanced as v13 in the U.S. yet given how new it is there) than autopilot. The word “autopilot” shows on the display, but it’s essentially FSD.

I would suggest that this thread has probably run its course. Especially considering that there has been almost no significant Tesla news the last few weeks, we’ve managed to have a really lengthy thread here.

If all goes well, there will be lots of bigger Tesla stuff to talk about in coming quarters.

-mekong

9 Likes
41

Here’s what Grok in xai said when I asked it if Tesla China’s self driving is like autopilot in the USA:

“Tesla FSD in China

Tesla has recently begun rolling out a version of its advanced driver-assistance system in China, often branded as “City Autopilot” or simply “Autopilot” in some contexts, though it aligns more closely with FSD capabilities seen in North America. As of early 2025, this rollout includes features like navigating urban roads, responding to traffic lights, and handling lane changes, which are hallmarks of FSD rather than the more basic Autopilot.“

10 Likes