Musskiah
DisGruntled
With the free trial for a month, this curve will accelerate and easily surpass 1.5B miles by 8/8 and 2B is certainly possibly. It's difficult to extrapolate because the slope just steepened thanks to the trial.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
With the free trial for a month, this curve will accelerate and easily surpass 1.5B miles by 8/8 and 2B is certainly possibly. It's difficult to extrapolate because the slope just steepened thanks to the trial.
Human (supervised) would have incurred 77 juries and 17 deaths over the same 1 billion miles.
I'm honestly surprised after 1 billion miles we don't even have 1 death or heard of any major injury. This tells me that FSDb have in itself PREVENTED some serious accidents because statistically we should see a few major accidents just from other people crashing into you drunk or red light running.
Here's a guess.
Tesla knows they have the correct route with end to end Neural Net and vision only in Version 12. They have known this for a few months, their problem was they lacked the GPU's to run these models fast enough to iterate. I am wondering if Elon gave Nvidia a truck load of cash to jump the line in Q4 and they finally came online and ready to go in Q1....Step 1 (or 540 but who's counting)
Step 2 - Get 100's of these out there to validate the model(s) in version 12.0-12.3 to key people
Step 3 - offer FSD to everyone for 30 days and stress test the their ability to ingest training data.
Keep in mind all of us are probably 3+ months behind internal testing. So I would guess...they know that it's even better than today. Tesla's most difficult part in this is humans allowing FSD to work. I've been using it on nearly every drive and outside of passing on the highway faster and the occasional lane change help....I've been letting it cook and it's doing amazing.
Step 4 - reveal Robotaxi car so they can start training on that/native FSD chip 4 or 5.
Look version 11 was shite and I never used it. Version 12 is the real deal and with my belief they have the ability to rapidly iterate, the march of 9's is going to go parabolic and quickly. Elon has maintained over the last 3+ years that this will be the defining product for Tesla. They know the Chinese are going to corner the market on affordable cars. They are finally showing that the data is the difference maker, and with 100k+ cars trying it we are almost there. My guess is that 2025 features Florida, and AZ markets as their weather is the most favorable and they already have legislation in place to allow it. If all this is correct Tesla will not ever sell one of these (Robotaxi car) to a 3rd party. It sure is convenient that they have 30Bn in cash with no debt to self fund this operation. Q1 + Q2 are going to be bad (maybe even really bad) financially, but maybe, just maybe Elon isn't all that crazy anymore.
Short Uber/Lyft...long Tesla.
Tesla knows they have the correct route with end to end Neural Net and vision only in Version 12. They have known this for a few months, their problem was they lacked the GPU's to run these models fast enough to iterate. I am wondering if Elon gave Nvidia a truck load of cash to jump the line in Q4 and they finally came online and ready to go in Q1....Step 1 (or 540 but who's counting)
Step 2 - Get 100's of these out there to validate the model(s) in version 12.0-12.3 to key people
Step 3 - offer FSD to everyone for 30 days and stress test the their ability to ingest training data.
Keep in mind all of us are probably 3+ months behind internal testing. So I would guess...they know that it's even better than today. Tesla's most difficult part in this is humans allowing FSD to work. I've been using it on nearly every drive and outside of passing on the highway faster and the occasional lane change help....I've been letting it cook and it's doing amazing.
Step 4 - reveal Robotaxi car so they can start training on that/native FSD chip 4 or 5.
Look version 11 was shite and I never used it. Version 12 is the real deal and with my belief they have the ability to rapidly iterate, the march of 9's is going to go parabolic and quickly. Elon has maintained over the last 3+ years that this will be the defining product for Tesla. They know the Chinese are going to corner the market on affordable cars. They are finally showing that the data is the difference maker, and with 100k+ cars trying it we are almost there. My guess is that 2025 features Florida, and AZ markets as their weather is the most favorable and they already have legislation in place to allow it. If all this is correct Tesla will not ever sell one of these (Robotaxi car) to a 3rd party. It sure is convenient that they have 30Bn in cash with no debt to self fund this operation. Q1 + Q2 are going to be bad (maybe even really bad) financially, but maybe, just maybe Elon isn't all that crazy anymore.
Short Uber/Lyft...long Tesla.
So the trial doesn’t just sell FSD, it samples more data from more drivers in more locations and situations to make FSD even better. And this just as Tesla is no longer constrained on training compute.
It’ll be interesting to see just how good FSD becomes by 8/8…
If it's a new machine, I suggest beta testing should start earlier...
Also, a random data point. I am driving to Dallas and saw 5 cyber trucks (4 on carrier going West and one waiting to supercharge)
Last edited:
Are you kidding? It's being trained right now with FSD v12.3.3. They've been training it for many years now.
Krugerrand
Meow
The Cat doesn’t agree with Elon about a lot of things. Like why he wants to save humanity in the first place is a mystery. I’d just as soon a zombie apocalypse happen and wipe most of mankind from the face of the planet.- and maybe when the cat finally doesn't agree with Elon on one single thing.
The difference between The Cat and so many others is that The Cat doesn’t need to let everyone know how it feels about irrelevant Elon bs. The Cat also simply doesn’t care, even a little bit, about anything that doesn’t have to do directly with the business, its financial strength, its expansion, and continued march to the goal. Most definitely doesn’t care about non-provable, opinionated statements being framed in any sort of an agenda way.
If Elon wants to pierce his nipples and tattoo ‘Free The Unicorns’ on his chest, don’t care. Vote for a Kardashian for president, don’t care. Tweet from his throne, piss off Baby Yoda, snort jelly beans, don’t care. Print up t-shirts that says Eat Cats and sell them for $69.69 each - okay, that one I might take some issue with, but if it said Eat Dogs, don’t care.
Yeah, some absurdity right there, but that’s how I see so many people who get bent out of shape about him - absurd.
2daMoon
Mostly Harmless
Folks, cool yer jetz.
He said the RT will be "unveiled" on 8/8, not that it will be put into service.
Keep in mind how much time passed between the CT unveil and first delivery off the production line.
Like the CT introduction, the hype for the RT will be stirring over some period of time, and this should raise awareness for Tesla and the concept of autonomous vehicles. Perhaps FSD-SuperV will get more attention.
The seed has now been planted. Fertilizer will be added on 8/8 and that seed will begin putting out roots. It will get the haters and the hopers talking, arguing, whatever, for however long it takes to make it ready for prime time. As happened with the CT.
It seems to me we probably won't have to wait as long as we did with the CT, and, however long the wait is, the anticipation will grow. Only it won't be focused on automobile consumers, rather it will get the attention of those bigger players who would like to have a significant piece of that pie.
If they can't buy the RT, what will they buy in the interim?
HODL
He said the RT will be "unveiled" on 8/8, not that it will be put into service.
Keep in mind how much time passed between the CT unveil and first delivery off the production line.
Like the CT introduction, the hype for the RT will be stirring over some period of time, and this should raise awareness for Tesla and the concept of autonomous vehicles. Perhaps FSD-SuperV will get more attention.
The seed has now been planted. Fertilizer will be added on 8/8 and that seed will begin putting out roots. It will get the haters and the hopers talking, arguing, whatever, for however long it takes to make it ready for prime time. As happened with the CT.
It seems to me we probably won't have to wait as long as we did with the CT, and, however long the wait is, the anticipation will grow. Only it won't be focused on automobile consumers, rather it will get the attention of those bigger players who would like to have a significant piece of that pie.
If they can't buy the RT, what will they buy in the interim?
HODL
Last edited:
2daMoon
Mostly Harmless
If it's a new machine, I suggest beta testing should start earlier...
You would think by now it would be using the Spank AI neural net,
in which case it will be Supervised Spank 
, not beta.
, not beta.
Last edited:
25% take rate for Robotaxi Network seems way too optimistic. I think it will be 10% or less. But even 1/3 of that graph is pretty impressive.
dhanson865
Well-Known Member
I thought he meant 25% of the take is the rate Tesla would get, poor shortening to put it on the slide as "take rate".
As in if they charge $1 a mile to the passenger, Tesla gets 25 cents per mile.
or if they charge 30 cents a mile Tesla gets 7.5 cents per mile.
Not as in 25% of the Tesla's on the road are used as robotaxis. I think his assumption is that 100% of them are robotaxis.
Are you saying that is too big a cut of revenue or are you on the "take rate" concept?
Last edited:
I thought it meant "25% of people will opt into the robotaxi network". That seems high to me.
Tesla taking a 25% cut seems very reasonable
dhanson865
Well-Known Member
OK, confirmed, poor label on the slide. See assumption #5
The assumptions:
1) Tesla sells their vehicles at cost and only makes money from FSD software subscriptions and Robotaxi revenue.
2) Tesla has sold a cumulative 4.9 million vehicles as of 2023 Q2 and will sell a cumulative 130 million vehicles by the end of 2035 (19 million in that year - 5 years later and 1 million less than the company's 20 million target by 2030). These vehicles are considered to be the "fleet."
3) The CERN model excludes all other revenue from Energy, Insurance, Supercharging, Service, DOJO and Bots.
4) All vehicles subscribe to FSD (no purchases) and the cost of monthly FSD subscriptions are $200.
5) Autonomous vehicles drive 60,000 miles per year, generate $0.30 per mile in revenue and Tesla takes 25% of the fleet revenue.
6) Net profit margins over time are 35%.
7) And investors give the company a P/E multiple of 30 for steady, growing, high margin earnings. None of the numbers in the model are discounted to the present.
MartinAustin
Active Member
I finally read the Reuters article, and they are doubling down... incorporating Elon's tweets. Wow, it is a scatterblasting hit piece of a calibre Kolodny could only dream about. Trying to hit so many targets 
Todd Burch
14-Year Member
Tell me you were out hiking all day without telling me you were out hiking all day.
At step 3, I was still thinking you were serious. Then I knew you were not. Though I still pictured that scene in my mind, with my MY and the front yard and road of my house.
Today is already Apr 5, not 1st. Please.
Musskiah
DisGruntled
Daimler Truck’s Chief Executive Takes Tesla’s Semi Down a Peg
Competing in big rigs is a ‘very, very huge undertaking,’ CEO Martin Daum says.
www.bloomberg.com
This sounds like 2014-2016 all over again when German automotive companies minimized the threat Tesla posed, and then...
Tesla Model 3 wreaks havoc on Mercedes-Benz and BMW in luxury sedan segment
Since the Tesla Model 3 began initial deliveries in Summer 2017, figures show that the United States’ ten most popular petrol-powered luxury sedans have seen a decline in popularity. All ten of the sedans have experienced a drop in sales when comparing numbers from 2017 and 2019. The sales...
www.google.com
Not sure why the x-axis has uneven intervals (some 10 months others 3 months), but assuming the lines/points are still accurate and putting in actual dates from the quarterly reports, the previous increase started in March 2023 with the rollout of single stack 11.x including freeways. FSD Beta reached ~100M cumulative miles around January 2023, so around 15 months to 10x to 1B miles. Here's some estimated monthly mileage numbers based on previous reports (these seem to be end-of-month numbers) and a simple extrapolation from the estimated partial April 2024 numbers forwards through August.
Of course actual mileage leading up to 8/8 will likely fluctuate based on how many people acquire FSD Capability after the 1-month trial as well as ongoing 12.x improvements that could result in even higher ongoing usage. But it seems like there's potential to double to 2B miles by August.
So, on the one hand, we have your question.
On the other hand, we have many people quite worried about Tesla's gap between production and deliveries, and their growing inventory. Estimates say 150,000 vehicles total...roughly a third of a quarter's production.
I'm not drawing any conclusions. I'm not even sure which hat I am wearing right now. Comedian? Fool? Wild optimist? Insightful genius? Intuitive seer? Random troll? Just an everyday handsome devil?
Similar threads
- Locked
- Locked
