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

[Thekiwi]

Weekend ATH cheers!

And with that, Elon tweeted about something I believe people are not paying enough attention to:


Here is my ATH hyper bull case:

• Tesla is going to become the new Google+AWS for the era of software 2.0

What is means:

• AWS generated 35B revenue last year and contributed more than half of AMZN’s profit, it’s a very lucrative business model.
• I think Cloud computing for Software 2.0 (aka AI in hype word terminology) would be even bigger pretty soon.
• Tesla would have opportunity to provide services on top of their physical world understanding, also able to provide computing 2.0 as service.

Here is why Tesla:

• Cloud computing for AI is still an area everyone is fighting for, but no clear winner yet.
• Tesla with FSD chip and now Dojo will be one of the best platform for inference and training for image/video understanding, aka physical world understanding.(THE best if you ask me)
• Tesla would have millions of eyes on the streets all the time. With Tesla’s data source lead, they will become the leader in real world understanding. Google’s attempts for physical world understanding for example street view etc would look like toys in comparison.
  
• Tesla with the vast amount of real world data, combined with battle tested hardware would be able to attract best talents to join (their mission statement helps too). This alone would be a very important factor for success.

I suspect when Tesla decided to design their own FSD chip, they already have this in mind. Marginal cost for producing chips is very low compared to R&D costs, once they have the capacity to build data centers with their own chip, the margin would be astonishing.

I assign this to a hyper bull case, not because I doubt Tesla’s capability of doing it, only because it seems to not align well with the current mission.

It might happen when Tesla expand their mission again though, let’s see.

Can you put this into english? What is "software 2.0" and how can Tesla FSD efforts lead to an AWS level cloud computing solution for it? Describe it to me as you would to a small child.

 

[printf42]

Can you put this into english? What is "software 2.0" and how can Tesla FSD efforts lead to an AWS level cloud computing solution for it? Describe it to me as you would to a small child.

For anyone who is not familiar with software 2.0 already, you need to do more research on FSD front. Search for Andrej Karpathy on YouTube and watch all video you can find.

Basic idea is, instead of telling the computer exactly what to do at each and every situation, you feed your program a vast amount of data with input and wanted(sometimes also unwanted) outcome, and let the program learn how to handle each situation.

Tesla FSD computer is a very specialized computer designed to handle image/video recognition and software 2.0. It have potential to be useful in many many other applications, not just FSD. Hence the potential for selling the services.

Traditional CPUs are hopeless in handling software 2.0 workload, even those GPUs from nvidia are not as good and efficient as what Tesla(and Google and Apple) is building.

Tesla FSD chip is already battle tested that it’s good for inference work load.
When Tesla got their Dojo ready and solve the training part, it could be a sea change on what is possible in deep learning workflows, if they are willing to open, there will be people lining up to try it.

 

[Tslynk67]

Interesting @Lycanthrope .

Were your numbers arbitrary or was there a rationale for how you calculated the numbers ? Each GTC order looks to be about 10-15% above the previous level.

What have you adjusted the numbers to now with the "old spliterrino" ?

I set these ascending orders to catch any sudden, violent short squeeze. My target price to "sell out" is $4k, so that's the starting point, but of course lots of people would sell at $4k, and most people set orders at such psychological price-points, so I took a number just below, in the 3900's, the 57 coming from "57 Varieties" from Heinz baked-beans, it's the random answer I always give people, 57, and that's where it's from.

Then I created an Excel and played around with %ages, but found that mechanism was a bit too exponential, an increment seemed better and I settled on 657 (there's the 57 again) as this gives a nice spread and, if all trigger, got me around $4m, which would be "enough".

Guess I could just divide the same numbers by 5 and x5 the shares to set it up after the split, or keep the tranches of 50 and have a finer grain of orders, but they are a bit cumbersome to setup each time, and I'm always terrified of accidentally putting a market sell not a limit, so I scrutinise carefully before committing each order.

One thing's for sure, I'm more bullish on Tesla than ever.

 

[Dutchie]

Possibly because Dojos been known about for quite a while already. It's not news.

Dojo was explicitly mentioned at Autonomy day early 2019.

Here's a reddit thread from late 2019 linking to other elon tweets about it from 2019, a presentation from Karpathey in 2019 about it, and the stuff from Omar about it.

Tesla's Dojo Computer : teslamotors




This one's open to tons of speculation on if he means more existing PW, or some next-gen product using something announced at battery day (and hence why orders are being pushed to 2021 as some are reporting).




That one was weird- since that update began rolling out weeks ago already... either he's been so busy with SpaceX he wasn't aware of that, or he means it's rolling wider than it has been and it's "coming soon" to the % of fleet that hasn't gotten that update yet.




I mean... that's always true





See above- this has been known for well over a year now (and Elons been asking AI folks to apply at Tesla for at least as long)




Also 100% already known info- though at least the 6-10 weeks remains "in line" with his previous time estimate for the initial release, so I guess "there's been no unexpected delay in the last couple months" is news?





This one is... eh, kinda also not news? He said a while back discussion the re-write they're expecting feature complete end of this year (which 6-10 weeks gets you) and then perhaps another year to get GOOD at it.... so again the only news I guess is there's no change from previous timelines given.




Again not at all news.... he's been telling folks this for WELL over a year- that the system will be able to avoid potholes, here's one of the many times he mentioned it from April 2019-
https://twitter.com/kkarthi002/status/1114721775877742592




This one approaches being actual news Or at least a timeline change/update to previously known news... personally I don't care about 2FA, but I do care if Tesla has finally gotten their internal IT out of the stone age, because it's been the cause of a lot of other problems (the cars getting sold at auction by Tesla with FSD and then having it pulled later for example- that's 100% an internal IT screw up.... as well as lots of other issues with sales, parts, service, etc that're well beyond scope of this thread).... so if they're on new, much better, internal IT systems now that's excellent news for the company.





I expect this'll get added to the games as a thing people complain about Tesla doing instead of improving FSD... Much like the games though this requires very very very little actual human effort to roll out so it's not a valid criticism.

It is the one time I'll be sad mine was built before the external speaker though.... I wonder how easy it is to add?




Yup, see above, this is already known stuff.... the cars getting campaigns to "send pics of X" has been going on before even Dojo was a thing- Greentheonly describes it in his "debunking shadow mode" stuff.

Elon tweeted that it will take another year before DOJO V1.0 will be ready. I thought that project was essential for software 2.0 and FSD. Software 2.0 will be released in about 10 weeks (as per tweet Elon) so before DOJO is operational. I thought I had a grasp of everything, but apparently not. Can someone please explain what DOJO has to do with FSD and why FSD will be released before DOJO is operational?

 

[Tes La Ferrari]

I set these ascending orders to catch any sudden, violent short squeeze. My target price to "sell out" is $4k, so that's the starting point, but of course lots of people would sell at $4k, and most people set orders at such psychological price-points, so I took a number just below, in the 3900's, the 57 coming from "57 Varieties" from Heinz baked-beans, it's the random answer I always give people, 57, and that's where it's from.

Then I created an Excel and played around with %ages, but found that mechanism was a bit too exponential, an increment seemed better and I settled on 657 (there's the 57 again) as this gives a nice spread and, if all trigger, got me around $4m, which would be "enough".

Guess I could just divide the same numbers by 5 and x5 the shares to set it up after the split, or keep the tranches of 50 and have a finer grain of orders, but they are a bit cumbersome to setup each time, and I'm always terrified of accidentally putting a market sell not a limit, so I scrutinise carefully before committing each order.

One thing's for sure, I'm more bullish on Tesla than ever.

I agree. Even though I am long TSLA, I don’t mind helping the mission by doing my part to hasten the demise of the shorts, if it is a sizeable profit via a short squeeze.

Your idea of chopping up your sell limit orders into tranches at progressively higher prices is a solid one. I was thinking about the prices that I might set mine up at...

and if some of the tranches got triggered on a squeeze, I’d buy back in on the next opportune time. I think the key for a long doing this is to set your limit orders fairly high. Ie I wouldn’t sell today if it hit 1800.

not advice

 

[Runarbt]

Elon tweeted that it will take another year before DOJO V1.0 will be ready. I thought that project was essential for software 2.0 and FSD. Software 2.0 will be released in about 10 weeks (as per tweet Elon) so before DOJO is operational. I thought I had a grasp of everything, but apparently not. Can someone please explain what DOJO has to do with FSD and why FSD will be released before DOJO is operational?

FSD run in the cars..Dojo run at Teslas upcoming supercomputer, and request/sort/process and learn from video all FSD cars produce.

Assume it will then feed FSD with what it learns, to improve FSD continously.

 

[new_sneakers]

Hi all,

Since some member posted earlier that Elon said something along the lines of "Tesla is betting the company on robotaxi's" during Autonomy Day, I went back and rewatched most of it.

However you feel about FSD, the presentation contains many little tidbits that are telling about the core business of Tesla. Since I was delighted to pick up on these (again), I thought I'd share.

Production rates
A lot has been said regarding production ramp. The chip guy mentioned that the cost reduction of having their own NN computer in cars times "a million cars a year" really adds up. He was referring to an anual production rate of Tesla vehicles, which we have not reached yet. Ipso facto: in 2020 or at the latest 2021 Tesla will produce 1 million cars a year. (By 2022 the next iteration of the FSD computer should be ready according to Elon ("three years from now" = april 2022))

Later on, Elon mentioned a Tesla fleet of "millions, tens of millions" of cars. Again it is clear to me he is envisioning where Tesla will be in a few years time and he is seeing it big.

In 2020 1M cars produced might not be possible, but ending 2020 on 1M annual production rate might get close. Fremont +GF1 = at least 360k (look at 2019) and China should match that. That's 700k cars in 2020. Any improvements and we'll end rather close to a million cars a year. (Personally I'd say we'll reach 1M capacity when GF4 goes online, which should be by the end of 2021 at the latest).

GF5 is still a question mark, so I won't speculate. But GF1/Fremont, GF3 and GF4 could produce 500k cars annually each by 2022, when the next FSD chip is built in to the cars. (They are made in Samsung, Texas, I forgot that detail) So in 2022 I'm guessing 1,5 million yearly production.

Elon did add the following regarding future production ramp:
In 2013 they produced 23k cars, in 2018 250k cars, which is a 10x increase "in five years" (it's actually in six years). Elon expects the same increase to happen over "the next five years", so (own calculation) around 2,5 million cars per year produced by 2023/2024.


Which models will get produced when?

I've seen some recent discussion regarding semi and model Y production. I think the Autonomy day was pretty straightforward about this:

View attachment 497029

2020 is indeed the year of the model Y and the semi. Semi might be postponed a bit more, since getting profitability and FSD up is more important in the short-term, and Model Y/3 are the cars to make that happen.

After semi there is of course Cybertruck, and after that just robotaxi's without steering wheels. Tesla expects to own their own fleet of robotaxi's counting roughly one million vehicles. Robotaxi = current model at first but later on without steering wheel/pedals, costing 25k to make, but worth 200k to Tesla.


Regarding FSD

Elon broke down FSD in three steps:

1) feature complete = FSD software can be activated anywhere and (theoretically) drive the whole trip. In april Elon expected this by end of 2019.
2) Tesla approved FSD = the point at which Tesla believes it is safe to fall asleep behind the wheel. In april Elon expected this by Q2 2020. (!)
3) Regulator approved FSD = the point at which FSD is allowed by law since it is believed to be more safe than humans and/or safe enough to be used by the public. In april Elon expected this by end of 2020.

The feature complete deadline (by end of 2019) was not met, so I'm keeping my eyes peeled for how many quarters this is delayed.

I had forgotten Elon's agressive timeline (Q2 for Tesla-approved FSD, crazy) but was amazed with the confidence with which he uttered it. I don't want to start an entire "will FSD happen"-thread, but I do believe Tesla-approved FSD in the USA (big emphasis) could be as close as early 2021.

Adam Jonas asked Elon (3:42:10): Can you share with us how much Tesla is spending on Autopilot or autonomous technology by order of magnitude on annual basis?

Elon replied: "it's basically our entire expense structure." (silence)

This is not betting the company though, since earlier in the Autonomy Day he said: during robotaxi ramp-up we expect to be cash flow neutral. After FSD is turned on (=regulatory approval) we will be hugely cash flow positive from then onward.



The above might be some random thoughts put together, but I thought you (TMC) would be interested. I am sticking with my investment, that's for sure!

In the shadows of battery day, it is worth remembering how much information about Tesla’s business can be gleaned from Autonomy Day. Thanks @jeewee3000 for this summary from January.

It is also worth remembering that Elon said last October that Plaid would enter production in summer 2020.

 

[Beezer_9]

https://twitter.com/elonmusk/status/1294903194997252096?s=21

Elon just posted this.
2020 theme song. Queen - Under pressure

 

[dakh]

Elon tweeted that it will take another year before DOJO V1.0 will be ready. I thought that project was essential for software 2.0 and FSD. Software 2.0 will be released in about 10 weeks (as per tweet Elon) so before DOJO is operational. I thought I had a grasp of everything, but apparently not. Can someone please explain what DOJO has to do with FSD and why FSD will be released before DOJO is operational?

There's training and inference. Inference happens when driving, this part uses a pre-trained neural network. Dojo will make training faster and cheaper. Current ways to train NNs also works, it just isn't as efficient as what they can do with their super-specialized hardware.

 

[Tslynk67]

I agree. Even though I am long TSLA, I don’t mind helping the mission by doing my part to hasten the demise of the shorts, if it is a sizeable profit via a short squeeze.

Your idea of chopping up your sell limit orders into tranches at progressively higher prices is a solid one. I was thinking about the prices that I might set mine up at...

and if some of the tranches got triggered on a squeeze, I’d buy back in on the next opportune time. I think the key for a long doing this is to set your limit orders fairly high. Ie I wouldn’t sell today if it hit 1800.

not advice

No, wouldn't sell today at 1800, but beginning of September, maybe...

 

[Oveeus]

I think that given what Musk has been doing on sharing some inventions on his many companies (SpaceX alloy on CT), we'd see more collab between them in the near future.

More specifically with SpaceX:
I believe Starlink will one day finds its way onto Tesla. In fact, I'd believe once Starlink is operational, Tesla would switch its connectivity to Starlink. The reason is quite simple, not only Starlink would suddenly have a huge customer base, Tesla kinda vertically integrates its connectivity... but its applications would also be incredible to say the least.

See how Google Streetview totally changed our way of using maps? Imagine instead of a yearly or bi-yearly update, you have a daily update? By the time Starlink is operational, there'd be millions of Tesla on road with millions coming every year. Being able to collect real world data in quasi-real time would be priceless. FSD would work much better when a pothole is reported by some cars so other cars can avoid it when passing the same spot.

Currently, it's cost-prohibitive for Tesla to transfer all interesting findings like traffic jam, pothole, debris, and more that Tesla cars see back to its center in realtime. Starlink solves that big bottleneck in the middle.

Not only can Tesla use that data, it can also sell that data feed to others. I'm sure Google would be the first in line to buy those.

 

[Christine69420]

Are you saying, the rest of us have to do something to get the extra shares??!!

No.

 

[Artful Dodger]

Seriously, how stupid can a company be? The bailout failed because they were unwilling to accept to unable to actually produce more electric cars?

FTFY. They have no ability to increase their battery supply.

Cheers!

 

[dakh]

I agree. Even though I am long TSLA, I don’t mind helping the mission by doing my part to hasten the demise of the shorts, if it is a sizeable profit via a short squeeze.

Your idea of chopping up your sell limit orders into tranches at progressively higher prices is a solid one. I was thinking about the prices that I might set mine up at...

and if some of the tranches got triggered on a squeeze, I’d buy back in on the next opportune time. I think the key for a long doing this is to set your limit orders fairly high. Ie I wouldn’t sell today if it hit 1800.

not advice

I just sell covered calls at certain strike prices. If my shares get called at those prices, great, that's what I wanted anyway. If not, I get paid to keep 'em.

 

[MikeAtkinson]

https://twitter.com/elonmusk/status/1294903194997252096?s=21

Elon just posted this.
2020 theme song. Queen - Under pressure

Seems to have been deleted.

 

[CaliBear1]

And we think there's a pothole problem now?!

Any road accident would be a potential nuclear disaster

Any road accident would be a potential nuclear disaster

 

[LN1_Casey]

Seems to have been deleted.

I saw it before it was, he had said, "Year 2020 theme song" or something similar. I'd be curious as to why it was deleted, didn't seem like something was wrong. Linking to a song from an official page is generally encouraged.

 

[CaliBear1]

What a great personalized license plate!

That decision confirms and further accelerates their eventual bankwuptcy.
But fine. The industry needs those that can EVolve.

Edit: Good on UK govt for adding the decarbonization requirements.

 

[HG Wells]

Exactly this psychology. Ask 100 random people on the street if a stock is more likely to go from 400 to 800 or 2000 to 4000 and I’ll bet money the majority say the former.

Well, that's obvious.
There are far far more 400 dollar stocks than 2000 dollar stocks.
Over a 5 year period Id bet it would be a order of magnitude more 400 dollar stocks dubling than 2000 dollar stocks doubling.

Now if you were talking about a single stock that is a different question, Then they are very close but the 400 would have a slight advantage as fewer people would buy shares as fractional trading is not offered at all brokerages.

 

[OrthoSurg]

I agree. Even though I am long TSLA, I don’t mind helping the mission by doing my part to hasten the demise of the shorts, if it is a sizeable profit via a short squeeze.

Your idea of chopping up your sell limit orders into tranches at progressively higher prices is a solid one. I was thinking about the prices that I might set mine up at...

and if some of the tranches got triggered on a squeeze, I’d buy back in on the next opportune time. I think the key for a long doing this is to set your limit orders fairly high. Ie I wouldn’t sell today if it hit 1800.

not advice

ALL the times I sold it managed to get higher.
That’s the magic with TSLA