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

[Singuy]

Let's assume @Artful Dodger 's theory about the dividend-split forcing the hands of the MM's to "legitimize" their naked shorting (by borrowing the shares and/or covering them). The current rally maybe an indication of that squeeze as it seems to be bigger than a usual reaction to stock-split (see AAPL). So what can we expect in the near-future ?
Here is my hypothesis (mind you I don't really know what I am talking about, so definitely do not take it as advice):

• Naked short squeeze continues until this Friday (21st) driven by MM's to legitimize their shenanigans
• Next week we trade mostly side-ways just following macros with usual high-beta multiple as the squeeze pressure is no longer in play, but MMs still cannot return to naked shorting until the split is executed (otherwise they would still need to cough up the extra 4 shares for each naked shorted one)
• The week following the split from 31st they can get back into naked shorting, plus all the traders who were front-running the split would be inclined to take profit, so I would expect a dip there
• Whenever the S&P500 committee pulls their *sugar* together and make the announcement of TSLA inclusion will trigger the next big run-up
• Battery day 22nd Sept will be another milestone which will trigger a runup and a following dip in the usual buy-the-rumor-sell-the-news style

So these are my expectations for the next month or so, although I have been known to be wrong about SP movements...

Can someone explain how naked short selling is profitable for MM? Seems like it's nothing but a dumpster fire strategy for them.

 

[ammulder]

With all this talk about GM doing an EV spin-off, it seems like the special GM task force from 2013 to study Tesla is finally coming up with results... Who remembers this?

General Motors takes a look under Tesla's hood

Famous last words (from that article, uttered in 2013):

This source, who asked not to be identified because of the sensitive nature of the task force and its work, said: “The fact that GM is studying Tesla should flatter Elon Musk, but it should also scare the *sugar* out of him. When you wake up the sleeping giant, look out.”
​

Mm-hmm. Elon really dodged a bullet there!

 

[OrthoSurg]

I'm dividing into smaller chunks and using them up in dips.Then I'm going to be strong and HODL

I had placed an order to buy 200 shares 2 weeks ago when it dipped to $1370. By the time I logged in it was up to $1430. Now I have that money burning in my hands

 

[humbaba]

I don't think Dojo is necessary for FSD, it just makes the training process faster so they can iterate/test to improve it faster than with their current compute setup.

For example Andrew Karpathy said that "A full build of Autopilot neural networks involves 48 networks that take 70,000 GPU hours to train."

So what you are saying is that it would take even longer without Dojo. As you aren't contesting my take on it that 2024 with Dojo seems a stretch, so without Dojo it would take even longer, right?

 

[StealthP3D]

Which reinforces my FSD skepticism. ARK's high valuation of Tesla is predicated on true autonomy (robotaxi network) by 2024.

You are wrong about that. ARK's "base case" 5-year share price target is based on zero FSD/robo-taxi revenue.

 

[humbaba]

You are wrong about that. ARK's "base case" 5-year share price target is based on zero FSD/robo-taxi revenue.

Ah, but that is their base case. Did I not reference their bull case? If so, my bad. Their base case is actually not that far from where were at.

Now, ARK is not accounting for Tesla Energy as far as I can tell. I'm not trying to argue for a low valuation or anything like that. Just trying to think about what lower bound is implied on timeline for FSD given what we know. And I think that it is fair to set a lower bound of Dojo being in production use for some span of time.

edit: yes, I said "high valuation" instead of bull case.

edit2: to be clear, when I say "FSD" I mean "robotaxi" not "feature complete" based on the features Tesla has actually advertised which might well come sooner

 

[Dangerous Fish]

I had placed an order to buy 200 shares 2 weeks ago when it dipped to $1370. By the time I logged in it was up to $1430. Now I have that money burning in my hands

Well, I don't think we can say "I'm not going to buy if they're more than $x".
I've bought as low as $185 and as high as $1835.
I'm still adding to my core amount even at today's prices.
Why not?

 

[Tslynk67]

Zuck is a smart guy who wanted to get laid and lucked into a new type of media. Elon is a real visionary.

Have to admit I'm with Zuck on this one...

 

[MP3Mike]

So what you are saying is that it would take even longer without Dojo. As you aren't contesting my take on it that 2024 with Dojo seems a stretch, so without Dojo it would take even longer, right?

How is saying Dojo isn't necessary for FSD not contesting you saying that it is?

Let's say they currently have 500 GPUs, which is only 250 AP2.5 units, that would mean a build currently takes ~6 days. The lack of Dojo is not going to stop them from getting to FSD.

 

[StealthP3D]

So what you are saying is that it would take even longer without Dojo. As you aren't contesting my take on it that 2024 with Dojo seems a stretch, so without Dojo it would take even longer, right?

It's more a matter of growing the training network as computer power becomes cheaper. FSD isn't the go/no-go situation it's often made out to be. Even after Tesla achieves what could be deemed to be FSD, they will still constantly improve it's safety and capabilities.

I don't think Tesla knows with any certainty at what point it will be sufficient for regulatory approval, simply that they will be constantly improving and refining it over the next 5 years so it makes sense to keep the training network state-of-the-art through the entire training process. It will always be a work in progress.

 

[dl003]

Famous last words (from that article, uttered in 2013):

This source, who asked not to be identified because of the sensitive nature of the task force and its work, said: “The fact that GM is studying Tesla should flatter Elon Musk, but it should also scare the *sugar* out of him. When you wake up the sleeping giant, look out.”
​

Mm-hmm. Elon really dodged a bullet there!

Being in America, and in reference to America, I automatically associate "sleeping giants" with 700lb dudes sleeping in a bed of hot Cheetos. You don't want to wake them up, that's for sure, but not for reasons hereby implied.

 

[JSML]

So what you are saying is that it would take even longer without Dojo. As you aren't contesting my take on it that 2024 with Dojo seems a stretch, so without Dojo it would take even longer, right?

I've learned not to understimate Elon. 2024 sounds about right to me for some form of robo taxis on the road.

My take is that Telsa will need the goverment's co-operation not only on autonomous driving approval but also general road rules to make autonomous driving a bit more easier to achieve. A few regions in California or Texas with robo taxis in 2024? I wouldn't bet against it.

 

[YAWD]

Man I wish if we could stop talking when FSD happens and only talk about if the rate of progress on FSD is accelerating or stagnating. How on earth does somebody predict timeline for a technology like FSD is beyond me. Its like somebody constantly nagging SpaceX on when are they getting to Mars

 

[lafrisbee]

The split can't come soon enough. My mind keeps refusing to process the numbers I am seeing correctly. I keep processing the rises and falls as if the stock price was much lower. Fifty Dollars used to mean something.
This compounding crap is real.

 

[LTC_RRR]

OT:

SpaceX launch this morning, happening now

I've seen them all and am still awestruck. Today's first: 6x reuse of the stage 1 vehicle and 2x use of the fairings. Incredible!

 

[NicoV]

Thinking about the timeline for Dojo has me wondering: how can Tesla be close to FSD when they are still years away from having their supercomputer? If they could really reach FSD by -- say -- end of 2021 then what would be the point of Dojo?

Then thinking about the scale of Dojo -- it will be a competitive supercomputer whether or not it ranks first -- if Tesla believes they need Dojo then what does that imply about pre-Dojo progress?

Which reinforces my FSD skepticism. ARK's high valuation of Tesla is predicated on true autonomy (robotaxi network) by 2024. No, I'm not losing any conviction (especially because I never had any to begin with when it comes to FSD) but if Dojo is currently at sim levels using FPGA what does that suggest about timeline before it can actually be made? And how long will it need to be operational before it can produce FSD? I realize the latter is an open ended question, but I would think someone familiar with chip design and production could hazard the former.

Don't get me wrong, I love my autopilot and am looking forward to (hopefully) having access to the rewrite this year or early next. But I just don't see how its plausible to have everything lined up and done by 2024 for a full self driving robotaxi network.

Remember Elon talking about the march of the 9s? As more and more training data is gathered to cover the corner cases, more computing power will be needed to build and validate the new neural networks. I guess that’s what Project Dojo is about: having enough compute power to enable more 9’s of reliability.

 

[StealthP3D]

Can someone explain how naked short selling is profitable for MM? Seems like it's nothing but a dumpster fire strategy for them.

The volatility of a stock like TSLA is great enough that short selling can still be profitable even though the overall trajectory is up. The fact that the same people manipulating the stock are also short-selling gives them a huge advantage in terms of when it's profitable to initiate and close positions. Market makers also have the best and most instantaneous data on what the market is doing as the orders are placed.

 

[Artful Dodger]

Thinking about the timeline for Dojo has me wondering: how can Tesla be close to FSD when they are still years away from having their supercomputer? If they could really reach FSD by -- say -- end of 2021 then what would be the point of Dojo?

You could drive legally at 16 yrs old w. just a few hours of training. Yet there still exist professional drivings schools and graduated racing licences. Why?

To get better.

 

[H Mak]

Thought it was successful lift off for $TSLA and are you watching Falcon 9.....

I can switch between trajectories

 

[Nocturnal]

Remember Elon talking about the march of the 9s? As more and more training data is gathered to cover the corner cases, more computing power will be needed to build and validate the new neural networks. I guess that’s what Project Dojo is about: having enough compute power to enable more 9’s of reliability.

Also, level 4 FSD vs level 5 FSD, several magnitudes of reliability requirements between them.