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

[Blue horseshoe]

https://twitter.com/x/status/1769931763252347334

Per Elon, three significant fsd updates coming every two weeks... should be shining really bright by late April early may.

 

[2daMoon]

Following up with a video from Tom Nash, discussing a Tom Lee interview and raving about the 6 Trillion dollars in cash sitting on the sidelines.

Particularly listen to how he describes the "small caps" in regard to Beta and fundamentals. Sounds like a company I like.

Bullish as... (Ferdinand?)

 

[heltok]

https://twitter.com/x/status/1769931763252347334

Per Elon, three significant fsd updates coming every two weeks... should be shining really bright by late April early may.

I assume that Musk is saying that for the 6weeks we will have 3 major updates released biweekly. I would assume
1. Actually Smart Summon
2. Reverse in Traffic
3. ???

Gonna guess 3 is Single Stack highway, maybe they will replace autopilot with FSD v12 lane keep assist.

Anyway this meme is very true:

https://twitter.com/x/status/1768868127138549841

 

[Musskiah]

Following up with a video from Tom Nash, discussing a Tom Lee interview and raving about the 6 Trillion dollars in cash sitting on the sidelines.

Particularly listen to how he describes the "small caps" in regard to Beta and fundamentals. Sounds like a company I like.

Bullish as... (Ferdinand?)

(War Tom Tom crescendo)

 

[skybluecgreen]

One factor in the lowering of forecasted q1 deliveries is the apparently VERY slow ramp of the Fremont refreshed model 3. Considering how easy the transition was for Shanghai, I am wondering why the delay for Fremont. Could it be Tesla is figuring a way to make the refreshed 3 eligible in the US for the 7500 credit it lost Jan 1 and is delaying production until they do? Doesn't seem to make sense to buy a model 3 in the US right now when you don't get the credit but do on the Y.

 

[AndrewZ]

Anyone have thoughts on any impact Nvidia’s new GB200 product has on tesla? It’s 4x better at training, and 30x (!) better at inference than there current top of the line H100 solution.

The new product of course will be great for companies working on AI products (like Tesla) - but seems like it may be time for Tesla to give up on Dojo development maybe? Seems to be so far behind now that it’s a waste of R&D resources, especially when Tesla can just use GB200 on demand via cloud providers for its FSD training.

I'd be curious what impact Nvidia's new product has on Nvidia's own autonomous vehicle and robotics efforts.

 

[Singuy]

I'd be curious what impact Nvidia's new product has on Nvidia's own autonomous vehicle and robotics efforts.

Nvidia doesn't know how to manufacture and most training techniques didn't even spawn from them. Just think of it as they provide TSMC the blueprint to make an computer capable of really fast matrix math solving. People all around the world build software around such computer to train AI in multiple different ways.

So if you are looking for autonomous vehicles and robotics, look elsewhere. Look at companies that actually build assembly lines, manufacture from raw materials, and developed or utilize training technics. Nvidia just provide people with the tool to train. They generally don't do any specific training themselves.

 

[jamaatta]

https://twitter.com/x/status/1769931763252347334

Per Elon, three significant fsd updates coming every two weeks... should be shining really bright by late April early may.

Elon's notion combined with Ashok's earlier tweet makes me truly eager to see what they have in the bag (of FSD). I guess those features are already baked in their alpha version of software..?

https://twitter.com/x/status/1769257128756380003

This: "The next set of 12.x releases should bring unprecedented progress.'

 

[Buckminster]

https://twitter.com/x/status/1769963505346609260

 

[Tiger]

https://twitter.com/x/status/1769918328506863803

 

[Huskyf]

Tesla's mission is to get all the other car manufacturers to start making BEVs as well as Tesla. Are you under the impression they want to have a monopoly on BEVs?

Yes Tesla can't alone change all ICE cars to Electrics cars they need partners

 

[elasalle]

Following up with a video from Tom Nash, discussing a Tom Lee interview and raving about the 6 Trillion dollars in cash sitting on the sidelines.

Particularly listen to how he describes the "small caps" in regard to Beta and fundamentals. Sounds like a company I like.

Bullish as... (Ferdinand?)

With 5+% in interest off money market funds.. no one is dying to get back in
Cheers!!

 

[jerry33]

The most confusing thing about hybrids, to me, is that all the other manufacturers, INCLUDING TOYOTA, dragged their feet fo so darn long.

Costs for hybrid components really didn't take until now to come down. Comparing the cost of a Prius to a Camry, for example, the additional cost of the battery and electric motors (paired with an arguably simpler transmission) was pretty minimal over the past decade.

The 1st generation Prius came out in Japan in 1997, and North America in 2000. The Prius has been very much "a thing" for over 20 years, and Toyota offered hybrid versions of several other models for quite some time.

But, over 20 years after the hybrid mass-production began, in 2022, I think roughly 25% of the cars sold by Toyota were hybrids -- about 2.6 million Toyota hybrids sold that year. They upped it considerably for 2023 to 3.4 million hybrids. Good numbers...but why on earth didn't they hybrid EVERYTHING?

Toyota never intended to make the Prius. They had a problem in Japan in the 1980s where no one other than those over 50 and government would purchase a Toyota (similar to station wagons and mini vans in the U.S.). It was so bad that a 40% market share appeared to be an unobtainable goal. So they cut through the normal channels and brought their best engineers together to create technology for the 21st century. The goal was to have new tech that could be put into future cars and to show that Toyota was not just an old folk's car. The Prius was a kind of bonus. The launch of the Prius improved their reputation in Japan and once they got back to their previously normal market share, they went back to business as usual. After the 2004 Prius, they stopped any significant development on the Prius. They created a hybrid version of the Lexus but it's not much more efficient than the gas versions (or wasn't up to the time I stopped following Toyota). Of course, Toyota is now forced to do something because of worldwide pollution regulations, so they now have more hybrids.

 

[hobbes]

Interesting discussion... it seems that the assumption is that the training data is only allowing the NN to watch the driving behavior of others, be it real-world video, or simulation data.

But I don't see why the training data can't include the rules of the road, distilled into some form (from the DMV booklet, etc...) the NN can ingest. This is what humans start with when they are trained. It can also be weighted (i.e.- when visibility is low, err on the side of the written rules, etc...)... this is also what humans tend to do: conservatively do the "correct" thing as uncertainty goes up.

What made me think of this was AlphaZero, the chess AI that Google acquired. It was trained by first giving it the rules of the game, then letting it play itself, developing strategy. It quickly became the dominant chess playing force on the planet.

Of course, chess rules are immutable, whereas driving rules are not, hence the weighting I mentioned above. And the driving space is much more varied than an 8x8 board, so it makes sense to train on real-world data, rather than just letting it solely drive and learn, but my point is that it starts with the rules.

So... undoubtedly FSD started with hard-coded procedural rules... but as Tesla move more towards E2E NN's, I don't see why rules of the road don't simply become another data input for the NN to ingest, rather than something hard coded for a specific rule...

FSD beta 12.3 is still doing the NHTSA prescribed stops down to zero mph at stop signs. I don´t think that comes from training clips either (otherwise it would not be that strict). And watching how humanly it is driving now I am afraid NHTSA will un-prescribe more of these human-like things like pragmatically crossing solid lines..

 

[Mengy]

Meet Kevin put out a very bearish vid on TSLA this morning, he says he is shorting TSLA:

In his analysis though he completely ignores Megapack sales ramping along an S curve right now, not sure why? I honestly agree with his bearish outlook on Tesla car production, they just aren't building out capacity fast enough right now to ramp it much faster IMHO, but Tesla is much more than just a car company! Particularly with FSD improving as fast as it is right now, man if I was short TSLA I wouldn't be able to sleep at night. Sure the next two years might be rocky for the stock but the future outlook for Tesla has never been brighter than it is right now IMHO.

Anyone else watch this video?

 

[Unpilot]

Meet Kevin put out a very bearish vid on TSLA this morning, he says he is shorting TSLA:

In his analysis though he completely ignores Megapack sales ramping along an S curve right now, not sure why? I honestly agree with his bearish outlook on Tesla car production, they just aren't building out capacity fast enough right now to ramp it much faster IMHO, but Tesla is much more than just a car company! Particularly with FSD improving as fast as it is right now, man if I was short TSLA I wouldn't be able to sleep at night. Sure the next two years might be rocky for the stock but the future outlook for Tesla has never been brighter than it is right now IMHO.

Anyone else watch this video?

No

 

[replicant]

Particularly with FSD improving as fast as it is right now, man if I was short TSLA

Why? What's your bullish timeline for FSD to contribute significantly to Tesla's profit in the next 3 years?

 

[CHGolferJim]

The most confusing thing about hybrids, to me, is that all the other manufacturers, INCLUDING TOYOTA, dragged their feet fo so darn long.

Costs for hybrid components really didn't take until now to come down. Comparing the cost of a Prius to a Camry, for example, the additional cost of the battery and electric motors (paired with an arguably simpler transmission) was pretty minimal over the past decade.

The 1st generation Prius came out in Japan in 1997, and North America in 2000. The Prius has been very much "a thing" for over 20 years, and Toyota offered hybrid versions of several other models for quite some time.

But, over 20 years after the hybrid mass-production began, in 2022, I think roughly 25% of the cars sold by Toyota were hybrids -- about 2.6 million Toyota hybrids sold that year. They upped it considerably for 2023 to 3.4 million hybrids. Good numbers...but why on earth didn't they hybrid EVERYTHING?

Some of Toyota's marketing points out that they can make ~90 hybrids, or 6 plug-in hybrids, for the same kWh capacity needed for one electric car. Sure, the battery chemistries are different, but a hybrid like the Prius really only needs 1-1.5 kWh of battery. So, it's pretty true and it SOUNDS like great marketing...but it doesn't explain why, 25+ years after their first production hybrid, Toyota was only making about 1.5 hybrids for every EV that Tesla made and sold. The implication is that Tesla was able to source 50x as much battery capacity as Toyota. Tesla didn't even EXIST when Toyota started with the Prius. Logic tells us that, if Toyota had wanted to, the definitely could have sourced enough batteries to make EVERY SINGLE CAR they sold a hybrid (or plug-in hybrid) for the last 10 years, easily. They just...didn't. To me, that's an incredible failure of their planning, marketing, suppliers/sourcing, and for the environment.

Toyota has finally starting to turn towards actually pushing their hybrids on a wide scale, and making more than just the Prius be 100% hybrid. I learned recently, for example, that the Toyota minivan - the Sienna - is only available as a hybrid starting with the 2021 model year. Amazing MPG and very simple AWD option makes it a stand-out among the ICE minivans of the world. I believe Toyota offers good hybrid options on the Toyota trucks now too, as well as the corolla and camry and Rav4...but still goes to the engineering effort to offer a bunch of non-hybrid/ICE-only versions too.

But, again, I really don't get why they really are dragging their feet. Toyota might approach 50% hybrids soon...but had they wanted to, they could have gone to 100% hybrid 10 years ago, I think. And all the other manufacturers that are pivoting to hybrids today will probably be at 10% hybrids in a few years. It's just embarrassing....it's a lack of effort, and a lack of caring to actually reduce the environmental impact....I'm glad they're "finally" doing it, but it feels too late now.

Could be wrong, but it seems like the reluctance to go 100% hybrid in the early days was mostly related to consumer resistance to the hybrid price premium. Most older large companies outside of the tech area seem unwilling to get out ahead (iow to “lead”) of current consumer preferences.

 

[Stretch2727]

A point of curiosity - those that frequently mention FSD being licensed to other auto companies - what would you like to see as the arrangements of such a license? What do you think is reasonable for Tesla to be asking for?

My personal opinion is this will be difficult to accomplish given the integration of FSD into the entire car electrical architecture. There is a lot of hardware that would need to integrated into any car that had the system.

Easier path is Tesla builds the car with some restyling and brand customization and sells to the OEM. Much like GM does with the Honda Prologue.

I just don't see any OEM going either direction as they give up much of the value chain and the reason they exist!

 

[Mengy]

Why? What's your bullish timeline for FSD to contribute significantly to Tesla's profit in the next 3 years?

I have no idea when FSD will get to Level 5 and substantially contribute to earnings, but that's exactly the point. It could be three years from now or it could be later on this year in 2024. Nobody knows, but I do know I wouldn't want to be short on TSLA when it does happen.