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

[Knightshade]

It's frustrating that the market doesn't understand that high-quality data is the gold, not chips

The beauty of selling the chips is you don't have to care about the quality of the data being processed on them.

 

[GOVA]

The beauty of selling the chips is you don't have to care about the quality of the data being processed on them.

Post of particular merit. Simple yet profound.

 

[elasalle]

The beauty of selling the chips is you don't have to care about the quality of the data being processed on them.

During the gold rush, many people made more money than the miners themselves. These included merchants who sold supplies to the miners .....

 

[Unpilot]

Broke though the 190 resistance....big call wall there for Friday.

 

[cusetownusa]

Trick Question:
What color is this light?

View attachment 1016486

Just checking...

lol...that is only a few miles from me.

 

[UncaNed]

I like Dillon's style, and this episode was a particularly good one with a lot of information on so many current topics. A snippet from the most recent Joe Justice video, extended Ford coverage, CT in a fog, as well as a funny dig at Ross G from a youtuber toward the end. Enjoy

EDIT- Replaced the wrong video with this one above-

Definitely worth watching!! Amazing info about cost reductions and constantly improvements, Tesla's agile manufacturing is vastly more agile than I could ever have imagined thanks to Tesla's AI.

 

[GhostSkater]

Too far fetched? But why? My partner and her brother (both way more computer literate than me) have showed me how they can ask AI to, say, create a fantasy landscape, or some crazy fantasy creature, and in short order it does. Then they can adjust the inputs, and the AI spits out a quite different yet equally alluring image in just a few seconds. Now imagine if you have CAD designs of every electric motor ever built stored in your own data centers, AND have real-time data on defects, friction points, lot numbers, date of manufacture, etc, and AI can draw what it thinks a more efficient motor will be. Isn't that, in part, what computer aided design is all about? And now just add AI to it. CAD engineering has done wonders in the bicycle industry, just fantastic wonders. The modern mountain bike today is so superior to anything ever built even 10 years ago, I often wonder how mountain biking as a sport ever took off, the old mountain bikes being so crappy, looking back now.

In the last earning call Elon said Tesla is not going to reveal as much anymore, going to keep their cards closer to their chest. I think that's wise, don't let the competition know what's coming. Shareholders will eventually be pleased and surprised, and the competition won't know what hit them and will fall further behind.

This will be really interesting, my capstone project was engineering a in wheel hub motor to replace a geared setup in a lightweight electric vehicle

Once you start digging how much parameters there is to tweak and how they influence each other it gets a bit crazy, I’ve made a huge mind map to try to make sense of it all, it helped but still too much for you to take, and keep in mind this is with just limited geometry variations and keeping everything traditional design wise


In the end the replacement motor maintained every performance spec, including efficiency, at least on paper, sadly I didn’t manage to get the fund to actually build it, I was going to make a prototype adapted for the company I was working, but you guys know how that ended

But something that I see becoming even better engineering wise in Neuralink in maybe a few decades, many times you came up with the perfect design in your head, it’s there, ready to be made, but you need to spend two weeks in CAD translating if from .brain to .sldprt

Join that with AI and AR and you have a winner, imagine working in design in 3D as if you had your parts in front of you, being able to manipulate, immediately asking a dedicated AI to setup and run a lifetime fatigue analysis in X scenario and so on?

 

[Featsbeyond50]

The board needs to release a statement on the comp package decision and how it plans to handle it. Hopefully they had been working on one incase of this decision, so should be ready to go.

I think they will simply file an appeal, then it will become news.

 

[Krugerrand]

Broke though the 190 resistance....big call wall there for Friday.

So, short lived then.

 

[Crystal01]

Maybe I'm getting old, but after the initial hype of generative AI, I haven't seen as many posts about ChatGPT or related memes.

And almost all of the phone AI capabilities I've seen (Galaxy S24) are gimmicky and have little real-world practicality or desirability (stuff like cutting out background people or moving your kid in a scene after you've shot a pic).

Generative AI and LLMs in their current form and architecture aren't good enough yet for profitable applications, imo (still need to cross-check and confirm outputs). Personally, I think the AI hype right now is a bubble, especially in relation to AI training chips, where only a few large / expert businesses will have the capability, motivation, and data to do training.

When I look at the current market for AI capabilities and products, it seems like Tesla is the only one trying to achieve meaningful applications and has the expertise and infrastructure to do it.

It'd be nice to see if I'm wrong on this broad perspective though. I just don't see a profitable and useful outcome for generative AI in its current form, unless there's some new approaches (which may end up using less training resources btw).

You are quite correct in my view. Much of the LLM based AI is pure hype. The apps are superficial and easy to copy so no moat. Microsoft will dominate in volume simply due to the installed base of ms office and there is a serious threat to google search. There will be increasing barriers as people protect their data and try for royalties on the training data respected groups like Gartner think this first wave will all be done in two years or less. It’s the vertically integrated game changing AI for things such as FSD and other process driven apps that will truly impact us all. I am happy for the chip guys on their current roll. But make hay now. It’s a bubble and won’t last long. Tesla’s lead is not recognised in all the noise but I am very convinced that it will get there and unlike others they have the scale to make a difference. Tesla has the data. It has the technology - or nearly there - and it has the defined market and app need. Patience is very hard at times but persistence will pay off.

 

[SOULPEDL]

So, short lived then.

Or... we made it!

 

[FEERSUMENDJIN]

I commented on some such developments a few weeks back. At the moment Land Rover, Rivian and BYD are all being included in such plans. BYD has some advantage because they can and do package off-grid solar, storage and a variety of vehicles. I was informed some weeks ago that there'd been discussions with Tesla, but no hint of how they're progressing.

This business, safari support integrates everything and has the advantage of facing gigantically expensive fossil fuels. Maria Mara and Serengeti, contiguous parks in Kenya and Tanzania, are having enormous rises in tourism, but fossil fuels for both are trucked in, from Nairobi and Dar es Salaam respectively, on dirt tracks for the most part.

For Tesla such projects can be of major benefit in spreading the word to wealthy people in other countries.
As previously mentioned Ethiopia has banned all imports of ICE, already in effect.

Just imagine what will happen as more remote areas adopt renewables only. They simply cannot afford the extortionately priced fossil fuels.

Ethiopia does not need ICE imports as it has infinite refurbishment of its huge LADA fleet. Not so great for emissions tho'

 

[M|S|M MYP]

Trick Question:
What color is this light?

View attachment 1016486

Just checking...

Your arrow looks more like its skewed more towards to the bottom color.

 

[SOULPEDL]

You are quite correct in my view. Much of the LLM based AI is pure hype. The apps are superficial and easy to copy so no moat. Microsoft will dominate in volume simply due to the installed base of ms office and there is a serious threat to google search. There will be increasing barriers as people protect their data and try for royalties on the training data respected groups like Gartner think this first wave will all be done in two years or less. It’s the vertically integrated game changing AI for things such as FSD and other process driven apps that will truly impact us all. I am happy for the chip guys on their current roll. But make hay now. It’s a bubble and won’t last long. Tesla’s lead is not recognised in all the noise but I am very convinced that it will get there and unlike others they have the scale to make a difference. Tesla has the data. It has the technology - or nearly there - and it has the defined market and app need. Patience is very hard at times but persistence will pay off.

Agree, but the little apps are chewing away at jobs etc. Everyone making and sharing their own GPTs since last month, it's like macro's on steroids. In theory, this should continue much like what Apps did for Apple (opinion).

My son's got a secret GPT he created already, and it could put into question the feasibility of some existing revenue streams out there. Everyone in the space of creating their own GPTs is likely discovering similar, low-hanging fruit. GPTs gonna be as common as Apps. These GPTs could become the building blocks for more advanced capability, just by connecting the dots.

 

[SOULPEDL]

Your arrow looks more like its skewed more towards to the bottom color.

I didn't place that arrow, it was on the web that way, saved me a step! Look at me, I'm efficient!

(Edit: Oops, my Histrionics Disorder was showing...)

 

[Discoducky]

Much of the LLM based AI is pure hype.

Which parts are "pure hype" exactly? Provide examples and I'm happy to discuss here or in another thread.

 

[2daMoon]

But something that I see becoming even better engineering wise in Neuralink in maybe a few decades, many times you came up with the perfect design in your head, it’s there, ready to be made, but you need to spend two weeks in CAD translating if from .brain to .sldprt

Join that with AI and AR and you have a winner, imagine working in design in 3D as if you had your parts in front of you, being able to manipulate, immediately asking a dedicated AI to setup and run a lifetime fatigue analysis in X scenario and so on?

As I understand it, this is exactly how the Tesla namesake's brain worked.

Nikola Tesla would work out an idea down to the smallest detail in his head, then go and build it without much in the way of notes or drawings.

 

[Krugerrand]

Or... we made it!

View attachment 1016541

But, short lived.

 

[Tony73]

It's frustrating that the market doesn't understand that high-quality data is the gold, not chips

Chipmakers are the picks and axes sellers of the the AI gold rush. The most successful ones are a lock to do very well.

 

[jkirkwood001]

You are quite correct in my view. Much of the LLM based AI is pure hype. The apps are superficial and easy to copy so no moat. Microsoft will dominate in volume simply due to the installed base of ms office and there is a serious threat to google search. There will be increasing barriers as people protect their data and try for royalties on the training data respected groups like Gartner think this first wave will all be done in two years or less. It’s the vertically integrated game changing AI for things such as FSD and other process driven apps that will truly impact us all. I am happy for the chip guys on their current roll. But make hay now. It’s a bubble and won’t last long. Tesla’s lead is not recognised in all the noise but I am very convinced that it will get there and unlike others they have the scale to make a difference. Tesla has the data. It has the technology - or nearly there - and it has the defined market and app need. Patience is very hard at times but persistence will pay off.

The latest Windows update has Copilot baked into the OS - you don't even need Office: