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

[aubreymcfato]

Nvidia makes most of their money selling hardware / chips. They seem to not only hold a significant lead in terms of the processing power for their chips but also when it comes to their efficiency. They already have competition in the form of AMD and Google, but there's a lot of inertia in the field, as Nvidia solutions are favored by the existing software solutions. In any case, the chances of tesla competing with Nvidia on chips are minuscule. Less money, less expertise and virtually no vertical integration or ability to push their solution like Google has.

What tesla could do is sell compute as a service. Essentially competing with Azure, AWS and GCP. But Tesla has nothing of the infrastructure or customers for this. There's no reason to believe they'll outcompete companies with loads of expertise and virtually unlimited financial resources in a game they've been playing for a long time.

This isn't like BEVs where you had actual technical developments that made electric vehicles feasible and a clear departure from existing architectures.

You're right... Unless there is a way to use HW4 e HW5 for training or SaaS. In that case, Tesla has something of a moat: hardware in ~6 million cars, whole software stack developed, years of experience. I don't understand why HW4 and 5 will be in the Texas data center: aren't those for inference? Is Elon not telling us something?

 

[Usain]

Can someone explain to me why Elon talks compute in terms of Energy?

He has also talked about it in terms of H100 equivalents. I forget what he has been saying when he refers to inference, but it's yet another measure.

 

[ToTheStars]

Because it is one number that describes it all? It is, also, your limiting factor. You have x available power for running chips and cooling, regardless of what kind of chips they are or how efficient they are. They used "H100 equivalents" to describe compute once, and people misinterpreted that, and complained about it as well.

Otherwise, you end up saying something like 100k TOPS of compute plus 150k BTUs of cooling. Which really isn't any more meaning full. Especially since they are installing NVidia H100s, Tesla HW4 chips, Dojo 1.x chips, Tesla AI5 chips, probably AMD xxxx chips, etc.

And you don't only need power for the chips and cooling, you need it for the networking, storage, and all the other necessary ancillary services/devices.

Great answer, I totally agree.
I would like to add something.
Elon likes to think from first principles, as we know. Energy and time seem like the most fundamental quantities you can think of to express computation in.
Of course we have to take into account efficiency as well (Moore's Law), but if Tesla is using state of the art chips to do the computing with (I think they do), power is the best quantity to compare.

 

[Captkerosene]

Nvidia makes most of their money selling hardware / chips. They seem to not only hold a significant lead in terms of the processing power for their chips but also when it comes to their efficiency. They already have competition in the form of AMD and Google, but there's a lot of inertia in the field, as Nvidia solutions are favored by the existing software solutions. In any case, the chances of tesla competing with Nvidia on chips are minuscule. Less money, less expertise and virtually no vertical integration or ability to push their solution like Google has.

What tesla could do is sell compute as a service. Essentially competing with Azure, AWS and GCP. But Tesla has nothing of the infrastructure or customers for this. There's no reason to believe they'll outcompete companies with loads of expertise and virtually unlimited financial resources in a game they've been playing for a long time.

This isn't like BEVs where you had actual technical developments that made electric vehicles feasible and a clear departure from existing architectures.

Thanks for responding.

I invested in Tesla five years ago after asking questions like: What are Tesla's competitive advantages over legacy auto? Will they go bankrupt? Is Elon going to implode? Those were the right questions. The correct answers to those questions earned me a ten bagger. The questions I'm thinking about now are different.

I don't think Nvidia got to a $3.3T valuation by just making chips. Isn't the value of Nvidia in what they can do with their chips? Isn't that where we'll be bumping heads with them? If, if ... we get other automakers to sign up for FSD won't we be the leader in the licensing of full-stack real-world AI? To be followed by the licensing of multiple different bot embodiments and applications? Isn't that a bigger TAM than just making chips? Isn't that where the "durable competitive advantage" lies in the long term? I don't want to build 10% of the bots out there - I want to own the software and hardware they run on. At least those are the questions I'm thinking about in terms of where we'll be in the next five years. Still not sure I'm even asking the right questions. Long way to go before I have any answers.

 

[FastEddieB]

Nice setup for tomorrow's triple witching 180 max pain price. Nothing to see here....move along...

For us old-timey, efficient market types, could you maybe explain what you're talking about here? And do you have any sort of prediction for TSLA stock price move tomorrow as a result?

 

[GatorMeat]

This year I replaced my roof for a new solar system and was surprised all of my quotes had a 50-year warranty on the shingles. It has been a long time since I needed a new roof so I wasn't aware these new composite shingles lasted that long.

no shingle lasts 50 years. the warranty is useless if you read it.

 

[jerry33]

For us old-timey, efficient market types, could you maybe explain what you're talking about here? And do you have any sort of prediction for TSLA stock price move tomorrow as a result?

Tomorrow is Friday. The MM and Hedge Funds will try to close at Max Pain because it costs them money if it's higher or lower.

 

[Southpasfan]

Thanks for responding.

I invested in Tesla five years ago after asking questions like: What are Tesla's competitive advantages over legacy auto? Will they go bankrupt? Is Elon going to implode? Those were the right questions. The correct answers to those questions earned me a ten bagger. The questions I'm thinking about now are different.

I don't think Nvidia got to a $3.3T valuation by just making chips. Isn't the value of Nvidia in what they can do with their chips? Isn't that where we'll be bumping heads with them? If, if ... we get other automakers to sign up for FSD won't we be the leader in the licensing of full-stack real-world AI? To be followed by the licensing of multiple different bot embodiments and applications? Isn't that a bigger TAM than just making chips? Isn't that where the "durable competitive advantage" lies in the long term? I don't want to build 10% of the bots out there - I want to own the software and hardware they run on. At least those are the questions I'm thinking about in terms of where we'll be in the next five years. Still not sure I'm even asking the right questions. Long way to go before I have any answers.

I agree with this completely. Which is why I am not sure I understand why one would want to build physical taxis when you can just own the software which everyone is forced to buy.

Its like Microsoft deciding the real dough is in making computers. Anyway, holding for the duration.

 

[MartinAustin]

Unfortunately it is affecting Tesla for registration of new vehicles in MI, WI and some other states.

https://twitter.com/x/status/1803881248781181227

Obviously this will not be acceptable for you-know-who and Tesla will try to use their own stack.

(USA specific) Often I see private companies getting an exclusive statewide contract to provide a service. Examples... buying custom plates, toll fees, car registration, paying traffic violation fines. If this is the same thing and CDK Global has a statewide monopoly on those new-car registration services, maybe it's time to change - dealerships and Tesla alike would want to have choices.

 

[GatorMeat]

Thanks for sharing. I'd say anyone who could even buy a Model X or multiple Tesla's would be in the upper income/asset class in America. I think that's a major disconnect here. Poor folks don't even own equities, let alone a ton of Tesla stock so when people post/talk here, it seems it's from an upper income bias. Folks who won't pay for FSD are morons according to some people here, not knowing that some folks live month/month. I'd say unless you know what it's like to be poor, most views aren't representative of the overall consumer (and why EV sales have fell off a cliff).

If you're in the Bay Area and have to charge on a 2-3 hour commute (and not allowed free at work), that can't be cheap I'd guess.

probably charge on taxpayer dime, like they sleep on taxpayer dime

 

[geometro]

I like Tom Lee, but can someone translate this?

https://twitter.com/x/status/1803816195000279187

 

[GatorMeat]

I like Tom Lee, but can someone translate this?

https://twitter.com/x/status/1803816195000279187

easy

 

[Waiting4M3]

He calls stock a granny shot because it's an easy bet, example here

 

[kbM3]

You are leaving out a critical detail. Musk had to grow valuation from $100B to $650B or he didn’t a single dime.

As for the “founder” nonsense, only people who know zilch about startups find it interesting.

Adoptive parents would find your quip about adopted children offensive.

You’re understating the accomplishment.

Elon generated a 13X return. Increasing TSLA from $50B to $650B market cap.

 

[GhostSkater]

When people talk crap about 4680 progress, remember it's a really easy problem to solve

https://twitter.com/x/status/1803905132423450868

 

[sunwarriors]

I got some smiley faces a few weeks back when I said that maybe Nvidia should be worried about us instead of us worrying about Nvidia. Given huge margins at Nvidia, isn't it possible that we can undercut them when it comes to training ... especially if we have better software? (I freely admit my ignorance in this area and ask the question again in the hopes it will stir some debate.)

https://twitter.com/x/status/1803849373018644880

Edit:
Then HW5, which has been renamed to AI5, in the second half of next year. The Tesla AI5 computer has ~10X the capability of HW4 computer and Tesla makes the whole software stack. Elon

Pretty much everyone is tired of paying NVDA for their chips:

Is Microsoft teaming up with AMD to develop AI chips? - gHacks Tech News

A report by Bloomberg suggests that Microsoft is working with AMD to develop chips specifically for AI applications and services.

www.ghacks.net

AWS Announces $230 Million Commitment for Generative AI Startups

Funds will help the next generation of startups turbocharge development of new generative AI applications and bring innovations to market faster

press.aboutamazon.com

Introducing Our Next Generation Infrastructure for AI | Meta

We’re introducing the next generation of Meta’s custom-made chips designed for our AI workloads.

about.fb.com

They all have billions of $$ to invest in it as well vs. the miniscule EV market back in 2012. I'm not saying TSLA can't do it, but it's a very VERY crowded space with a ton of very well free cash flow companies who can throw hundreds of billions at it as well. Just need to decide is this the fight they want since TSLA already has their own chips, but needs to buy from NVDA as well?

I didn't read any of those articles, but you can search and simply assume everyone is entering this space.

 

[Singuy]

Pretty much everyone is tired of paying NVDA for their chips:

Is Microsoft teaming up with AMD to develop AI chips? - gHacks Tech News

A report by Bloomberg suggests that Microsoft is working with AMD to develop chips specifically for AI applications and services.

www.ghacks.net

AWS Announces $230 Million Commitment for Generative AI Startups

Funds will help the next generation of startups turbocharge development of new generative AI applications and bring innovations to market faster

press.aboutamazon.com

Introducing Our Next Generation Infrastructure for AI | Meta

We’re introducing the next generation of Meta’s custom-made chips designed for our AI workloads.

about.fb.com

They all have billions of $$ to invest in it as well vs. the miniscule EV market back in 2012. I'm not saying TSLA can't do it, but it's a very VERY crowded space with a ton of very well free cash flow companies who can throw hundreds of billions at it as well. Just need to decide is this the fight they want since TSLA already has their own chips, but needs to buy from NVDA as well?

I didn't read any of those articles, but you can search and simply assume everyone is entering this space.

There are 2 aspects to this race.

1. Time it takes to develope your own AI model like Chatgtp, FSD, robotics, etc etc so you can be first to market

2. Hardware acquisition and cost used for training and inference

Right now you buy Nvda so you can have a chance at winning #1 because it's drop and play. Most AI research and training technics are done on Nvda hardware using their software ecosystem. It takes 9 months to get a server room running and another few months to optimize the hardware. It takes 5 years to go from hardware development ->chip and another year to deploy.

So yes everyone is buying nvda today or else you'll twiddle your thumb for half a decade waiting for your inhouse hardware while the rest of the world has their AI model out already.

Hardware acquisition we have seen is a b%$tch. Everyone rushes to buy the newest because they have at least 5x the TCO so you end up waiting and paying a premium, but you must because of everything I said above. Think of right now as a coke bottle with a thin neck, where Nvda runs the show. However everyone really wants an alternative because the race is a lame one to be in.

 

[AnbaricBike]

Its in the context of the fans. The fans are there for waste heat, and thats going to be strongly related to the energy used to power the data center.

Seems like they have enough thermal processes at that plant to make use of that heat. More heat is the last thing Austin needs.

 

[Blue horseshoe]

They made quick work of the wiper motor replacements and about 30 owners took delivery today in Marina del Rey. They did a whole synchronized show for us which was fun! Got a rivet accelerator to boot!

 

[FreqFlyer]

View attachment 1058317
They made quick work of the wiper motor replacements and about 30 owners took delivery today in Marina del Rey. They did a whole synchronized show for us which was fun! Got a rivet accelerator to boot!

Saw two of them on a carrier full of Model Ys northbound on Tollway 130 leaving the factory.