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

[UnknownSoldier]

Even though financials and guidance are amazing, I'm thinking we won't see a nice bump in stock price. Would be happy to be wrong but nothing we do is ever really good enough for Wall Street.

 

[adiggs]

I share your skepticism of believing Elon when he says "FSD this year, for sure, honest, no really I mean it this time." I'll believe it when I see it, but I know it's close. It seems close for 2022 because the beta is in 60,000 cars in January, with zero crashes so far, whereas there have been an average of 106 fatalities every day this year, via human-driven cars.

Hummer EV is delivering to customers as of Dec 2021 (TINY volume initially of course)







I'm guessing you haven't driven the beta? I have, for a while now.

With a human ready to intervene it's pretty cool. It works well a lot of the time. Sometimes I can take a half hour or longer trip without having to intervene at all, especially if it's very little traffic around.

But it's nowhere near ready for use without a human actively ready to intervene.

I've had it lane change into oncoming traffic. Twice. And not in the old 2020 version, but recently. . More than a few times it simply refuses to proceed at an intersection without you confirming it's ok....or takes several minutes to proceed creeping the whole time, sometimes into the oncoming traffic lanes for a bit before going.


That said- Elon claimed on the Lex podcast recently they're doing Yet Another Fundamental Rewrite That Will Change Everything And Make RTs Real in v11, and he mentioned it again on the call today.

We've heard that several times- eventually I guess it'll be true, but it hasn't been for any of the previous rewrites so far.

This is where I'm at. Especially the rewrite thing.

I've worked in a broad sense in machine learning, just not with video. I am familiar with the generalized workflow, and more specifically with the ease of finding a / any model in a set of data (there are always patterns when you go looking for patterns), while in some circumstances being unable to find a good model at all. Or a good model that still isn't good enough to make use of. Finding good models that can be used to solve the business problem can be amazingly difficult.

Even more fun - with open ended completion timelines! Sort of like research - you'll know you're "done" when it actually works, without knowing how long until it works or even IF it will work. Some of these more cutting edge projects don't fit very well into a project management approach. Actually they work well with an agile mindset but when the people running the project need a completion date, there isn't one. These projects can be a few days away from working or a few years, and might even prove to be impossible.

I've been through a few of these fundamental rewrites where the structure of the problem, and the details of how its being solved are needed. Somebody that says, today, that this really is the very last fundamental rewrite of this FSD software is really just guessing.


Net net for me - I like tangible reports on distribution progress. I.e. that there are now 60k people/customers running the software - that is useful. Even better if its reported as "people running the software that don't work directly or indirectly for Tesla". If there are 50k Tesla employees in that 60k, then its misleading (I don't expect that to be the case).

Accident rate on City NOA - that'd be wonderful. These are both measures of actual software in use by people that don't work for Tesla. And City NOA vs. what we have today, as accident rate on freeway / highway driving is much lower than in town.

Ready for general distribution to the customer base in x months - guesswork and not helpful. We've been hearing this for 2+ years now it seems, and every time its just around the corner we get lots of people repeating that they'll be surprised if it doesn't happen in 2020 2021 2022.


For an investor my not-advice is that if broad FSD deployment is critical to your investment thesis, then reconsider whether Tesla is a good investment for you. I agree that there is a ton of upside because of FSD / driver assist. That makes it an amazing option on successful development.. I personally put Tesla comfortably #1 in this race, with maybe Mobileye #2, and most of the others trying to figure out what a starting line looks like. The others still have the work to do to realize that fundamental assumptions are wrong, and its time for fundamental rewrite #1 #2 #3 #4.

That doesn't mean there is a timeline to general deployment today any more than there was a timeline to general deployment last year or the year before.

 

[Webeevdrivers]

That'd be helpful, but I think you're wrong on that. When Tesla moved their headquarters to Austin, they didn't close their office space in Palo Alto. In fact, they leased another 350K sq.ft from H.P.

Tesla News and Reviews | InsideEVs

Get the latest on new Tesla models including spy shots, first drives, in-depth reviews, concept car photos, new features and more.

insideevs.com

I suspect due to the sheer volume of data, the labelers need to be co-located with the training CPUs (bandwidth constraints).

This is just my WAG, but that's an awful lot of new office space for which we know of no other purpose. Does anyone know where the server farms are located for FSD training? Follow the Fiber...

Cheers!

What does a labeler do. Thank you for your patience.

 

[Artful Dodger]

I dont even see where he got these numbers but I am pretty dense. Any accountants around to decipher this?

Lol, that's just an attack on the "61 days of accounts payable" and "10 days of accounts receivable". All it means is that Tesla is receiving parts, building the car, and selling it before they have to pay for the parts.

Elon told us this literally 4 years ago, that's the main reason to build at least one Gigafactory on each continent where they sell cars. The idea is, building more cars results in INCREASED working capital rather than consuming working capital.

TL;dr GoJo is a clown (3rd class).

 

[Zhelko Dimic]

Yeah, he send that tweet way back on Dec 3, 2021. A lot didn't happen since then... he could have skipped.

Elon should hire a script writer, these mad-libs aren't informative. Nothing new today from the "roadmap".

It's good we learned to not expect much...
When he announced he'll give roadmap update, I felt (and commented here) that was improvisation and an excuse to join back, as he missed being in a spotlight.

 

[Paul_SF]

Dave is on

 

[Discoducky]

For an investor my not-advice is that if broad FSD deployment is critical to your investment thesis, then reconsider whether Tesla is a good investment for you.

I normally love what you write, but with this sentence, I think it might be better to say that "If you don't believe Elon will EVER deliver full FSD..."

I think we all participate in this shared amazing experience that is Tesla/Elon as we do believe he will deliver...someday.

I told my AP team once, when we were working with SpaceX on their Dragon capsule UI, that developing a rocket that can land itself is super easy RELATIVE to FSD. And that time I couldn't dream of making a rocket land itself, but I was TOTALLY bought in on creating FSD. Elon is bought in and will make it happen, it just isn't a matter of will or effort or patience, it is just a matter of time.

 

[JRP3]

Were people expecting a new reveal and summer launch?

After the newly leaked prototype pics and video I was hoping for something more positive. At this point it sounds as if they still don't know how to make the Cybertruck.

 

[RobDickinson]

After the newly leaked prototype pics and video I was hoping for something more positive. At this point it sounds as if they still don't know how to make the Cybertruck.

When you are going to make millions of a large expensive complicated thing and you want to make it well and as cheap as possible it takes time, validating all that tames time. Packaging all that up in teslas usual slick way takes time and effort.

 

[Zhelko Dimic]

id prefer the main narrative be unadulterated explosive growth, built upon their fundamental strengths (engineering, ops, logistics)

and, bonus, 4680 in the products right now, since that is the future foundation of products.

- let fsd/robotaxi and optimus be the whip cream on top, a year or whatever time is needed down the road

Yeah, these answers could age badly...
If analysts don't give him credit for FSD, and discount revenue from Model 2, what happens?
He's betting that he can sell 3M+ of a single car platform.
If anyone can do it it's Elon, and I know that if things don't go well, Tesla will adjust in a nick of time...
But it's asking much of analysts' imaginations to endorse this plan...

 

[Creekstalker]

Even though financials and guidance are amazing, I'm thinking we won't see a nice bump in stock price. Would be happy to be wrong but nothing we do is ever really good enough for Wall Street.

When in doubt zoom out, Eeyore! Wall Street may not respond correctly tomorrow, but even those dullards eventually catch on. If you've been holding for more than a couple years, you are doing better than 99% of all investors. This is a great day!

 

[jerry33]

What does a labeler do. Thank you for your patience.

Think of the pictures that you "label" on Captcha. This is a stop light, this is a truck, etc. (this is not a stop light, this is not a truck). That's basically what a labeler does.

 

[Knightshade]

When you are going to make millions of a large expensive complicated thing and you want to make it well and as cheap as possible it takes time, validating all that tames time. Packaging all that up in teslas usual slick way takes time and effort.

For sure. I'd far rather they get it optimized before trying to make a lot of em.

But between this, delays to semi and roadster, and how badly late the S/X refreshes were I hope we can put an end to all the "ELON IS TOTES SANDBAGGING EVERYTHING" idea.

Elon says prototype to production is really hard. Time to believe him.

So instead they're smartly doubling down (or at least substantially more than 50%ing down ) on what they've already got the making of figured out- albeit with significant improvements like the structural pack on the Y, and letting entirely new stuff work out all the kinks in design/engineering/cost reduction throughout the year, rather than rushing to production.

 

[ggies07]

What some are piecing together here is, Tesla will introduce FSD and robotaxi software, and Model 3 & Y owners will be able to offer their cars for rides for as little as $5 a ride - or less - depending on the distance and so on. A dollar a mile? There are a lot of people who can't afford a $25,000 new car and would have trouble paying for a $16,000 new car. But they are always open to buying an $8,000 used car or a $5,000 used car. Such cars aren't as safe to drive and usually come with a list of repairs or things that need to be done on them. If you can ride in a new-ish $40,000+ smooth+quiet electric car that drives itself, won't get into a crash, is very safe for you if someone else crashes into you, and you don't need to insure it or pay for parking it anywhere, and it only costs you $5, you are going to do it and work out how to avoid buying your own car. This is why the market for the $25,000 car isn't as strong as we think it might be. Anywhere globally. Because Tesla are working on FSD first. We should be thinking of the $25,000 car as the "$25,000 individually-owned and driven" car. In that light, it doesn't seem as important.

Yeah, except for all those lower middle class families with 2+ kids that do sports every weekend and go out. No way they will use this every.single.time in 2-5 years. This will take at least 10 years for society to get used to it ONCE it is normalized. So until then, we have all these families still driving their gas powered cars, polluting the environment, because A) Tesla won't make a cheaper car B) we sure as hell know none of the other automakers will be making as many cars as Tesla in the next 10 years. Hell, even GM's trajectory is for 1 million by 2025.

It's the wrong strategy for their mission (because the earth is warming too quickly), but it's the right strategy for the next transportation system.

 

[Zhelko Dimic]

You're still here?

I certainly pay lot more attention to his posts than many others'

 

[ByeByeJohnny]

Some great numbers reported today. But today totally did not live up to the hype. Pretty much back where we started this morning.

I just wished Elon would't hijack these calls to talk about FSD every time. And now Optimus to. At this point almost nobody cares what he says about the timeline. He's a pretty smart guy. How can he not see that.

Anyone not a Tesla Bull don't believe any of those will happen, in a financial meaningful way at least, inside five years. Actually half of the Tesla Bulls don't believe so either.

I'm more and more convinced Elon's a masochist and thrives on being ridiculed in media.

 

[RobDickinson]

I just wished Elon would't hijack these calls to talk about FSD every time. And now Optimus to. At this point almost nobody cares what he says about the timeline. He's a pretty smart guy. How can he not see that.

Honestly Agree with this, FSD or Optimus are not relevant to the Q on Q business update until they are finished shipping products.

 

[strago13]

I agree, but it really only has to beat the timeline for demand not keeping up with production of the current product lineup. At least, that's how I look at it. And if it isn't solved, then Tesla would pull other demand levers. I hope that doesn't happen though.

Do they even need additional factories if they solve FSD? I took away from the call no. Hence all the vague answers about new locations. I think you’re right Tesla is hedging their bets on solving FSD this year negating the need for 25k car, additional factories, etc

 

[BornToFly]

Actual headline - "Rivian stock jumps after report saying factory is ramping production." They shut down to try to ramp to 200/week, which would be a whooping 10,000 vehicles/year if they could ramp instantly. They are up 2%. Tesla ramping to over 1.5M vehicles this year with the best margins in the industry and massive profits - down in AH.... You can't make this stuff up.

 

[Discoducky]

Do they even need additional factories if they solve FSD? I took away from the call no. Hence all the vague answers about new locations. I think you’re right Tesla is hedging their bets on solving FSD this year negating the need for 25k car, additional factories, etc

I think they do, as the mission is to accelerate the advent of sustainable transportation, so I'm thinking they need to hit a run rate of 10 to 15 million/yr to saturate the market in 10 years, assuming robotaxi is launched by late 2024.

There, I've gone on record!