Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

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

This site may earn commission on affiliate links.
Elon goes to China for FSD approvals -- oh! he's back and enticing us to vote for his comp plan

Elon fires entire teams in middle of the night -- that's just Elon being Elon
BOTH are Elon.

It’s just the most amazing coincidence that so many participants disappeared for green days since ER and then just as suddenly reappear for a red day.
 
What does Grok know? He's he up to date to the latest news? I would never trust it for these questions.
according to Grok

Grok obtains up-to-date information by having access to real-time data from X (formerly known as Twitter), which is a unique and differentiating feature compared to other AI models like ChatGPT. This access allows Grok to provide up-to-the-minute insights, including breaking news and live tweets, which can be a significant advantage for users seeking the latest information on various topics. This real-time data access is available to Grok users as part of their premium subscription to X, ensuring they have the most current and relevant information at their fingertips.
 
Management rolling tells me they were not selective enough in identifying unnecessary resource as directed. It's gotta be tough to fire people, fortunately I've never had to do that.

I don't view those Assembly skills as difficult, more routine by now. If they had well defined workmethods, perhaps even recorded by an emersive bodysuit, it should not be difficult to re-establish what's needed, and re-hire pure essential employees (vs let go who they don't need). Maybe that's actually easier than watching your collegues and friends get walked one at a time - which is something I did experience at one time.

We are not getting the whole story, that's the whole point. It is intended to create questions and worry. I'm cool waiting for the rest of the story and view this as a despirate move to scare investors who should (and normally do) reward layoffs.
It’s hard firing ‘good’ employees. It’s remarkably easy firing not so good employees. I’ve done both.

1,000,000% we don’t have the whole story. And yet -
 

Absolute madness, this is one area where Tesla had the foresight to invest in and as a result it is now the leader.
Arguably EVs would not have been as viable without the network tesla built.
Logically you continue to build on this strength. Even if team needs reducing, you keep some people to continue the pipeline because as the cumulative number of EVs grows more superchargers will be needed.
I await the leaders explanation.
 
Logically you continue to build on this strength. Even if team needs reducing, you keep some people to continue the pipeline because as the cumulative number of EVs grows more superchargers will be needed.
That’s what YOU think should be done. It’s clearly not how Elon sees it and he has way more information than you.

I await the leaders explanation.
He already told you why. Did you not read the various email excerpts posted here?
 
It's not that this is actually impossible. Self-modifying code was a common technique in the early days of computing. It's now mostly a lost art. As for the checksum, it's possible to generate a new one and store it so that the system would still work on reboot. I can think of other ways to get around it as well.

But no sane developer would attempt such a thing for an application like FSD. Like @Knightshade correctly points out, it would be a nightmare to debug. And furthermore, it would invalidate all the data gathered from the fleet because no two cars would be running the same model.
Back in the day I happened to write a little bit of self-modifying code. In good old GW Basic, an interpreter, which allowed for such things.

But we're not necessarily talking about that.

As a made-up thought experiment, suppose that one has a subroutine that has a bunch of parameters, passed to it by some calling program. Some of those parameters are pointers to data locations that will be returned to the calling program; some of those parameters are values or arrays or something to be operated upon by the subroutine.

The point is that a value passed to said subroutine might be some kind of analog slider. For the thought experiment, this value might be, "how aggressive should this car be on an unprotected left turn?" Said it to a low value, and the car's a pansy and waits for hours for traffic to be clear in both directions for a mile back. Set it to a high value and the Car Risks Its Life And The Lives Of Innocent Bystanders. The subroutine might check this value and make sure it's within a safe range. But, having said that, a varying value, set at some nominal value at compile time, but allowed to float based upon "X", where "X" might be based upon map data.. or the car's history in some way.

Calling this kind of process, "Self modifying code" isn't quite right. But it could alter the behavior of the car.

FWIW, I've spent more hours digging through highly stable network clock nodes, both in hardware and software, than I can easily remember. One of the characteristics of such nodes is a concept called, "Holdover", which is used to set the frequency of a node when all input references are lost. When one first powers up a node, the holdover value is set to something nominal, but is then (during initialization) cranked to accommodate offsets in hardware reference clocks (which might be stable, but not accurate), gains in VCOX types devices (which are subject to manufacturing variations), and such like. Telecom hardware is full of adjustments like this; some SDH gear with which I'm familiar might have a dozen or so control loops that minimize error rates and such, all of which get adjusted initially, and might be kicked into "fast adjustment" modes under the right circumstances. (And how that interacts with slewing system clocks is what keeps mixed signal engineers up at night.)

None of this is exactly new technology; we're talking multiple decades, here. And we're also talking checksummed code. But the effect of all this is different behavior over time. There's no particular reason to think that the FSD code base doesn't have loops like this. Or that some of the NN weights might be designated for these kinds of control loops.

Now, as to whether Tesla actually has stuff like this baked into the FSD software: Who knows? But reports from the field (that is, users) hint that that might be the case.
 
That’s the way you want to play this? Are you sure? Even after one of your buddies tried to dig his way out by pointing out TSLA is up significantly since ER call and that today most everything is red.
Looking at the present stock price some are taking your point C. Congrats, you are a Tesla stock analyst savant.

A stockholder, asking why at the Church of Elon does not make one a heretic.


To others-
Regarding the layoff news, currently, it is without context. A lot of institutional knowledge just walked out the door and yes that matters. I am also long in Boeing, and it is finding out quickly that experienced workers matter. Some are rumoring that a Super Charger sale may be in the works. Maybe, who knows? Or The Super Charger network and expansion is plug and play and could be run more efficiently with a smaller team.

On the flip side, the headcount at Tesla is a little high, so some pruning is in order.
 
My only thought on letting all the supercharging engineers go is that Elon is focusing on the Wiferon induction charging tech acquired in Sept 2023. In hindsight, this could seem rather obvious if FSD is solved soon'ish (6 to 12 months).

Also, I'm super bummed about Daniel Ho as we worked together and it was a joy to see him in the Tesla video for Highland. I'm sure he'll land a great role somewhere!

 
I've been extremely interested in Supercharger developments over past several months. I have been scrubbing some data and built my own Supercharger Report which I've refreshed daily.

As of today, we have 6,304 locations with 58,000 stalls around the world.

Asia Pacific: 2,620 locations, 15,440 stalls.
Europe: 1,186 locations, 14,729 stalls.
North America: 2,498 locations, 27,831 stalls.

Based on this, I think that with the latest developments, SC development/deployment paths need to be modified.
Perhaps Elon figured out that there is no need for 500 folks doing the work. Me thinks that the initial SC stage is long past. There are things coming that only Elon and his crew know.

In any case, see the maps. The next phase need to be different now and Elon is doing it again.

EU.PNG
Australia.PNG
CA.PNG
 
New theory, what if Buffalo Plant has been whispering UAW with Mr. Fain? We would not know if this is the reason as that would be illegal. I'm also seeing Buffalo on their radar as having 1,808 employees - so <1/3 (or it's old data, as IDK).

Recall a year ago, something similar happened with this factory. Also consider China ramp, margins, maybe Buffalo is no longer needed ATM. So, move their equipment to Nevada or Texas, and bring in the Robots? Maybe.

FYI, this story is ~ 1 Year Old.


And this one's interesting, only 2 months old (but I don't have access to read the story).

1714489929086.png
 
I've been extremely interested in Supercharger developments over past several months. I have been scrubbing some data and built my own Supercharger Report which I've refreshed daily.

As of today, we have 6,304 locations with 58,000 stalls around the world.

Asia Pacific: 2,620 locations, 15,440 stalls.
Europe: 1,186 locations, 14,729 stalls.
North America: 2,498 locations, 27,831 stalls.

Based on this, I think that with the latest developments, SC development/deployment paths need to be modified.
Perhaps Elon figured out that there is no need for 500 folks doing the work. Me thinks that the initial SC stage is long past. There are things coming that only Elon and his crew know.

In any case, see the maps. The next phase need to be different now and Elon is doing it again.

View attachment 1043137View attachment 1043138View attachment 1043139
Would the Drive-in restaurant SuperCharger under construction in LA be under this team? Who oversees the SC construction at the Buffalo plant?
 
  • Helpful
Reactions: jkirkwood001

Absolute madness, this is one area where Tesla had the foresight to invest in and as a result it is now the leader.
Arguably EVs would not have been as viable without the network tesla built.
Logically you continue to build on this strength. Even if team needs reducing, you keep some people to continue the pipeline because as the cumulative number of EVs grows more superchargers will be needed.
I await the leaders explanation.

Remember, Elon is playing 4D chess here. If everyone can use NACS, you have no advantage with your brand of EVs, nor with your EV charging so if it's bad, it's bad for every EV car owner and really, who cares? It's all the same stuff. Who cares how great charging will be when you can be driven around with the fully autonomous RT so you won't even need a car now. /s

It's all about AI/RT/FSD now.
 
New theory, what if Buffalo Plant has been whispering UAW with Mr. Fain? We would not know if this is the reason as that would be illegal. I'm also seeing Buffalo on their radar as having 1,808 employees - so <1/3 (or it's old data, as IDK).

Recall a year ago, something similar happened with this factory. Also consider China ramp, margins, maybe Buffalo is no longer needed ATM. So, move their equipment to Nevada or Texas, and bring in the Robots? Maybe.

FYI, this story is ~ 1 Year Old.


And this one's interesting, only 2 months old (but I don't have access to read the story).

View attachment 1043141
I believe Tesla has received tax breaks tied to employment numbers. To move out of Buffalo may require a check from Tesla. Not to be heartless, but long-term may be advantageous.

I would be more concerned about UAW's efforts at Fremont. Not saying the UAW has a snowball chance in Death Valley in July but the recent layoffs and staged VW win might give a tiny bit of momentum.
 
There's nothing I've seen in videos or data that makes me think they're on the cusp of getting where Waymo was 4 years ago in terms of capabilities. That is running a level 4 system. Not to mention level 5. So although V12 is a huge step forward, there's still a long way to go. And there are no guarantees on whether it's even possible for them to reach level 4 with the current approach. And I'm surprised people are pushing back so hard on the latter part. How can they be guaranteed to reach a milestone with a solution that NOBODY else has tried before?



From the data published by another poster, I can see that the MSRP to invoice difference is anywhere between 3 and 8%. So on average you can assume a conventional manufacturer loses 5.5% top line compared to Tesla's direct model and a fraction of that would've been gross margin. But I do get your point (and agree) that it's a complicated calculation to make. If you have any dealership network that is publicly listed, I'd love to delve a bit into their reports and see what I can find.




The bolded part is a strong statement to make. Conventional L2 systems are pretty good at reacting to frontal collisions, detecting cross traffic. What they can't do is take evasive maneuvers, but statistically you end up with just a few cases over an impressive number of miles driven. Don't want to discount those situations, every life is important, but people seem to discard those situations as "it will never happen to me".



Because I think the only timeline that matters from a financial perspective is the one for widespread L4 deployment. And although V12 is certainly a step forward, the timeline for widespread L4 deployment is certainly greater than 3 years out.
Here is the Europe+UK list as of 2023:
here is the US list:

All of those huge dealer groups have fairly complex accounting issues. Ones such as Penske, on both lists, have major consolidated leasing operations, in Penske's case one fo the largest commercial vehicle lessors in the world.

Where the problems of understanding dealer margins become clear is in the major collision repair, short term leasing/renting, warranty repair, service etc all are artfully managed to disclose what they want to disclose without revealing too much of OEM relationships. All of those dealership groups, on both sides of the Atlantic have huge volumes of used vehicles, from off lease, trade in, manufacturer supply all of which are not included in any typical new car MSRP/'dealer cost' calculus. In real world terms the large dealers have numerous extra benefits from sales quota 'spiffs' (extra common), floor planning incentives, equipment and facility finance, including subsidies.

All that accumulates to increase the cost of a dealer distribution system. The net of all that makes new car MSRP/'dealer cost' an entirely managed, financially insignificant, component of large dealer profitability.

Because Tesla has no dealers their actual distribution costs are probably about half that of typical OEM.
OTOH Tesla accounting recognizes a sale only on final delivery with payment in full. Other OEM's tend to recognize a sale when accounting transfer to a dealer/distributor happens (most often when shipped from a factory). That also means Tesla inventory levels include both OEM and dealer inventory so there si an apples to oranges comparison. Since in some cases OEM's also report dealer inventory that tends to distort actual inventory cost dynamics.

All this becomes very complex and generalizations are fraught because each is different. VAG, for example, owns some dealers that also serve as handy ways to solve sales issues. Not too long ago they allegedly sold large number of otherwise unsalable ID3 to their own dealers, so reported sales, even though it was actually to themselves. That sort of 'stuffing' long ago had Buick Reatta, which was virtually unsalable, 'sold' to Alamo Rent-a-Car, with "favorable terms". It was easier when a former GM executive ran the company, and GM mostly financed the company at the time.

These and countless other examples demonstrate the strange and often symbiotic relationship between OEM's and Dealers. After all the gimmicky accounting and incentives ar counted it's fair to say that dealer distraction costs most OEM's somewhere around 20-30% of gross sales with variation between countries, brands and models.

Some, like Geely have approached this by having dealers, where permitted, on fixed sales commissions of 6-8% plus pre-designated warranty fees, and then retain title themselves until customer delivery.while allowing used cars and out-of-warranty to be anything they wish. That has, in such countries, tended to help sell BEV's, especially. The side benefit is that they dictate service intervals to allow dealers more income, even when BEV services are not really necessary. Customers don't know that.

And on it goes, with more and more layers. Luckily for all of us with Tesla WYSIWYG. The sometimes irritating corollary: Tesla does not hide incentives and price changes. Every customer sees it. That allows some customers to be offended. Those might be happier if they are ripped off in secret.

Lastly: Leasing. The secret sauce of high profits and hidden costs. That is another story, Every significant dealer plays with this. Simultaneously reducing reported profits while avoiding taxes. In nearly every major country these are quite different, but have huge significance. Corporate tax rules in many countries make dealers wealthy while often masking their actual direct roles. In many large dealer groups leasing, whether openly disclosed or not, account for more than half of net profits. Tax rules nearly always allow taxable income to be far lower than it would otherwise be.

I will not further comment on Finance and Insurance (F&I) except to say a common dealer nickname in some quarters "Trash and Trinkets" T&T
 
Last edited: