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

[Opus BC fan]

It's been a good day and a good week for TSLA trending.
I am anticipating an upside breakout soon (that's Tesla "soon")
Have a great weekend!

 

[hobbes]

I took delivery of a Model 3 SR MiC in Køge, Denmark a week ago. The lot was PACKED with shiny new cars and many, many customers were there to pick up their new Tesla. Especially Model 3s. The end of quarter push this time is impressive (again).

It's amazing. Totally in love

View attachment 1060531

And we thought "the wave" was a thing of the past...

Electric Vehicle registrations in Europe: 15 countries, majority of BEV market

 

[hobbes]

Kansas Panasonic plant update - production begins 2025 Panasonic applies lessons from Tesla Giga Nevada to new Kansas facility

It even begins in spring 2025!

The infrastructure for wing one is finished, and Panasonic has already started installing cell production equipment inside it. It will continue construction on wing two while installing equipment in wing one.

Panasonic Energy aims to open its Kanas facility by spring 2025.

 

[snellenr]

Refers to inflammatory, deleted post (and factually absolutely wrong, too)

John Carney, who appointed Kathaleen St. J. McCormick to the Delaware Chancery Court, is only the Governor of Delaware. ​

 

[sunwarriors]

Yes, this is a new kind of software development. It is indeed hard to anticipate the speed of progress.

But I don't see why a few weeks delay is a big deal. From my decades of experience in software development I can tell you that releases are almost never on time.

I've said this before, but I stopped caring about all these minor releases. I think most people just don't care. Just wake us up when it's completely unsupervised and there is an actual story to tell. Musk has stated FSD back in 2019, that it should be ready in 2020 to drive you from west to east coast completely. Maybe he can sit in the back of a Tesla and do it or see how far he gets to shut everyone up.

There is a lot of hate with Waymo here, but even when hitting a pole, it's still an actual real self driving service you can order NOW. Open to everyone in SF. Lots of testing in AZ. Another lawsuit again with Tesla employee dying, that FSD about to hit a train, another ~15? lawsuits pending.

The good thing is like the Cybertruck, so much competition in autonomy that Tesla will probably be forced to release something, even if it doesn't work perfectly. Even Toyota is getting in the game in China (yeah, not much confidence there even from me with their record).

 

[EQC_]

...Another lawsuit again with Tesla employee dying...

From a human perspective, it is bad when any person is injured or dies. Full stop there.

In this case, it looks like the part of this lawsuit that is most damaging to Tesla is that it makes the news and many people won't read the details.

Just so we're clear, I believe the key facts of the lawsuit you're mentioning here are:

1. Employee was driving a Tesla.
2. The employee was not on-duty, but on a golfing day with friends.
3. Employee complained about FSD on the drive to the golf course.
4. There seems to have been lots of drinking at the golf course, then the employee and a witness/passenger got back in the car to drive home.
5. Employee had a blood alcohol content that was measured at 3 times the legal limit in the autopsy.
6. The only apparent witness to the crash was a golfing buddy and passenger, who had a "similar" blood alcohol content, and said the employee seemed fine to drive.
   1. The suing family members want to use that statement as evidence that the deceased was not too drunk to drive.
7. That witness reports that FSD or Autopilot was activated, and was doing crazy things, and wouldn't give back control to the driver
   1. As anybody who has used FSD or Autopilot has observed, the car immediately gives back control if you even tap the brake or move the steering wheel more than a few degrees.
8. The vehicle apparently left the road and hit a tree at high enough speed to cause the vehicle to catch fire, resulting in employee's death.

 

[rentierparasit]

It's about quantity and quality. But more data does indeed equal better results.

Look at Karpathy's talk below and pay special attention to the slide at about 25:45 (see below set at proper time).

He presents the LLM scaling law which says performance of an LLM is a smooth, well-behaved, predictable function of N, the number of parameters and D, the amount of training data. He also mentions that there is no known limit to this scaling law.

I see no reason that the LLM scaling law would not apply to Tesla's end-to-end FSD.


Obvious? We the public don't have good data on disengagements per mile. But Elon does. Elon is claiming 5x to 10x intervention improvement between 12.3 and 12.4. I don't think he's lying. He is basing this on internal metrics.


I agree with you here. There is no reason to celebrate any failure. And when (not if) FSD finally goes unsupervised there will be plenty of risk.

But I do think it is fair to say that it looks like Elon has chosen an approach that is superior to Waymo and Tesla has a much better chance at success.

Hi, Usain --

I'm not picking on you in particular, but I found this a bit triggering:

> We the public don't have good data on disengagements per mile.

It's a mystery to me that Elias' tracker isn't more broadly referenced. I guess you could argue that it isn't "good" data, but it's *some* data, and it's possible to discuss how much it might be off by and which direction. I think the key stat for robotaxi purposes is city miles to disengagement. I find Elias' numbers plausible, since they're in rough accordance with TMC trip-report anecdotage. I think E has something like 95% of trips of without a critical DE. That's in line with, every now and then someone pops up and says, "Had a scary moment the other day ...", which, yeah, that sounds about right.

Is there any reason to not use the tracker as a starting point for guessing about the timing of L4 autonomy?

Yours,
RP

 

[Usain]

Hi, Usain --

I'm not picking on you in particular, but I found this a bit triggering:

> We the public don't have good data on disengagements per mile.

It's a mystery to me that Elias' tracker isn't more broadly referenced. I guess you could argue that it isn't "good" data, but it's *some* data, and it's possible to discuss how much it might be off by and which direction. I think the key stat for robotaxi purposes is city miles to disengagement. I find Elias' numbers plausible, since they're in rough accordance with TMC trip-report anecdotage. I think E has something like 95% of trips of without a critical DE. That's in line with, every now and then someone pops up and says, "Had a scary moment the other day ...", which, yeah, that sounds about right.

Is there any reason to not use the tracker as a starting point for guessing about the timing of L4 autonomy?

Yours,
RP

No problem @rentierparasit.

I've talked about this before, but I don't consider Elias' tracker to be a good source of data. There is just too much variance in the participants, routes, subjectivity in reporting, etc.

I think it's great that Elias is doing what he is doing. Maybe it's the best we've got. But I just don't put any stock in it as a measure of progress.

I definitely wouldn't use it to guess the timing of robotaxi-level autonomy or of unsupervised FSD.

I suspect that Tesla has decent data based on performance against simulations. They probably also have good data based on their own professional drivers where they control for things like routes and strict reporting standards. Unfortunately, Tesla doesn't share this data with the public other than vague references from Elon.

So really, we're just left with our own impressions of progress from version to version. Plus, Tesla's recent level of commitment to the robotaxi effort is a good proxy for gauging their confidence in the FSD program.

 

[cusetownusa]

Tesla seems to be close to running out of FS Cybertruck buyers. I Just got my invite to order a FS CT with reservation that was placed October 2022 and I read they are already up to February 2023 reservation holders. I will be passing on my order.

 

[Darkfox021]

Tesla seems to be close to running out of FS Cybertruck buyers. I Just got my invite to order a FS CT with reservation that was placed October 2022 and I read they are already up to February 2023 reservation holders. I will be passing on my order.

6+ months of FS seems to be a good run. Elon already indicated non-FS orders should start next quarter.

 

[Blue horseshoe]

6+ months of FS seems to be a good run. Elon already indicated non-FS orders should start next quarter.

They've sold over 11,000 for an extra $220M in FS upcharges. Not bad

 

[aubreymcfato]

We now have the best month in Norway since March 2023 (which is undefeatable, best month ever and not even close).
We'll see what Germany and France do, but if they are not miserable as we expect we could have a good end of quarter. Italy and Czech republic will increase YoY, but not enough to go on par with 2023.
I'd be super happy to gnaw that gap a bit tho.

 

[CorneliusXX]

From a human perspective, it is bad when any person is injured or dies. Full stop there.

In this case, it looks like the part of this lawsuit that is most damaging to Tesla is that it makes the news and many people won't read the details.

Just so we're clear, I believe the key facts of the lawsuit you're mentioning here are:

1. Employee was driving a Tesla.
2. The employee was not on-duty, but on a golfing day with friends.
3. Employee complained about FSD on the drive to the golf course.
4. There seems to have been lots of drinking at the golf course, then the employee and a witness/passenger got back in the car to drive home.
5. Employee had a blood alcohol content that was measured at 3 times the legal limit in the autopsy.
6. The only apparent witness to the crash was a golfing buddy and passenger, who had a "similar" blood alcohol content, and said the employee seemed fine to drive.
   1. The suing family members want to use that statement as evidence that the deceased was not too drunk to drive.
7. That witness reports that FSD or Autopilot was activated, and was doing crazy things, and wouldn't give back control to the driver
   1. As anybody who has used FSD or Autopilot has observed, the car immediately gives back control if you even tap the brake or move the steering wheel more than a few degrees.
8. The vehicle apparently left the road and hit a tree at high enough speed to cause the vehicle to catch fire, resulting in employee's death.

There is an ancient (60's and 70's when drink driving was a "skill rather than a crime") joke in Australia that the whole reason for a tree's existence is to stand on the side of the road, minding its own business, then to jump out in front of you on the way home from the pub.

It sounds pretty clear this person was hammered which means they weren't in an appropriate capacity to drive or supervise the car driving.

 

[CorneliusXX]

Tesla seems to be close to running out of FS Cybertruck buyers. I Just got my invite to order a FS CT with reservation that was placed October 2022 and I read they are already up to February 2023 reservation holders. I will be passing on my order.

Another example of "limited edition" being limited to how many can be sold.

 

[jkirkwood001]

6+ months of FS seems to be a good run. Elon already indicated non-FS orders should start next quarter.

It is good, but note that Canada still hasn't received Cybertruck. At 10% of the US population, they could have another 1,000 FS if / when they open the door. But I'm guessing that won't get approved before they switch to the non-FS in the US. And after that, buying a new 'Foundation Series' would kind of be a joke, I suppose, so Canadians will never get the option.

No big loss, I was going to decline and wait too. In fact, you could argue your non-FS resale value could be higher since owning a FS could be interpreted as "model with the kinks still weren't worked out"!

 

[nativewolf]

I've said this before, but I stopped caring about all these minor releases. I think most people just don't care. Just wake us up when it's completely unsupervised and there is an actual story to tell. Musk has stated FSD back in 2019, that it should be ready in 2020 to drive you from west to east coast completely. Maybe he can sit in the back of a Tesla and do it or see how far he gets to shut everyone up.

There is a lot of hate with Waymo here, but even when hitting a pole, it's still an actual real self driving service you can order NOW. Open to everyone in SF. Lots of testing in AZ. Another lawsuit again with Tesla employee dying, that FSD about to hit a train, another ~15? lawsuits pending.

The good thing is like the Cybertruck, so much competition in autonomy that Tesla will probably be forced to release something, even if it doesn't work perfectly. Even Toyota is getting in the game in China (yeah, not much confidence there even from me with their record).

There is not so much hate on Waymo, there are a couple of posters that are embarrassing to anyone with a brain that try to trash waymo (why I don't know) and others such as @Doggydogworld who are sound skeptics on all RT related things. The embarrassing folks post all day (have no lives, literally post over a dozen times a day) but have no real knowledge of ai, have not considered the business model, when confronted with truth they pretend it doesn't matter and change arguments, then 2 pages later go right back to lying about the same things.

At least when @DarkandStormy makes a post it is humorous even as he trolls away. It's the professional posters/uber bull types that destroy a conversation on RT/AI. Waymo has a clear and easy to understand strategy, is tremendously well capitalized, is cautious, has a great vehicle picked out (which will likely be powered by catl new 4680 cells- the irony), and is first mover. No one knows what happens now, will Waymo feel confident enough to launch the actual RT fleet? Geely has started hiring people to support operations in CA. @Doggydogworld doesn't believe Waymos model will turn out to be profitable (I think that is a fair summary) but I'm not so sure about that. In any case when DW makes his arguments that are rational and don't change if it makes Tesla look good or bad.

 

[nativewolf]

Agree to disagree @Usain. I'm pretty confident about my calls . Also I don't agree with your interpretation of what altman is saying, he also said it's foolish to try to anticipate AI more than 1-2 years into the future.

You haven't addressed a number of the points I made, but you seem to be convinced tesla has the right approach, while I genuinely believe they're manufacturing a growth story to keep the stock price above the required threshold. The timing of the announcements screams that. Occam's razor and all. 10 years in, just as we're consistently seeing negative growth on delivery metrics , fsd is suddenly ready? I don't buy that.

He never addresses points that demonstrate facts contrary to his vision for stock price , he has no idea how Tesla plans to make money with RT, he is just about hype. He doesn't understand AI, nor does he understand anything about Waymo. He claims Waymo will not be able to compete and will go under.....as if Google has ever had issues competing and claims that expensive vehicles will make Waymo uncompetitive- it is such a laughable failure of an analysis that I can only assume he is a paid shill. Heck google was even able to compete with Apple on smart phones (and yes extending that to RT is a very neat idea upon which to have a real conversation thus it should take place in the AI thread).

 

[rentierparasit]

No problem @rentierparasit.

I've talked about this before, but I don't consider Elias' tracker to be a good source of data. There is just too much variance in the participants, routes, subjectivity in reporting, etc.

I think it's great that Elias is doing what he is doing. Maybe it's the best we've got. But I just don't put any stock in it as a measure of progress.

I definitely wouldn't use it to guess the timing of robotaxi-level autonomy or of unsupervised FSD.

I suspect that Tesla has decent data based on performance against simulations. They probably also have good data based on their own professional drivers where they control for things like routes and strict reporting standards. Unfortunately, Tesla doesn't share this data with the public other than vague references from Elon.

So really, we're just left with our own impressions of progress from version to version. Plus, Tesla's recent level of commitment to the robotaxi effort is a good proxy for gauging their confidence in the FSD program.

Hi, Usain --

> I don't consider Elias' tracker to be a good source of data.

Ehh, this is very much in "agree to disagree" territory, there's no "proof" either way, no objective definition of what "good" would even mean, and no way to measure how far away from "good" the tracker is. This is an area where you could argue hammer-and-tong for months about every single little thing -- as people on the Internet have been known to do.

My personal opinion is that the tracker #s are in the ballpark -- say, within 2x of the "true" #s, whatever "true" even means here -- and that it certainly has value longitudinally, like, "Did miles/critical intervention increase by an order of magnitude from 10.x to 11.x?" -- I would just look at the tracker and say, "Definitely not." If somebody wanted to claim that 11.x was twice as good as 10.x on that metric, I'd say, "Perhaps, but there's no evidence in the tracker for that."

> we're just left with our own impressions of progress from version to version.

Ha! I live in NYC. I've spent a bit of time in taxis taking my cancer-patient girlfriend to and from the hospital, but don't remember the last time I was in a private car -- it would have to have been pre-COVID. So all I have to go on is other people's anecdotes. One could argue that the tracker is just distilled anecdotes; maybe true, but that's still something!


Yours,
RP

 

[NicoV]

And we thought "the wave" was a thing of the past...

View attachment 1060659

Electric Vehicle registrations in Europe: 15 countries, majority of BEV market

For the 6 real-time reporting countries together we’re now above Q1.

 

[sunwarriors]

Another example of "limited edition" being limited to how many can be sold.

Would this leave to a pretty massive depreciation as regular editions start rolling out? I saw some video of someone trying to sell theirs to avoid this big hit.