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

[Singuy]

If quoting your post back to you and pointing out none of the claims are in line with facts is annoying, maybe post less wrong things?




When you said one of the reasons low sub rate was that nobody wants to use beta SW, despite it not being beta anymore, that was a waste of everyone's time.

I simply corrected you.





Where are you getting that # from?

Are you under the hilariously wrong impression they put data center chips in the cars?


Again- Tesla does not make chips and neither does Nvidia.

They both pay an actual chip fab for that.



Legit- not sure what happened, but the quality of your posts from a fact or data perspective had gone very badly down hill.

Okay, sorry, The FSD chip that was fabbed by TSMC/Samsung who bought their EUV machine from ASML and got the PCB board made somewhere in China I don't care to look up. Probably rather chew glass than to continue this conversion of you strawmening around.

 

[cliff harris]

Tesla's cost is so low they give it away for free on every car.

IMHO this is underappreciated. Any FSD wannabe is going to initially be very conservative, and at the MOST might consider a placement for 100,000 FSD-capable in-car units.
Thats just laughably low by Tesla standards, who are probably using almost identical hardware in the 3 and Y, and to some extent the S,X and CT as well?
Even in some laughably unlikely future scenario where magically some other company solves FSD and has the tech to deploy it to a suddenly produced fleet of competing EVs, they simply will not be able to compete on cost with Tesla.

I think other companies wont be just licensing FSD software. They will have to pay Tesla to design, and install the compuet hardware in their cars, and likely handle all their over-the-air needs as well.
Essentially they will just become panel-bashers making different shaped surrounds for the real product, which will be Tesla computer hardware

 

[advocate8]

Why? The average person doesn't drive remotely enough miles 98% of days to sleep while the car is driving. Most trips are short. I think for many being able to (legally) read a book or email while in the car would be a nice to have- not an immediately replace existing car to have--- (and hey, you can do that TODAY in a Mercedes in very specific conditions- the ones most likely for it to be useful (stuck in slow highway traffic for a while) and everyone here seems to think that's Super Dumb.

99.7 percent of those fully freed from driving will just get on the net and waste more of their lives shopping, looking at porn, arguing on forums etc.
The remaining 0.3 will be working due to deadlines.

 

[EQC_]

I agree that sleep in the back for LONGER drives is a nice to have, but I go back to the cost of whether the uptake will be that high depending on price. $299/month (random $$ posted here often) is way too high for me to pay for that. ....

This is a big hypothetical and a long way off...but I imagine the occasional ~400 mile trip, something like the LA area to the San Francisco area. For that, especially for a small family traveling, and especially overnight while they sleep, $300 would be worth it for even just one trip.

Spoiler: Current and future travel options

The way I see it:

• Current Option 1: Drive yourself. You'll basically kill a day driving. However, especially if there is more than one person (ie: a couple, a family, or a group for work), this is the most economical.
• Current Option 2: Fly. Depending on your proximity to the airport, this means time spent driving to the air port and parking (an hour or more), time between arrival and departure (they recommend 2 hours...but I know many people don't take that seriously), the flight itself (about an hour), then time spent rounding up luggage and getting a rental car afterwards. Direct costs are the airfare, parking at the airport, and the rental car, and when you add things up, it doesn't actually save THAT much time relative to driving yourself. This probably makes sense for a single traveler...airfaire and rental car isn't too much. But, if a couple people need to do the travel, or if it's a family, the airfare adds up and starts to make driving a big money saver.
• Far Future Hypothetical Option 3: FSD, overnight. I'm picturing a "personal car/minivan sized" version of the Night Bus in Harry Potter, with FSD...so yes, very far off. We'd need a vehicle better set up for sleep, and automated superchargers, and we'd probably need a hotel-alternative that lets you check in in the morning to freshen up. By travelling overnight at each end of the trip, you can probably eliminate the "lost" day trevelling by human-driven car or flying, and possibly cut off a night or two of hotel. Simpler, more convenient, and more direct than having to move into and out of airports too.

The average person doesn't drive remotely enough miles 98% of days to sleep while the car is driving.

Very true. Even carpooling to work, on a commute that's well over an hour, the best I can do is doze uncomfortably on the days somebody else drives. Even in the morning, you still have to get up, get dressed and mentally alert, and then get in the car to ride.

On the other hand...if Optimus can carefully lift me out of bed and put me in the FSD car still asleep...and then, during the drive while I sleep, Optimus will dress me, give me a quick shave and hairstyle while I sleep, and prepare me a quick breakfast and coffee before gently waking me as the car arrives in the work parking lot....

 

[mongo]

Why? The average person doesn't drive remotely enough miles 98% of days to sleep while the car is driving. Most trips are short. I think for many being able to (legally) read a book or email while in the car would be a nice to have- not an immediately replace existing car to have--- (and hey, you can do that TODAY in a Mercedes in very specific conditions- the ones most likely for it to be useful (stuck in slow highway traffic for a while) and everyone here seems to think that's Super Dumb.

Certainly there's some outliers but not enough to be terribly meaningful on rate of replacement for the overall vehicle fleet in any short time period.

The ability to wake up 700 miles from where you fell asleep without needing a plane/ train/ bus ticket would do wonders for tourism.
More than 700 if you consider early/ late drowsy period. We'd use 3 drivers to cannonball from Michigan to Florida for vacation. Worked, but hurt a bit.

 

[Knightshade]

Okay, sorry, The FSD chip that was fabbed by TSMC/Samsung who bought their EUV machine from ASML and got the PCB board made somewhere in China I don't care to look up. Probably rather chew glass than to continue this conversion of you strawmening around.

I notice you still haven't provided any sources for any of the #s you keep making up and instead keep moving the topic away from whatever was last debunked.

As I suggested, if you find being called out on making claims you can't actually support, and throwing out numbers that are clearly fictional and when asked for sources you change the topic, maybe do less of that and try providing actual factual things in your posts going forward?

 

[Knightshade]

The ability to wake up 700 miles from where you fell asleep without needing a plane/ train/ bus ticket would do wonders for tourism.
More than 700 if you consider early/ late drowsy period. We'd use 3 drivers to cannonball from Michigan to Florida for vacation. Worked, but hurt a bit.

Sure- for folks who already own such a vehicle it'd probably ding air (and rental car) revenue a bit.... Right now if I'm going anywhere much more than 400-500 miles I just fly and (if needed- but sometimes it's not) rent a car on arrival. Actual L5 self driving could replace that for sure.

I don't think anyone's running out to buy a new car based on they can save a couple short-haul plane tickets a year though.

 

[Singuy]

I notice you still haven't provided any sources for any of the #s you keep making up and instead keep moving the topic away from whatever was last debunked.

As I suggested, if you find being called out on making claims you can't actually support, and throwing out numbers that are clearly fictional and when asked for sources you change the topic, maybe do less of that and try providing actual factual things in your posts going forward?

My post gets people the general idea (ie, Nvidia makes high margins on their chips). Is it 80%? They don't exactly break down their margins from Auto so it's hard to know, but it's not 20% or 10% when historically they have been making 60%-65 margins on their revenue despite being massively diluted by their gaming/console division and without the massive datacenter revenue they have now.

And are their Orin hardware datacenter chips? Why yes they are a cut down version of their inference chips put into data centers.

 

[AudubonB]

Endit.

You know who you are.

Now.

 

[mongo]

Sure- for folks who already own such a vehicle it'd probably ding air (and rental car) revenue a bit.... Right now if I'm going anywhere much more than 400-500 miles I just fly and (if needed- but sometimes it's not) rent a car on arrival. Actual L5 self driving could replace that for sure.

I don't think anyone's running out to buy a new car based on they can save a couple short-haul plane tickets a year though.

Anecdotes != data, but we are...
Trip to Canaveral for Falcon Heavy launch took most of two days in an ICE vehicle with iron bladders. Model X at the time would have been three days with reasonable hours/day of driving. FSD would get it closer to just over a day. Major cost savings versus parling fees, two plane tickets, and rental car. Plus, overnight FSD saves a hotel room.

 

[EQC_]

...

They absolutely did not.

You can tell because it's not even compatible with Cybertruck 6 months after launch... (not even basic AP, let alone FSD),

So integrating it into other OEMs cars would be a years long endeavor--- and Tesla told you this on the Q1 call too.

100% agree that integrating with another OEM would take years.

We know the development cycle for cars. Even if some other OEM committed TODAY to add FSD to 100% of ONE of their models, it would take years before they could design and manufacture that car with all the needed items integrated: 8+ cameras, the FSD computer, a method of activating FSD, plus fully automatable steering, lights, acceleration, braking/regen, windshield wipers, etc. etc. all connected and controllable by the FSD system.

And, of course, that OEM would still hedge their bets. They wouldn't want to put all the FSD stuff on 100% of even one model...they'd want to be able to offer a cheaper version with no FSD hardware. Maybe some version with only forward-looking collision avoidance so probably some other driver assist system. And then the FSD hardware and software would only be on the top-end version. All that complexity for tiers/options would make the full design and development cycle take even longer.


For the Cybertruck:
Do we know what needs to happen for Autopilot and FSD to work on it?

In my (very dumb) brain, it seems like applying the cameras in (roughly) the same locations on a new model should really just require that some parameters about the size of the vehicle need to be updated within the models FSD uses. Thinking a little more deeply, I could also imagine that the FSD system also needs to know the dynamics of that car -- FSD needs to predict how it will respond to braking/acceleration/turning controls in a wide variety of situations, including varying road surfaces, varying weather (wet roads, etc.), road angles (hills and banks), load and distribution of weight in the vehicle, etc.

Tesla being so software-oriented, I would think they'd have computational models of their vehicles for both the size/dimensions and for the vehicle dynamics noted above. But maybe they need a bunch of miles of real data to prove/ground truth those models?

Or maybe some of the "new" stuff in the Cybertruck -- perhaps the rear wheel steering -- requires additional real-world data since that wasn't used in previous Teslas?

Is there something I'm missing...some other "real world" data that Tesla would need to collect for a new car model before FSD can be enabled?

Hopefully, once FSD can handle a range of car sizes -- from something smaller than a Model 3, up through the Cybertruck and the Semi -- hopefully it is just a few parameter changes to enable it on any arbitrary new car that has all the hardware in place.

 

[uscbucsfan]

Anecdotes != data, but we are...
Trip to Canaveral for Falcon Heavy launch took most of two days in an ICE vehicle with iron bladders. Model X at the time would have been three days with reasonable hours/day of driving. FSD would get it closer to just over a day. Major cost savings versus parling fees, two plane tickets, and rental car. Plus, overnight FSD saves a hotel room.

I don't think that's a realistic scenario. People aren't going to spend 24+ hours on the road only stopping to charge every 2-4 hours. It's just not comfortable. Most will still get hotel rooms, but with less stress and more road time than usual.

 

[uscbucsfan]

100% agree that integrating with another OEM would take years.

We know the development cycle for cars. Even if some other OEM committed TODAY to add FSD to 100% of ONE of their models, it would take years before they could design manufacture that car with all the needed items integrated: 8+ cameras, the FSD computer, a method of activating FSD, plus fully automatable steering, lights, acceleration, braking/regen, windshield wipers, etc. etc. all connected and controllable by the FSD system.

And, of course, that OEM would still hedge their bets. They wouldn't want to put all the FSD stuff on 100% of even one model...they'd want to be able to offer a cheaper version with no FSD hardware. Maybe some version with only forward-looking collision avoidance so probably some other driver assist system. And then the FSD hardware and software would only be on the top-end version. All that complexity for "options" would make the full design and development cycle take even longer.


For the Cybertruck:
Do we know what needs to happen for Autopilot and FSD to work on it?

In my (very dumb) brain, it seems like applying the cameras in (roughly) the same locations on a new model should really just require that some parameters about the size of the vehicle should need to be updated within the models FSD uses. Thinking a little more deeply, I could also imagine that the FSD system also needs to know the dynamics of that car -- FSD needs to predict how it will respond to braking/acceleration/turning controls in a wide variety of situations, including varying road surfaces, varying weather (wet roads, etc.), road angles (hills and banks), load and distribution of weight in the vehicle, etc.

Tesla being so software-oriented, I would think they'd have computational models of their vehicles for both the size/dimensions and for the vehicle dynamics noted above. But maybe they need a bunch of miles of real data to prove/ground truth those models?

Or maybe some of the "new" stuff in the Cybertruck -- perhaps the rear wheel steering -- requires additional real-world data since that wasn't used in previous Teslas?

Is there something I'm missing...some other "real world" data that Tesla would need to collect for a new car model before FSD can be enabled?

I think it's logically the SBW/Rear wheel steering more than anything else.

That being said, the Highland still isn't able to use FSD if Google is correct. So maybe camera positions is more difficult, but things progress and training progresses.

 

[Singuy]

100% agree that integrating with another OEM would take years.

We know the development cycle for cars. Even if some other OEM committed TODAY to add FSD to 100% of ONE of their models, it would take years before they could design manufacture that car with all the needed items integrated: 8+ cameras, the FSD computer, a method of activating FSD, plus fully automatable steering, lights, acceleration, braking/regen, windshield wipers, etc. etc. all connected and controllable by the FSD system.

And, of course, that OEM would still hedge their bets. They wouldn't want to put all the FSD stuff on 100% of even one model...they'd want to be able to offer a cheaper version with no FSD hardware. Maybe some version with only forward-looking collision avoidance so probably some other driver assist system. And then the FSD hardware and software would only be on the top-end version. All that complexity for "options" would make the full design and development cycle take even longer.


For the Cybertruck:
Do we know what needs to happen for Autopilot and FSD to work on it?

In my (very dumb) brain, it seems like applying the cameras in (roughly) the same locations on a new model should really just require that some parameters about the size of the vehicle should need to be updated within the models FSD uses. Thinking a little more deeply, I could also imagine that the FSD system also needs to know the dynamics of that car -- FSD needs to predict how it will respond to braking/acceleration/turning controls in a wide variety of situations, including varying road surfaces, varying weather (wet roads, etc.), road angles (hills and banks), load and distribution of weight in the vehicle, etc.

Tesla being so software-oriented, I would think they'd have computational models of their vehicles for both the size/dimensions and for the vehicle dynamics noted above. But maybe they need a bunch of miles of real data to prove/ground truth those models?

Or maybe some of the "new" stuff in the Cybertruck -- perhaps the rear wheel steering -- requires additional real-world data since that wasn't used in previous Teslas?

Is there something I'm missing...some other "real world" data that Tesla would need to collect for a new car model before FSD can be enabled?

I think it's their safety validation process that takes 3-6+ months. It is the reason why Tesla seems to hold these new "blow your mind Vxx" fsd for months before release. Remember how people speculated that V12 was going to be worst than v11 hence they wouldn't release the thing for months on end despite the hype?

 

[cliff harris]

For the Cybertruck:
Do we know what needs to happen for Autopilot and FSD to work on it?

TBH none of us know, but I would guess that before they make FSD live on a new model like the CT, they will, at the very least, run FSD in shadow mode on a bunch of CTs for quite a while. They may know everything they need to know in order to enable it, but nothing will keep the safety experts happy as much as being able to say 'FSD was simulating what it would do in 50,000 miles of real world cybertruck operation, and we didn't detect any vast discrepancies between the proposed driving, and what the humans did'.

Even with that, I wouldn't be surprised if they enabled highway-only autopilot for CT first, and let that run an additional month or so before switching on city-streets FSD. I'd rather Tesla be too cautious than not cautious enough here. From an investor POV, cybertruck subscription revenue is likely yo be trivial until at least Q1 2025 anyway.

 

[uscbucsfan]

cybertruck subscription revenue is likely yo be trivial until at least Q1 2025 anyway.

It will be 0 even if turned on today. FSD is included with the Foundation Series. There's no hint of non-Foundation coming this year.

Q1 2025 is more like Q2-Q4 2025.

 

[Papafox]

Percent of TSLA selling tagged to shorts was 69% today, I've never seen it that high. Also, the stock price got away from the big option sellers even with that effort and closed more than $2 above their 175 line in the sand. Bullish!

UPDATE:
The newest shorting report for TSLA's Friday now says 59%. That's still very high and represents a failure of their attempts to keep TSLA below 175 for the big options close this week. Still bullish.

 

[Max Plaid]

Sure- for folks who already own such a vehicle it'd probably ding air (and rental car) revenue a bit.... Right now if I'm going anywhere much more than 400-500 miles I just fly and (if needed- but sometimes it's not) rent a car on arrival. Actual L5 self driving could replace that for sure.

I don't think anyone's running out to buy a new car based on they can save a couple short-haul plane tickets a year though.

Don't agree with this... I only fly if I have to due to environmental issues, plus I just dislike the whole experience. My wife Will only fly if she has no other option, she's terrified of flying. So we drive pretty much every where. If I had a car that could do that while I slept, or read a book, that would be extremely valuable to me

But I'm in Europe, we don't have FSD, just an old version of EAP, so can't judge the progress

 

[sunwarriors]

This is a big hypothetical and a long way off...but I imagine the occasional ~400 mile trip, something like the LA area to the San Francisco area. For that, especially for a small family traveling, and especially overnight while they sleep, $300 would be worth it for even just one trip.

Or just catch a flight that's 1.5s hours for something that's safer than driving? Even if FSD is safer, there are many bad drivers on the road as we probably all can agree on. I checked Southwest and there are flights as cheap as $46/1 way. You still need to charge on the way there as well as charge after arriving which isn't free (and expensive in CA).

Having a car at your destination is nice, especially if it's your own car with all your stuff, but I am not sure I agree with you that there will be many takers at $300/trip, add in all the charging/time/higher risk driving vs. a flight, even for a small family.

 

[indysurfer]

Heads should roll. I'm not even sure if I'm kidding or not...

I don't think it cost Tesla a bunch of money or time to push out a free 30 day demo to everyone. If anything, it has probably been a net positive with all of the new data and input they are getting from the fleet.

I assume they will likely provide free demos again as new features get added or perfected (Banish?)

When the system gets really good, I imagine Tesla will do some real TV advertising too. So few people (even Tesla owners) have a clue about the abilities of FSD.

BTW - I have at Tesla tech working on my Y right in the driveway now fixing an ultrasonic sensor so I can finally try out Summon and Autopark.....hopefully.