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

[EVNow]

Disengagement rate is a useless metric for a non-geofenced car and I don't know why people keep using this as a way to compare autonomous vehicles. There are routes FSDb will have a disengagement 100% of the time within 100 ft and there are routes you can drive for 1000 miles and not have any disengagements.

Disengagement rate is a very useful metric. The metric has nothing to do with geofencing.

It is very simple - when FSDb does something wrong - I disengage (or have to intervene). If left to itself, the car might get into an accident. So the disengagement rate is a proxy for miles per errors a.k.a. accident rate. Accident rates determine when an AV is ready for true driverless operation ... needs to be atleast a few thousand miles per intervention.

Ofcourse it would be best if Tesla published the numbers after actually sorting thorugh all the disengagements and interventions to figure out what would have actually caused an accident, what are just inconveniences or downright embarrassments for Tesla on the road. But they probably get millions of that per day and they can't actually analyze all the data manually. Anyway, they won't publish the data because miles/disenagement is so pathetically low and the media will laugh at them.

That is also the reason you are trying to completely discount the most important metric - because it suits your narrative. Not because it is actually not useful.

This is like saying why Gross Margin or EPS is actually not useful.

 

[SOULPEDL]

The issue was that S-K did too good of a job copying the LG technology. They even duplicated the "feature" that folded the cathode so that it would short out in the battery causing a fire. (Similar to the problem that GM and Hyundai had to deal with in the LG cells that they used.)

Holy Cathode! So now Ford is using LGs or other? If so, that was quick - as if they knew it was coming?‍

 

[MC3OZ]

My concern with FSD is the current pace of AI development. If Tesla solves FSD/ Robotaxi and the general state of AI is advanced enough where it can be deployed in every fleet within a year or two after Tesla is able to deploy it, then their competitive advantage will melt away.

The concern is not so much “Will FSD be solved?” Spoiler: It will.

The concern is how quickly after Tesla is it available to the competition.

At that point, then Tesla’s big advantage is in the library of data collection they’ve been doing and in having a fleet they can deploy it to—assuming HW3 is capable of supporting it.

3 big advantages Tesla has

1. Data - FSD probably performs better in locations with more training data.
2. Cost - Robotaxis cheaper than most of the competition.
3. Production volumes - especially Gen 3.

Without the data competitors are relying on simulation or have a geofenced solution.

Some are predicting Tesla is a decade away. I think we are closer but the project has been running for around a decade.

The language model AI is performing well, but is different to FSD.

At worst ChatGPT gives a wrong answer, with FSD the worst case is a crash. I thonk ChatGPT is hosted on internet, FSD needs to run in each car.

Users are happy if ChatGPT rexponds in a few seconds, for FSD every microsecond counts.

Perception is also harder, moving through a 3d world in varying light conditions as opposed to mostly correct words in a reasonable sentence.

My conclusion- it is a hard problem, that is why it is taking Tesla so long, and competitors will find it equally hard, and hard to find staff with the ability to make a difference.

 

[SOULPEDL]

Disengagement rate is a very useful metric. The metric has nothing to do with geofencing.

It is very simple - when FSDb does something wrong - I disengage (or have to intervene). If left to itself, the car might get into an accident.

Correct, however it's not quite that simple. Read the answer to my riddle today for the many other reasons and why I believe was just reduced by about 25%.

Meanwhile, I think I need a nap again... for real. Busy week, Assy shipped from China etc. Plus, having quit coffee and stuff, my body thinks I'm in a time change now and I'm experiencing the curse of that long nap that most fear.

 

[Ogre]

3 big advantages Tesla has

1. Data - FSD probably performs better in locations with more training data.
2. Cost - Robotaxis cheaper than most of the competition.
3. Production volumes - especially Gen 3.

Without the data competitors are relying on simulation or have a geofenced solution.

Some are predicting Tesla is a decade away. I think we are closer but the project has been running for around a decade.

The language model AI is performing well, but is different to FSD.

At worst ChatGPT gives a wrong answer, with FSD the worst case is a crash. I thonk ChatGPT is hosted on internet, FSD needs to run in each car.

Users are happy if ChatGPT rexponds in a few seconds, for FSD every microsecond counts.

My conclusion- it is a hard problem, that is why it is taking Tesla so long, and competitors will find it equally hard, and hard to find staff with the ability to make a difference.

Yeah, I'm looking at how fast AI is advancing and I don't think it's a decade away even for the competition.

Tesla likely 1-3 years. Competition needs a fleet for data collection, until they have that, they are at ground 0. That said, AI and hardware is much more advanced and it won't take them 8 years or more to get where Tesla is.

Hypothetical best case for competition:
Someone like VW partners with someone with auto ambitions and the technology skills to do this—maybe Apple??—and they install the hardware on 100% of their cars. They might develop a broad enough data-set 2-3 years after deployment.

There are a lot of big ifs in that hypothetical though. Someone needs to be willing to invest millions(billions?) in hardware on a bet which won't pay off for that at least a couple years. Right now legacy auto doesn't have the money to invest in this. They are already struggling with margins and profitability.

 

[AZRI11]

Tesla herding CATL to the U.S.

Biden administration response to this news.

 

[Olle]

My concern with FSD is the current pace of AI development. If Tesla solves FSD/ Robotaxi and the general state of AI is advanced enough where it can be deployed in every fleet within a year or two after Tesla is able to deploy it, then their competitive advantage will melt away.

The concern is not so much “Will FSD be solved?” Spoiler: It will.

The concern is how quickly after Tesla is it available to the competition.

At that point, then Tesla’s big advantage is in the library of data collection they’ve been doing and in having a fleet they can deploy it to—assuming HW3 is capable of supporting it.

That is a valid concern. Let's say that FSD becomes available to all manufacturers simultaneously.

Depending on when that happens, Tesla might at that time have amassed a huge advantage in volume and cost of robotaxis and charging infrastructure and dominate the market for a while.

 

[insaneoctane]

Ford (and Taycan) both announced price increases on the Pro, Lariat, and Platinum. Battery issue "resolved" and orders open again on (Edit: F-150 Lightning). Still not clear what it was exactly.

Ford announces another price increase for the F-150 Lightning

Ford has announced a price increase for its entry-level F-150 Lightning Pro. The automaker has increased the price of this trim level, which is designed for commercial use, from $57,869 to $61,869. The price hike […]

driveteslacanada.ca

Porsche Taycan prices increased for the 2024 model year

At a time when most automakers are lowering prices on their electric vehicles (EVs), Porsche has increased pricing across their entire Taycan lineup for the 2024 model year. The price increases have taken effect in […]

driveteslacanada.ca

Also, had to look it up BTFD (Buy the Fu***ng Dip). Michael, didn't understand us diehards nor the condition of the planet it seems.

"Ford said the standard-range Lightning Pro, a lower-cost version of the truck optimized for fleet use, will now start at just under $60,000, not including shipping. That’s roughly 50% higher than the Lightning Pro’s original starting price at launch last spring."

 

[M|S|M MYP]

"Ford said the standard-range Lightning Pro, a lower-cost version of the truck optimized for fleet use, will now start at just under $60,000, not including shipping. That’s roughly 50% higher than the Lightning Pro’s original starting price at launch last spring."

Classic bait and switch or.....is it a case of "Hey, we didn't know how much this was really gonna cost to build"

 

[Singuy]

Disengagement rate is a very useful metric. The metric has nothing to do with geofencing.

It is very simple - when FSDb does something wrong - I disengage (or have to intervene). If left to itself, the car might get into an accident. So the disengagement rate is a proxy for miles per errors a.k.a. accident rate. Accident rates determine when an AV is ready for true driverless operation ... needs to be atleast a few thousand miles per intervention.

Ofcourse it would be best if Tesla published the numbers after actually sorting thorugh all the disengagements and interventions to figure out what would have actually caused an accident, what are just inconveniences or downright embarrassments for Tesla on the road. But they probably get millions of that per day and they can't actually analyze all the data manually. Anyway, they won't publish the data because miles/disenagement is so pathetically low and the media will laugh at them.

That is also the reason you are trying to completely discount the most important metric - because it suits your narrative. Not because it is actually not useful.

This is like saying why Gross Margin or EPS is actually not useful.

Macro disengagement rate which Tesla has is what matters. PERSONAL disengagement rate is a useless metric and yet everyone uses their own anecdotal experience to judge FSD's progress. There are people who disengage every half a mile because they live in the heart of Manhattan. Then there are people who lives in North Dekota and thinks FSD is almost ready for robotaxies. Macro disengagement rates account for all of this. Your personal disengagement rates is as significant as a drop of water in the ocean. This is what happens when you don't geofence and I am 100% sure you understand this and you are arguing for the sake of arguing.

 

[2daMoon]

Biden administration response to this news.View attachment 923188

I herd that they'll calm down when the band plays Rawhide... (or, perhaps, Stand By Your Man)

 

[2daMoon]

That is a valid concern. Let's say that FSD becomes available to all manufacturers simultaneously.

Depending on when that happens, Tesla might at that time have amassed a huge advantage in volume and cost of robotaxis and charging infrastructure and dominate the market for a while.

If, by "dominate the market," you mean that Tesla will be the only company selling FSD to the other OEMs, then I concur.

 

[Singuy]

Production has slowed down, Fremont is in clean up mode. Lathrope has over 100 megapacks sitting on the logistic lot.
Perhaps the slow down is to give people a break for the end of Q push. Or perhaps they are introducing highland soon, in time for the SR+ demand cliff due to the IRA credit cutting in half.

 

[Artful Dodger]

Biden administration response to this news.[CATL to build U.S. factory]

This new CATL factory should be recast as reshoring LFP battery production to the USA after the bankruptcy of A123 Systems in Oct 2012. That company was founded at M.I.T. and headquartered in Massachusetts, then bought out in bankrupcy by a Chinese company. Lithium Iron Phosphate cathode (LFP) is an American invention.

So now that the 20-yr term U.S. patents on LFP have expired, China no longer has an competitive advantage in IP over American companies producing batteries with that same U.S. technology:

A Brief History of LFP, Patents, Licencing Costs, Pricing // and Tesla | The Limiting Factor (Sep 08, 2021)

 

[Artful Dodger]

That is a valid concern. Let's say that FSD becomes available to all manufacturers simultaneously.

Depending on when that happens, Tesla might at that time have amassed a huge advantage in volume and cost of robotaxis and charging infrastructure and dominate the market for a while.

Other carmakers catching up to FSD eventually is really not a valid concern. Elon told us literally years ago that all carmakers eventually would have self-driving technology. Tesla's long-term advantage will be in manufacturing:

Elon Musk: "Tesla's Long-Term Competitive Advantage will be in Manufacturing" | Cleantechnica (2020-08-21)​

1 week later, he sent this tweet:

Elon Musk on Twitter: "@scottwww @Tesla Pace of innovation is all that matters in the long run" / Twitter​

'Ladies and Gentlemen of the jury, I present the "unboxed plant" w. Tesla 'bots' (Mar 01, 2023)

[drops mic]

 

[ZeApelido]

Someone has been taking too many Stellantis crazy pills.

Stellantis CEO: Earth Doesn't Have Enough Lithium For EVs

At a Stellantis-hosted event, CEO Carlos Tavares remarked that there might not be enough lithium to replace every internal-combustion vehicle with an EV.

jalopnik.com

 

[Singuy]

Someone has been taking too many Stellantis crazy pills.

Stellantis CEO: Earth Doesn't Have Enough Lithium For EVs

At a Stellantis-hosted event, CEO Carlos Tavares remarked that there might not be enough lithium to replace every internal-combustion vehicle with an EV.

jalopnik.com

Earth has enough tombstones for car companies tho.

 

[nativewolf]

Earth has enough tombstones for car companies tho.

Just a simple wooden marker will do, I have a sawmill

 

[theschnell]

Norway still going gangbusters with the Model Y today. Can they reach the magical 10K in the quarter? This would be greater than the total ever sales of many EV models. Registreringer av nye elbiler i Norge This is incredible that in an EV saturated market that Tesla can get nearly 50% of all sales !!!!

As goes Norway, so goes the world (eventually)

 

[theschnell]

When I first started investing in Tesla Toyota was the company I was concerned about. It made sense that the Prius experience could have been used as a launch pad. Turns out I was wrong, so far at least. (not that I expect anything to change)

I wonder if the Prius through Toyota off the right track in the sense that with the possible exception of the years 2005 - 2008, I don’t think they were ever in high demand or very profitable.

Maybe they thought pure BEVs would also be low margin…

If Toyota produced them, they might have been…