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

[strangethingintheland]

Another interesting interview on autonomy.

A few key points:

• Elon thinks autonomous cars should be worth 5-10x non autonomous cars for the next 5-10 years.
• Elon expects Teslas to be an appreciating asset from here.
• The new FSD computer is able to run cameras at full frame rate full resolution non cropped and it still has headroom on one computer. The FSD computer is 2 SoCs with full redundancy. It is like a twin engine commercial aircraft, it will work best if both systems are operating, but it can operate safely on one. Currently there is no need to use both SoC.
• In 6 months hands on wheel detection may be removed for parts of autopilot. For this he thinks he needs statistical evidence that incidents per mile are 200% better on AP than human driving. Need to asses probability of crash, injury, permanent injury and death.
• Driver vigilance/attention is very quickly going to be a moot point. Maybe this year but definitely next year human intervention is going to reduce safety.
• Elon thinks the race for self driving cars is already "game set and match" to Tesla. "I don't want to be complacent or overconfident, but that is literally how it appears right now, I could be wrong but it appears to be the case that Tesla is vastly ahead of everyone."

I listened to the interview. The appreciating asset idea doesn't quite seem right. I believe that the cars can become revenue generating assets. But for the car to appreciate in value would imply that you couldn't buy a new Tesla for less, implying that Tesla will be continually raising the prices of new cars. Am I missing something? Granted, a revenue generating car is a pretty bold statement, but I think that's closer to what Elon is actually trying to say. Of course, I've been wrong before.

 

[Remus]

I dunno. Call Elon's view the rosy view of the future of Tesla autonomy, or vehicle autonomy in general.

My view is different.

• I'll stipulate that vehicle autonomy is inevitable. I'll grant that it is coming, nothing will stop it.
• I'll stipulate that in time it will be ubiquitous, the way smartphones popped up, scaled, and became ubiquitous, changing the world.
• I'll stipulate that Elon is probably right, and that Tesla is vastly ahead of everyone.
  
• BUT... I think autonomy technology will eventually--sooner than we realize--be a commodity, just like smartphones.
• The question is: will Tesla's autonomy technology be Android or iOS? Will it run in 90% of vehicles, or 10%?
• Another question: Will it matter to Tesla shareholders? Same question applies to Tesla's overall global vehicle market share by the 2030s. Would 5% share be enough? 10%? If 10% got you to $800 billion market cap, would that be victory?

I see a future where many old cars are junked, and most of the rest are converted to be autonomous via add-on kits, with a small minority still driven manually, until laws ultimately ban manual cars from public roads, or at least require special (expensive) permits. I see a future where Tesla most likely becomes be the iOS of autonomous technology, in a crowded market of a thousand commodity Android variants. That crowded market consists of offerings from, who knows, Apple, Microsoft, Amazon, Google, and probably some other company not known yet.

This week's move to bundle Autopilot by default in M3 orders is the beginning of the commodification. Years from now, the public won't really care who makes their cars autonomous. There will be a "good enough" level that lets other autonomy suppliers thrive. Will autonomy-powered car-sharing networks really be the key to Tesla climbing to $4000/share as ARK believes? Maybe. Will competitors figure out autonomy/car-sharing networks that are "good enough" to compete with Tesla? I'm sure they will. In the tech biz, there are always fast followers.

In the end will the public care about the Tesla version versus others? Some will, most I expect won't, just like most people buy Android smartphones because they're "good enough" and they don't care to pay the tax for premium Apple products. But enough will buy into the Apple branding/positioning of Tesla to give it a shot at a near-trillion-dollar valuation. Which is why I'm in for the long run. But I don't think the long run is Tesla dominating everything. The pie will be huge and many players can have healthy slices.

I don't see full autonomy achieved anytime soon, but, if it's here, then pretty much the winner take all. Because to scale out you just need money.

If Tesla demonstrates that a human driver can manage as little as 5 cars, then Wall Street will beg Tesla to take their money.

 

[Carl Raymond]

I dunno. Call Elon's view the rosy view of the future of Tesla autonomy, or vehicle autonomy in general.

My view is different.

• I'll stipulate that vehicle autonomy is inevitable. I'll grant that it is coming, nothing will stop it.
• I'll stipulate that in time it will be ubiquitous, the way smartphones popped up, scaled, and became ubiquitous, changing the world.
• I'll stipulate that Elon is probably right, and that Tesla is vastly ahead of everyone.
  
• BUT... I think autonomy technology will eventually--sooner than we realize--be a commodity, just like smartphones.
• The question is: will Tesla's autonomy technology be Android or iOS? Will it run in 90% of vehicles, or 10%?
• Another question: Will it matter to Tesla shareholders? Same question applies to Tesla's overall global vehicle market share by the 2030s. Would 5% share be enough? 10%? If 10% got you to $800 billion market cap, would that be victory?

I see a future where many old cars are junked, and most of the rest are converted to be autonomous via add-on kits, with a small minority still driven manually, until laws ultimately ban manual cars from public roads, or at least require special (expensive) permits. I see a future where Tesla most likely becomes be the iOS of autonomous technology, in a crowded market of a thousand commodity Android variants. That crowded market consists of offerings from, who knows, Apple, Microsoft, Amazon, Google, and probably some other company not known yet.

This week's move to bundle Autopilot by default in M3 orders is the beginning of the commodification. Years from now, the public won't really care who makes their cars autonomous. There will be a "good enough" level that lets other autonomy suppliers thrive. Will autonomy-powered car-sharing networks really be the key to Tesla climbing to $4000/share as ARK believes? Maybe. Will competitors figure out autonomy/car-sharing networks that are "good enough" to compete with Tesla? I'm sure they will. In the tech biz, there are always fast followers.

In the end will the public care about the Tesla version versus others? Some will, most I expect won't, just like most people buy Android smartphones because they're "good enough" and they don't care to pay the tax for premium Apple products. But enough will buy into the Apple branding/positioning of Tesla to give it a shot at a near-trillion-dollar valuation. Which is why I'm in for the long run. But I don't think the long run is Tesla dominating everything. The pie will be huge and many players can have healthy slices.

Ark Invest keep referring to autonomous taxis as a 'natural monopoly'. Water is a natural monopoly, it makes no sense to have two pipe networks. I'm not convinced robotaxis are a natural monopoly. Happy to be convinced otherwise.
I do see Tesla winning the race and gaining massive first mover advantage. Even if narrowly beaten at the start, Tesla still win the race with a leapfrog move, by deploying to a huge preexisting fleet.

 

[M3Rider]

I think that's delusional. Here in Texas lawyers run "Hit By a Company Vehicle?" ads because awards can be 10-100x higher for the same injury. No company can afford to operate a robotaxi service without a liability shield. And lawmakers won't create liability shields for a 2x improvement. Especially not a 2x Muskstat. You'll need something like 10x verified by independent 3rd party, at a minimum.

Don't know how this will work out, but those companies that are sued survive somehow today? Insurance will definitely be involved and the rates should correspond to payouts?

Yeah, this seems like a new field to be explored, but putting a wall on the way of improving safety in the form of 100x the normal payout does not make much sense, that would be blocking the progress. It seems Elon is confident they can work something out with regulators of they show 2-3x improvement.

 

[j6Lpi429@3j]

some things I pay commodity prices for. Like I dont give a damn who makes my socks. But if I'm paying for an AI to drive me around and stop me dying, you bet your ass I want the best. I'll have the tesla autonomy service thanks, not the walmart version.

 

[MarcusMaximus]

some things I pay commodity prices for. Like I dont give a damn who makes my socks. But if I'm paying for an AI to drive me around and stop me dying, you bet your ass I want the best. I'll have the tesla autonomy service thanks, not the walmart version.

There’s no investment I’ve found that pays off more than investing in good socks. Being able to walk/run comfortably makes an enormous difference in what you can accomplish. Whatever else you skimp out on, socks are too important.

 

[generalenthu]

The 2023 numbers are very close to mine. I saw FCA’s CO2 targets to be 91 g/km (95 is the overall average EU target, but the manufacturer's individual targets are biased due to average fleet vehicle weight).

I need to review again how those EV multipliers work. I thought it was something different than how you used it but I may have been mistaken.

Tesla cars are heavier, bumping up the target, which is a second order benefit.

 

[tinm]

some things I pay commodity prices for. Like I dont give a damn who makes my socks. But if I'm paying for an AI to drive me around and stop me dying, you bet your ass I want the best. I'll have the tesla autonomy service thanks, not the walmart version.

I don't disagree. But I'd argue that in the tech world, Microsoft is the "Walmart version" and Microsoft is currently the world's most valuable company. History shows that they are notorious fast followers. They will find a way to *try* to compete in autonomy. They may fail miserably like they did with smartphones. I expect they'll fail. But I think in time, given the sheer size of the world vehicle market and the opportunity to be a player in it, lots of companies will try to do what Tesla is doing and some of these companies will find some degree of success in the long run. It's inevitable. I'm not yet sold on a "winner take all" outcome for vehicle autonomy. Few things ever work out that way in tech.

 

[kbM3]

I listened to the interview. The appreciating asset idea doesn't quite seem right. I believe that the cars can become revenue generating assets. But for the car to appreciate in value would imply that you couldn't buy a new Tesla for less, implying that Tesla will be continually raising the prices of new cars. Am I missing something? Granted, a revenue generating car is a pretty bold statement, but I think that's closer to what Elon is actually trying to say. Of course, I've been wrong before.

IF Tesla can achieve their network before everyone else, each car would become very valuable. I would assume this would cause upwards pricing pressure on all of their vehicles. If someone can buy a $45k car that can generate that much per year, I would guess demand would go through the roof and thus lead to price increases or super long waitlists.

 

[carterm2]

Time travel...

a monthly chart showing a rising channel. One way to look at the "trend".

 

[M3Rider]

some things I pay commodity prices for. Like I dont give a damn who makes my socks. But if I'm paying for an AI to drive me around and stop me dying, you bet your ass I want the best. I'll have the tesla autonomy service thanks, not the walmart version.

Btw, it's not just AI. The car safety rating matters too. When two cars collide, it does matter which one is built better.

 

[heltok]

I listened to the interview. The appreciating asset idea doesn't quite seem right. I believe that the cars can become revenue generating assets. But for the car to appreciate in value would imply that you couldn't buy a new Tesla for less, implying that Tesla will be continually raising the prices of new cars. Am I missing something? Granted, a revenue generating car is a pretty bold statement, but I think that's closer to what Elon is actually trying to say. Of course, I've been wrong before.

Uber said if Tesla makes a FSD car they would buy 500k/year of them. If we assume that Tesla is able to provide FSD this year and they get legal approval in some states there will be a very high initial demand for them as they will be worth 0-100%(lets say 50%) of like 2 full time human drivers = $50k/year. If you could buy a car that you knew would make you ~$50k/year how much would you be willing to pay for it?

Maybe after 22April we will see some companies stockpiling black Model 3 LR as a speculative investment.

 

[BioSehnsucht]

Elon stated he currently has the FSD that stops for red lights. I would assume the 10% usage is given for the real Beta FSD, not for AP.

It's more likely that the 10% usage is with the existing NN for HW2/2.5 that has reduced resolution/etc but running on HW3 (and perhaps with running two copies for redundancy), since that is more inline with the claimed performance increase from HW2/2.5 to HW3.

 

[KarenRei]

Pier 80 needs a ship:

Morning Cornelia was supposed to arrive yesterday.

 

[Dan Detweiler]

I listened to the interview. The appreciating asset idea doesn't quite seem right. I believe that the cars can become revenue generating assets. But for the car to appreciate in value would imply that you couldn't buy a new Tesla for less, implying that Tesla will be continually raising the prices of new cars. Am I missing something? Granted, a revenue generating car is a pretty bold statement, but I think that's closer to what Elon is actually trying to say. Of course, I've been wrong before.

I believe the "appreciating asset" comment was referring to the ability for the car to become better through OTAs.

Dan

 

[Remus]

some things I pay commodity prices for. Like I dont give a damn who makes my socks. But if I'm paying for an AI to drive me around and stop me dying, you bet your ass I want the best. I'll have the tesla autonomy service thanks, not the walmart version.

And the first mover also have the biggest data pool for further improving their technique.

It's like Google, they get better as more people use them, and more people use them as they get even better.

It will be winner take all IF Tesla get there first. Because naturally you want the taxi to be an EV for low operational cost, nobody gives a dam about panel gaps, colors and options, it's the miles per kWh and the battery degradation that matter the most.

 

[KarenRei]

New lot clearing east of Lathrop over the past several weeks:

 

[Remus]

I don't disagree. But I'd argue that in the tech world, Microsoft is the "Walmart version" and Microsoft is currently the world's most valuable company. History shows that they are notorious fast followers. They will find a way to *try* to compete in autonomy. They may fail miserably like they did with smartphones. I expect they'll fail. But I think in time, given the sheer size of the world vehicle market and the opportunity to be a player in it, lots of companies will try to do what Tesla is doing and some of these companies will find some degree of success in the long run. It's inevitable. I'm not yet sold on a "winner take all" outcome for vehicle autonomy. Few things ever work out that way in tech.

But this thing has a network effect. Microsoft never conquered web search despite of Bing got pretty good, Google never conquered social network

 

[kbM3]

And the first mover also have the biggest data pool for further improving their technique.

It's like Google, they get better as more people use them, and more people use them as they get even better.

It will be winner take all IF Tesla get there first. Because naturally you want the taxi to be an EV for low operational cost, nobody gives a dam about panel gaps, colors and options, it's the miles per kWh and the battery degradation that matter the most.

Kind of ironic all of that attention to acceleration and driving dynamics to make one of the world’s greatest driving sedans, will pretty much go to waste

 

[Causalien]

To take one example of road hazard detection:

It would seem that the car has to classify every object in the road along a spectrum of how safe it is to hit, vs. how imperative it is to avoid.

To get to 99%, it may be enough to identify pedestrians, wheels, plastic bags, roadkill and potholes.
To get to 99.99%, there may be hundreds of different road hazards, going up to 100’s of thousands as you march to 99.999%.

Am I missing something? Does someone know how Tesla can handle edge case explosion?

Assuming TSLA ai is learning from human driving. The programmers will be identifying which drivers are safe drivers to learn from and in special accidents, accept or reject the reactions.
This way, the NN eventually learns to identify patterns that are similar and responds from the millions of cases that human drivers have responded before. So a lot of the edge cases, like trafic cones of different shapes will have similar reactions. The NN itself sometimes evolve weird structures to do the same things humans do. Take for example the identification of the traffic cone. Human will write a code that scans for a triangle shape to identify something as a traffic cone, a NN might evolve some sort of FFT algorithm to analyze the visible red specturm with a low pass filter to get rid of noise response that somehow only uses 2 gates by taking advantage of gated clock propagation delays between two xor gates. Algorithms that humans cannot make sense of but works. These type of special NN evolutions that relies on hardware tends to break as hardware changes. Evident in how autopilot seems to regress a bit with each upgrade before getting smooth again.

The important part is that someone needs to be there to tell the NN it is wrong or right on more important evolutions, but in everyday situation, the NN can just assume no accident = right.