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

[Causalien]

I feel that part of the reason of this analyst meeting was to make clear how nonsensical the valuation of Uber is, with a competitor that will be able to undercut Uber and Lyft on price, and be very profitable at it. And that Tesla deserves that Uber valuation.

Either tsla rise up to include Uber valuatoin or Uber comes down to TSLA's autonomy only valuation.

 

[tinm]

One thing that struck me about today's event: how different Jobs and Musk are.

Think back to Jobs announcing the iPhone in 2007. It was a monumental/historical event, a game-changing milestone that fundamentally changed the industry and launched "mobility." Sure there had been hints it was coming with Palm Treos and Blackberries, but the iPhone blew everyone away and set Apple on a path that would take it to a trillion-dollar market cap.

Now think of Elon's presentation today. I'd argue the event, in hindsight, has the potential to be seen as a game-changing milestone that fundamentally changed an industry and launched "robotaxis." Sure there have been hints it's coming, with Waymo et al. But what Tesla showed off today was pretty mind-blowing and sets Tesla on a path that, maybe?, takes it to a trillion-dollar market cap.

Now think if Jobs had announced the iPhone in 2006, or 2005, a year or two years before it was out and purchasable. And imagine how Wall St would have reacted. They probably would have been "meh" and "show me." Jobs waited until the iPhone was imminently available before announcing it. People realized this was gonna be obtainable in days or weeks. Musk is announcing a robotaxi network that's still a ways out, obviously to alert the market about the difference between Lyft/Uber (both IPOs) and Tesla's competing offering. But Wall Street and the media are at peak skepticism/FUD when it comes to Tesla. Elon could have announced that oh yeah, the our HW3 proprietary chip's spare trillions of compute operations available beyond FSD in the cars will be used for a cure for cancer, which will be done within 18 months of rollout. I don't think even that would move the stock. I'm not sure anything from today is going to be a big boost to the stock any time soon until Tesla shows it is executing brilliantly. No more fumbles, like announcing full closure of stores one day and then walking back that announcement 48 hours later. The market is still reeling from that, for good reason.

 

[powertoold]

Imagine Tesla sticks to $35K+ cars and then $15K FSD capable econobox BEVs start hitting the market. How much will riders pay extra to be brought to the airport or work by Tesla whom they know make much more money, as the novelty wears off?

I doubt any other car manufacturer will be able to beat Tesla in efficiency, safety, durability, OTA software, etc. etc. any time soon. People car about safety and reliability. They'll definitely pay more for Tesla for the time being.

 

[Remus]

The mission = changing the WORLD to sustainable transport. Can’t do that if you don’t sell to everyone.

You got it backwards, how would an ordinary joe get the car? Rich people can hire an army to buy all cars from Tesla for them, leaving others with no cars.

 

[Cloxxki]

Was I the only one cringing when Elon felt he had to re-state that all cars made today will be able to FSD? Wasn't AP1 supposed to be that, then 2, then 2.5? And even then, which member of the audience wasn't up to speed with this yet?

 

[powertoold]

Now think if Jobs had announced the iPhone in 2006, or 2005, a year or two years before it was out and purchasable. And imagine how Wall St would have reacted. They probably would have been "meh" and "show me." Jobs waited until the iPhone was imminently available before announcing it. People realized this was gonna be obtainable in days or weeks. Musk is announcing a robotaxi network that's still a ways out, obviously to alert the market about the difference between Lyft/Uber (both IPOs) and Tesla's competing offering. But Wall Street and the media are at peak skepticism/FUD when it comes to Tesla. Elon could have announced that oh yeah, the our HW3 proprietary chip's spare trillions of compute operations available beyond FSD in the cars will be used for a cure for cancer, which will be done within 18 months of rollout. I don't think even that would move the stock.

The tangible announcement from today is that the FSD computer is going to be in all Teslas going forward. That's pretty big...

 

[Cosmacelf]

A proper simulation of human neurons need a proper hardware specific design for it. Not saying that Tsla have it down (since nobody knows what they have), but at least theirs is probably optimized to use the best type of variable for the calculation. According to neuro scientists, a neuron has a voltage component, a frequency and a phase component to their signals. Technically possible to evolve by allowing the NN to evolve its own transistor resistor level logic. But Mongo told me that's not how it was designed.

No one is simulating how human neurons work in a production NN system. Human neurons learn on the fly and adjust their weights as they learn continuously. Tesla’s (and everyone else’s production system) learns offline in huge data centers and then just the network weights are downloaded as a new net to the car.

When Elon says every car is learning, what he means is that your car might be asked to upload some data which would then be used to retrain the master Tesla NN. Unlike humans, real time learning does not happen in production NN.

 

[Remus]

Were you at the test drive?

that's irrelevant. As Karpathy said in his presentation, the important thing is to get to the tail 9s. you are not gonna experience all those 9s in a test drive.

 

[Hock1]

where’s my refreshed model S&X at?

Did anybody else catch this?: EM said something like: "we need to convince everyone that there is no sense in buying any other car than a Tesla". Then under his breath, he seemed to say "starting tomorrow". Did I hear that correctly?

 

[mblakele]

It’s a programmable CPU, with a small amount of instructions specific for neural network calculation (6 or 7 instructions, they were on a slide). The next slide was about their compiler that generates the specific set of instructions to calculate the entire network solution. (There was even a typo in the title stating ‘complier’ instead of ‘compiler’). They even mentioned some specific optimizations in their compiler. The next chip will have more instructions.

I'd call the NN part an NPU not a CPU. I'm not sure why future generations would add more instructions to the NPU. It's possible, but I'd assign higher probability to die shrinks, clock boosts, higher core counts, and more SRAM. To recap, the HW3 SoC includes:

• code security module
• ARM-based CPU cores
• GPU core(s)
• redundant Tesla NPU cores, which include SRAM
• plan safety module

More, faster, smaller cores might be helpful for those middle three modules. Future iterations might also focus on cost reduction, if this iteration turns out to be powerful enough for the job.

 

[tinm]

Today signalled another difference between Apple/Jobs and Tesla/Musk.

For Apple, the journey is the reward.

For Tesla, the destination is the reward.

The journey is gonna have a tough time of it in a robotaxi world. In fact in Elon's dreams, the robotaxis all just get in tunnels ASAP and drive 150mph+ before surfacing. And even that is just a stopgap to what Elon clearly really wants: teleportation.

In Elon's mind there is no journey. You go from destination to destination. Anyone who wants a journey, get a horse.

 

[Cosmacelf]

that's irrelevant. As Karpathy said in his presentation, the important thing is to get to the tail 9s. you are not gonna experience all those 9s in a test drive.

It isn’t irrelevant since Humans are irrational.

 

[SOULPEDL]

Just reading headlines on yahoo finance :

-
Elon Musk Boasts Tesla Built the ‘Best Chip in the World,’ Drops Nvidia

Elon Musk says Tesla driverless taxis coming next year, touts self-driving chip

Tesla promises ‘one million robo-taxis’ in 2020

The thing I think would help the share price is to see the same titles but with an external person/entity name instead of Elon/Tesla. At this point, it is just too easy to say that they are over-promising.

So ur assuming everyone missed the first part on the mission and how EVERY Elon prediction has come true? Timing aside while still leading. You are hearing competitors say one thing in public, they are freaking in private... IMO. I know I would.

 

[anthonyj]

Did anybody else catch this?: EM said something like: "we need to convince everyone that there is no sense in buying any other car than a Tesla". Then under his breath, he seemed to say "starting tomorrow". Did I hear that correctly?

I think he was saying that starting tomorrow FSD will be the main focus of marketing a Tesla car. Right now the focus is that electricity is cheaper than gas, but tomorrow they will push FSD as why someone should buy Tesla over BMW etc

 

[erha]

The bad news is, I believe, version 2 of the NN chip will have to start from scratch as an expansion of Neurons will probably mean it will have to restart learning from a random state.

Good news is, they can just hook up version 2 to version 1 and have it train the kid. Maybe that's what the training simulator they are building is for.

I’m no expert here, but the learning is only a couple of hours to a couple of days. The hard part is in labeling data. That’s part is already done.

 

[ZeApelido]

Question:

I’m not entirely sure what was meant when they mentioned unsupervised learning and "Project Dojo."

Were they referring to:

a. Some kind of bootstrapping with the training, they mentioned Self-supervision

b. Some kind of competitive network evolution, such as Generative Adversarial Networks or Genetic Algorithms

c. Unsupervised learning as might be found in biological brains, such as an Adaptive Resonance Theory based network or a behavioral conditioning circuit such as a Gated Dipole

Any thoughts or guesses?

One simple answer might be they will just upload a crap ton of data, without doing any manual labeling. And they will just assume the driver's input is the True condition.

E.g. the driver's input is the label. Even though in some cases the driver will do the wrong thing, this is such a small % of the data that they expect to get most of the way to good accuracy without worrying about that.

BTW I surmise the reason they are going to do this is they are still using the NNs mainly for perception. Which may be good enough. But I would guess the next gen chip is expected to do more end-to-end deep learning for driving - this requires much more data and compute power (both training and inference).

That's their next-level goal.

 

[sparcs]

Wall Street doesn't know how to value the robotaxi promises and is skeptical of the timeframe that it might take for Tesla to execute that strategy. Plus, now you have Panasonic limitations on the number of vehicles that Tesla can even produce.

Hopefully Q2 guidance for a profit is still on track so that we don't see as much short term pain.

 

[mblakele]

No one is simulating how human neurons work in a production NN system. Human neurons learn on the fly and adjust their weights as they learn continuously. Tesla’s (and everyone else’s production system) learns offline in huge data centers and then just the network weights are downloaded as a new net to the car.

When Elon says every car is learning, what he means is that your car might be asked to upload some data which would then be used to retrain the master Tesla NN. Unlike humans, real time learning does not happen in production NN.

That's a key point: learning takes place offline. Neural nets don't change while they're running.

That said, Lex Fridman's latest interviewee Ian Goodfellow mentioned one possible reason to change this: security. He sees it as a potential security problem that a given NN always returns the same result for the same input. A neural net that does some form of online learning might be more secure. However I think that's some years off.

 

[Doggydogworld]

The moment one other company manages to get Level 5 going, it's a price war right there.

Yep. Also, robotaxi battles are local. Tesla may have enough density to do battle in CA, but a million cars spread across a thousand cities does not equate to critical mass.

Imagine Tesla sticks to $35K+ cars and then $15K FSD capable econobox BEVs start hitting the market. How much will riders pay extra to be brought to the airport or work by Tesla whom they know make much more money, as the novelty wears off?

$15k FSD econobox BEVs are a ways off. But with fast charging and a large fleet you can dramatically reduce battery size, which will reduce upfront cost and can reduce TCO if your batteries can do enough cycles (e.g. LTO, maybe LFP).

The bigger issue is customer convenience. Would you rather ride in the back seat of a Model 3 or a minivan? Frumpy BEV econoboxes designed specifically for rideshare can offer a superior rider experience.​

 

[Anstandswauwau]

And they have ML expertise that far exceeds Tesla. And they have an actual Robotaxi service. And they're legally allowed to operate with an empty driver's seat in CA. And they've driven thousands of miles on public roads without anyone in the driver's seat. And they have enough cash to fund any possible business plan.

Where they fall short is business plan. Tesla customers have for years paid thousands of dollars for a product that didn't even exist. Google/Waymo has nothing like that. They better get their a** in gear, or all the "Ands" listed above won't matter.

Well, contrary to popular belief, Waymo is neither in the "car selling"- nor in the "ride hailing"-business.

Instead Waymo is supposed to become to cars and Robotaxis what Android is to smartphones – a licensable piece of tech. And they very well be able to pull this off, again!

Sure, you can laugh at Waymo's current capabilities and required hardware, yet, they are still miles ahead of what most traditional automakers can do today. If said automakers fall behind even further, they might one day have no other option than to license whatever tech is available on the market.