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

[MC3OZ]

What happens when a passenger walks away and leaves a door or the trunk open.

They should get a message on their phone via the app telling them to close it...

If they don't do that, then ban them from booking Robotaxis in future.....

Next step is the car needs to plead with nearby pedestrians to close the door

If none of that works, it is a difficult problem, but out of the hypothetical 1000, cases per day the solution above probably works for at least 900, for the other 100 a Robotaxi employee or Optimus needs to travel to the scene and fix the problem.

If the car can't leave a parking spot because a door is open, there is a good chance someone else wants that spot and will close the door. If no one wants the spot the the car waits for Optimus to jog over, preferably in 1980s jogging gear with a headband,.

 

[2daMoon]

So, is Dojo dead or what? By the time it's done will be actually still cheaper/better than an off the shelf NVDA system?

The way I understand it, Dojo is apples to Nvidia's oranges. Both systems have unique properties and both are needed.

Someone more articulate on the subject might give us the Cliff's Notes again?

 

[Knightshade]

Those are foreseeable and solvable problems.
What happens when a passenger walks away and leaves a door or the trunk open.
I ‘m sure a robotaxi will be able to close a door or trunk automatically. Today’s model 3/Y can’t.

Current 3 and Y both have power-closing trunks... so they can today if Tesla wishes to let the RT software do that.

There's potential pinch liability in play, but nothing technical prevents it.

 

[jim0266]

The way I understand it, Dojo is apples to Nvidia's oranges. Both systems have unique properties and both are needed.

Someone more articulate on the subject might give us the Cliff's Notes again?

"In early October, the U.S. government rolled out extensive new restrictions on China’s access to advanced semiconductors and the equipment used to make them. The restrictions require a hard-to-get license for the sale of advanced semiconductors to entities within China, largely depriving the country of the computing power it needs to train artificial intelligence (AI) at scale."

Can Tesla build DOJO in China to somehow bypass these restrictions? Otherwise where will Tesla get the hardware needed to train FSD in China?

 

[Knightshade]

"In early October, the U.S. government rolled out extensive new restrictions on China’s access to advanced semiconductors and the equipment used to make them. The restrictions require a hard-to-get license for the sale of advanced semiconductors to entities within China, largely depriving the country of the computing power it needs to train artificial intelligence (AI) at scale."

Can Tesla build DOJO in China to somehow bypass these restrictions? Otherwise where will Tesla get the hardware needed to train FSD in China?

Tesla doesn't build chips anywhere. They pay folks like TSMC and Samsung who actually own chip factories to do that.

To answer your question though -sure if they want to invest like 20 billion dollars in a state of the art fab and wait a couple years to get production up to speed, then constantly lose massive amounts of $ annually because fabs are insanely expensive to operate and Teslas own chip needs wouldn't come close to making the investment worthwhile.... (companies who use MASSIVELY more chips annually than Tesla, and have more cash on hand- Nvidia and Apple for example- don't even build their own fabs- it's an insane business).

Doesn't seem a great choice.

 

[Singuy]

The way I understand it, Dojo is apples to Nvidia's oranges. Both systems have unique properties and both are needed.

Someone more articulate on the subject might give us the Cliff's Notes again?

Yes, both systems are very different. Dojo with the right optimization can train gigantic models beyond the speed of Nvidia. Nvidia uses a monolithic chip per pcb and communicating to another chip requires their pretty fast nvlink. However just like when you run out of memory in windows and much slower virtual memory is used which is much slower than real memory, there are models in the future that are so big...like say AGI of optimius, that DOJO will surpass nvidia's performance or it might be the only system that can train it end to end. With nvidia, you have to break certain aspect of training into pieces because bandwidth bottleneck appears every time it needs multiple gpu cores, or worst needing to jump to a different rack using switches.

Dojo was built from the ground up to act as if all the cores are within one gigantic cluster. This is what I think when Musk said it's a gamble worth taking.

Also worth noting that Nvidia is working toward something like this as well, but they are in the business of selling chips. So they have planned obsolescence fully knowing models will become bigger and bigger. They will give people just enough to want them to buy the new stuff while Tesla doesn't do such things.

 

[Singuy]

Tesla doesn't build chips anywhere. They pay folks like TSMC and Samsung who actually own chip factories to do that.

To answer your question though -sure if they want to invest like 20 billion dollars in a state of the art fab and wait a couple years to get production up to speed, then constantly lose massive amounts of $ annually because fabs are insanely expensive to operate and Teslas own chip needs wouldn't come close to making the investment worthwhile.... (companies who use MASSIVELY more chips annually than Tesla, and have more cash on hand- Nvidia and Apple for example- don't even build their own fabs- it's an insane business).

Doesn't seem a great choice.

I think there's a possibility that Dojo can be imported into China for Tesla use only as it's the only US owned enterprise. High chance they can get away with it. Also technically they are not selling chips to China as Dojo is not being sold to anyone except it's hardware developed inhouse. I think this will give Tesla a huge advantage over peers in China when it comes to end to end training(after they steal Tesla's homework) since no one is allowed to buy Nvidia hardware.

 

[UkNorthampton]

They should get a message on their phone via the app telling them to close it...

If they don't do that, then ban them from booking Robotaxis in future.....

Next step is the car needs to plead with nearby pedestrians to close the door

If none of that works, it is a difficult problem, but out of the hypothetical 1000, cases per day the solution above probably works for at least 900, for the other 100 a Robotaxi employee or Optimus needs to travel to the scene and fix the problem.

If the car can't leave a parking spot because a door is open, there is a good chance someone else wants that spot and will close the door. If no one wants the spot the the car waits for Optimus to jog over, preferably in 1980s jogging gear with a headband,.

For several years, the Starship (yes, that's the company name) delivery robots near me have been asking people to press the buttons at pedestrian crossings or asking for a push if they get stuck.

 

[insaneoctane]

Waiting to buy back in at $160 or $150 after P&D and earnings later in April.

I haven't sold anything, but I will be buying more at $16x and $15x if they return. I know it's unpopular in this thread to want lower SP, but the stock has been depressed for so long, I'd personally enjoy briefly revisiting those prices one more time for some final buys before the inevitable ascension....but I understand not everyone feels the same.

 

[Xepa777]

I haven't sold anything, but I will be buying more at $16x and $15x if they return. I know it's unpopular in this thread to want lower SP, but the stock has been depressed for so long, I'd personally enjoy briefly revisiting those prices one more time for some final buys before the inevitable ascension....but I understand not everyone feels the same.

I need tesla to stay depressed until the crypto boom hits full swing so I can sell those positions and plow back into Tesla lol. Sold my entire Tesla Ira postion the day after Cybertruck pricing was revealed and saw how unserious it was lol. Not even worth arguing with the religious cultists here. Made (and still am) making great money on that repositioning. Tesla is in for a rough year and no one cares about FSD or Optimus until it’s actually working (so like 2028). The only catalyst is the announcement or leak of the model 2 which is optimistically at the end of this year. Even then, Tesla has damaged its reputation on scaling for Wall Street projections with how it’s performed the past two years, the rise of cheap byd mass market cars already, and the advent of sodium battery based cheap cars.

Only coming out of my hole so I can reference this post in a few years like how I can always reference to my past posts on FSD.

 

[raffiniert]

who will charge the cars between fares?

Induction.

Who will clean the cars?

Could be done by non-Tesla-employees that just volunteer and get paid to clean a car in credit for future drives or money. Works for the e-scooter lenders here in Europe as well (they collect the scooters when the app tells them to, carry them up into their flat, connect it to their own outlets and return them to the street once charged for a bit of money).

 

[JusRelax]

There are several confirmations on X that FSD Beta 12.3.1 is now being rolled out to Canada:

https://twitter.com/x/status/1772880032454848772

https://twitter.com/x/status/1772879382505230693

https://twitter.com/x/status/1772877944697848121

 

[Optimeer]

„Mr.Tesla“ on X has 12.3.2 and says it is another hughe improvement. Further Auto Wipers are fixed now and the Auto Park functionality was added. 12.3.2 could go wide since it was merged on the 2024 branch according to Chuck Cook.

The rate of improvement was very high lately and will probably be for at least a month. I don‘t think this will be unnoticed for very long.

From my experience as a shareholder since 2018, during unexpected good news, the stock prize goes down for at least some days. Then, after the big dogs have buyed in, the stock price starts to rise.

Soon we will get the P&D numbers, which likely will be spun negative by the media. Also for the quarterly earnings, there will be points which can be spun negatively. So we could see a shareprice of 15x again.

Also BYD was down 6 % because of missed earnings expectations, a good excuse to send TSLA down today.

But.. like we say in Switzerland.. There is light at the end of the tunnel!

 

[raffiniert]

So we could see a shareprice of 15x again.

I read that as "15 times" and we will get there!
Cheers from NW-Switzerland!

 

[Optimeer]

The situation now feels like: Tesla did exaclty know what they wanted to do for a year, but couldn‘t, due to lacking training capacity. Now the training capacity is online and all the queued tasks get done.

 

[jeewee3000]

They should get a message on their phone via the app telling them to close it...

If they don't do that, then ban them from booking Robotaxis in future.....

Next step is the car needs to plead with nearby pedestrians to close the door

Much easier solution: the robotaxi keeps the meter running until the fare closes all doors/trunk/frunk at destination. Simple and effective.

 

[heltok]

So, is Dojo dead or what? By the time it's done will be actually still cheaper/better than an off the shelf NVDA system?

https://twitter.com/x/status/1772874074911359073

Seems like it was a large amount of compute coming online from various sources. Imo this is what Tesla are doing well, being agile. When everyone struggles to get nVidia chips, they have already diversified to alternate suppliers, doing it inhouse and getting things up and running fast. And they are not afraid to make the billion dollar investment into compute. Meanwhile where is Toyota, VW etc with this? Have we heard anything?

 

[-=buzz=-]

So, is Dojo dead or what? By the time it's done will be actually still cheaper/better than an off the shelf NVDA system?

I don't think so:
1)
Nvidia's latest GPU which isn't sold yet has a system level architecture that's much closer to the Dojo architecture than before, so Tesla seems to be on the right path from that perspective.
2)
Dojo is still on silicon version 1, as far as we know. The next iteration promises bit gains due to fixes and a general uplift in compute power.
3)
Supply. They are already buying every GPU they can. Producing their own increases the available resources.
4)
Nvidia has a Gross margin of ~75%
Even if Dojo is only half the speed, it's about break even on compute/$

 

[2Pearls]

Has the long-awaited rotation finally begun? (speculation only - and perhaps some wishful thinking too)

View attachment 1032265

My wishful thinking says the rotation has begun. But holding TSLA since 2013 and wanting to HODL says:

View attachment 1032266

I need tesla to stay depressed until the crypto boom hits full swing so I can sell those positions and plow back into Tesla lol. Sold my entire Tesla Ira postion the day after Cybertruck pricing was revealed and saw how unserious it was lol. Not even worth arguing with the religious cultists here. Made (and still am) making great money on that repositioning. Tesla is in for a rough year and no one cares about FSD or Optimus until it’s actually working (so like 2028). The only catalyst is the announcement or leak of the model 2 which is optimistically at the end of this year. Even then, Tesla has damaged its reputation on scaling for Wall Street projections with how it’s performed the past two years, the rise of cheap byd mass market cars already, and the advent of sodium battery based cheap cars.

Only coming out of my hole so I can reference this post in a few years like how I can always reference to my past posts on FSD.

I mostly agree with you and @insaneoctane and @Mengy . I have been a major TSLA investor since 2013 and I have held through all the ups and downs including the disaster from October 2022 till January 2023. All stocks go up and down, even the best ones that have been staples within people’s portfolios for their lifetime...TSLA is a little more volatile than most. But, if people think that Q1 2024 P&D (Which does not look good based on 99% of the present reporting) and subsequent earnings later in April are fully baked in, and the stock will stay in this range of $170-$185, and not take a 5%-10% dip, even if it is short lived, then they are living in a bubble with blinders on.
Do I want a huge dip, no, but I live in a reality based world.

 

[Usain]

Back of the envelope:

There are about 3 trillion vehicle miles travelled annually in the US alone.

Let’s say (generously) a robotaxi does 10x the ~15K miles the average American drives annually or 150,000 miles.

It would take 20 million robotaxis to cover those miles.

One could make a case that not only will the "old Teslas" have an opportunity to drive some of those miles, they could be worn out driving them.

Also, just as there are managerial services that take care of Airbnb’s for property owners, there will likely be third parties on whom Tesla owners can rely to do the legwork.

My intent is not to bandy hypotheticals with you, rather I’m pointing out that there will be a lot nuance as well as surprises in the evolution of the robotaxi business.

There is for some of course value in figuring this business out before others both for those looking to profit and those that will be disrupted.

Totally agree that there are lots of ways this could play out. We're all just spitballing here.