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

[phantasms]

@Artful Dodger
Has your conviction on the split announcement this month changed one way or the other?

In 2020, initially the Annual shareholder meeting was scheduled for July-07, later moved due to Covid.
Anyone has any idea on when the notification was published on the initial schedule in July? How many days prior to July-07-2020 was it published?

Original July 7th shareholder meeting letter is below and I believe was published on May 28th

tsla-def14a_20200707.htm

www.sec.gov

July 10th - New shareholder meeting/battery day date announced.
Aug 11th - Split announcement

Personally I haven't seen evidence of an impending share split but I have been closing out my covered calls faster than usual lately. lol

 

[Blue horseshoe]

At 22:20 Elon talks about one of his many mistakes. He uses the model 3 production hell as an example and specifically discusses a production chokepoint. There was a fiberglass mat on top of the batter pack that was slowing production so Elon first tries to better automate the process. Then he tried to accelerate. Then he tried to optimize.

Finally he asks the battery safety team what are these mats for? They say for noise and vibration. But Elon says you're the battery dept so he goes to noise, vibration, and harshness (NVH) team and they tell him it's for fire safety. At this point Elon questions if he's in some sort of simulation and he's trapped in a dilbert cartoon.

Getting a lot more information on what Elon is talking about when he says production is difficult. This video is definitely worth watching.

[Edit] at 31:45 Elon mentions they use model 3 motors to move the grid fins. I thought that was cool.

 

[Sofie]

FWIW, Florida Tesla Enthusiasts had a meetup at Babcock Ranch some time ago:

Tesla Owners Florida - Florida Tesla Enthusiasts tour Babcock Ranch, the city of the future (Public)

Just another reason I love my Florida

 

[JRD1]

So this chip that was just shown was reported awhile back in 2020. Reports are calling this HW 4.0, but not from the packaging we just saw. No way Tesla is putting a 25 chip cluster into a single FSD computer for a car.

Tesla Partners Up With Broadcom To Equip FSD System With 7nm HPC Chip Packaged Without Substrate​

"Apart from being built on the 7nm node, the HW 4.0 chip will also utilize TSMC's Integrated FanOut (InFo) packaging technology that is designed to reduce overall package surface area and provide lower thermal resistance. Additionally, the processor will also be packaged using TSMC's latest System on Wafer (SoW) technology that removes the need to place a substrate and a printed circuit board (PCB) in the entire process. Via SoW, the HW 4.0 will be placed directly on the heat dissipation module, and today's report suggests that Indium Corporation will provide the thermal conductivity materials.

Initial production slated to begin later this year will churn out 2,000 silicon wafers with the chip, with each wafer having a 12-inch diameter and consisting of 25 HW 4.0 chips."

Tesla To Equip Vehicles With 7nm HPC Processor In 2022 Built With Broadcom & TSMC

Broadcom and Tesla have developed a new HPC 7nm chip dubbed HW 4.0 for Tesla's vehicles reports from China suggest. Take a look here!

wccftech.com

I think that means they will get 25 chips out of the wafer - not that 25 chips go into 1 FSD computer

 

[MC3OZ]

I think that means they will get 25 chips out of the wafer - not that 25 chips go into 1 FSD computer

We could read this as the initial production run is 2,000 x 25 = 50,000 chips....

Initial production slated to begin later this year will churn out 2,000 silicon wafers with the chip, with each wafer having a 12-inch diameter and consisting of 25 HW 4.0 chips.

From memory, at Autonomy Day HW 4.0 was listed as being 3X faster than HW 3.0.

I'm more interested in price and power consumption, because HW 4,0 might be cheaper and use less electricity, and that may be important for future compact models.

People are assuming HW 3.0 can't do the job, or that there will be a simple upgrade path from HW 3.0 to HW 4,0.
The most efficient solution for Tesla is not upgrading cars with HW 3,0.....
In fact avoiding or delaying the need to upgrade is worth expending considerable NN development resources.

At AI day we may see specialist hardware for Dojo, and we may get an update on HW 4.0

 

[Knightshade]

People are assuming HW 3.0 can't do the job, or that there will be a simple upgrade path from HW 3.0 to HW 4,0.
The most efficient solution for Tesla is not upgrading cars with HW 3,0.....

Green has found (since mid 2020)- and IIRC James Douma confirmed, that Tesla had run out of compute on node A in HW3, and had to spill stuff over (increasingly so since then) to node B.

Which is fine for L2 ADAS since you don't need full node redundancy- you have a human as your backup.

But it's a non-starter for L4+ where a human can't be required to even be in the vehicle.

Thus (barring some future complete re-write that vastly cuts down what needs to run, by a lot- which seems the opposite of the direction they've been going in running more, and more complex and larger, NNs over time) HW3 is insufficient for L4+ FSD.

Thus they're going to need to upgrade everyone (who fully purchased FSD anyway) to HW4.

I find the fact HW upgrades are NOT included for subscription customers to be significant here.... (the # of HW2.x folks impacted by this is relatively small... the # of HW3 folks that'd be impacted is much, much larger)


That said- they can certainly delay such upgrades until the point it's matters---which is either when they feel confident of having real L4, or if they flat out run out of compute for even the more advanced L2 stuff on HW3.... they did that with the free 2.x->3 upgrades too, delaying the upgrades until it mattered at all.

 

[dl003]

At 22:20 Elon talks about one of his many mistakes. He uses the model 3 production hell as an example and specifically discusses a production chokepoint. There was a fiberglass mat on top of the batter pack that was slowing production so Elon first tries to better automate the process. Then he tried to accelerate. Then he tried to optimize.

Finally he asks the battery safety team what are these mats for? They say for noise and vibration. But Elon says you're the battery dept so he goes to noise, vibration, and harshness (NVH) team and they tell him it's for fire safety. At this point Elon questions if he's in some sort of simulation and he's trapped in a dilbert cartoon.

Getting a lot more information on what Elon is talking about when he says production is difficult. This video is definitely worth watching.

[Edit] at 31:45 Elon mentions they use model 3 motors to move the grid fins. I thought that was cool.

This is the best kind of advertisement for TSLA. I dare anyone to find another CEO who can be such a total embodiment of his company like Elon.

 

[Singuy]

I think that means they will get 25 chips out of the wafer - not that 25 chips go into 1 FSD computer

This rumor poorly translated story is suppose to give us a sense of the dojo package we saw today which has 25 chips interconnected as one full module. The original source did not mention anything about hw4 fsd computer.

台積電7奈米 獲博通HPC訂單 - 財經要聞

晶圓代工龍頭台積電再傳接單捷報。業界傳出，全球IC設計龍頭博通（Broadcom）與電動車大廠特斯拉（Tesla）共同開發的新款高效能運算（HPC）晶片，將以台積電7奈米先進製程投片，並採用台積電整合型扇出（InFO）系統

www.chinatimes.com

 

[ls7corvete]

I believe that the more claimed losses, the greater is the total judgment amount (and lawyers' fee), if the verdict favors the litigants. A lawyer here may know better.

The ambulance chasing law firm is making it rather difficult for potential litigants to opt out. Simply returning the postcard, emailing an opt out, or phoning the firm is insufficient. I phoned, and was told to visit the webpage linked below, and send them a postal letter answering all of the detailed questions in section #9:

Tesla Securities Litigation

I think it's time for another class action suit for individuals that feel opting out is a burden. All tsla share holders are included, unless you fax in pages 9-14 filled out in full and stamped by an officiary.

 

[MC3OZ]

Green has found (since mid 2020)- and IIRC James Douma confirmed, that Tesla had run out of compute on node A in HW3, and had to spill stuff over (increasingly so since then) to node B.

Are you sure?

My impression was code to gather data on predefined triggers, was running in Node B.

Tesla also did an acquihire of a group very talented in optimising NNs and running them on the bare minimum of hardware...

A fairly normal IT / engineering process is, getting it working first, then optimize it...

I agree it might not be possible to avoid hardware upgrades, but I still think it is worth trying...

 

[t153]

I don't think this has been determined?

Will Tesla converting 10.000 stalls in the US get the same funding as building a new stall from scratch? Somehow I doubt it.

Just making up numbers here. $7,5 billion for 100.000 stalls = $75,000 per stall.

So with 10,000 Tesla stalls already existing in the US, with an adapter and some software as only cost for converting them, Tesla could maybe make $750 million almost overnight if they qualify.

If those qualify I imagine existing stalls that other companies have will cry foul and demand some retroactive money to despite this being slightly different. Not saying they should, but they would try. There could also be plenty of media backlash from the government 'handing' this money to Tesla.

On the other hand, if already existing Tesla chargers don't qualify why should Tesla open up access to those? If funding is only for truly 'new' chargers shouldn't Tesla put all new ones in a Tesla Supercharger Network 2 which is open for everyone, while keeping the existing ones in Tesla Supercharger Network 1 keeping those as Tesla only.

Yea, I personally don’t think this would apply to existing superchargers, which is why I was saying that it’s unknown whether the grant money would be retroactive or not. To me it looks like it would only go towards building out new stations. Can’t see the RFP being awarded only to Tesla for exactly the same reasons you’re outlined. That’s why I was saying that it would be great for Tesla to take the money and expand their network in areas where there may not be enough demand for them to do this with their own $, but it could be a chicken and an egg thing; if the chargers are deployed then perhaps the demand for their cars will go up in those areas too. If they can do it “for free” then it will only serve to improve their network with no real downside. Just as an example, imagine tmobile or any other carrier got free govt money to install their towers in some remote areas - it would help them and their customs, but even other carriers who have roaming agreements with tmo.

 

[ls7corvete]

Yea, I personally don’t think this would apply to existing superchargers, which is why I was saying that it’s unknown whether the grant money would be retroactive or not. To me it looks like it would only go towards building out new stations. Can’t see the RFP being awarded only to Tesla for exactly the same reasons you’re outlined. That’s why I was saying that it would be great for Tesla to take the money and expand their network in areas where there may not be enough demand for them to do this with their own $, but it could be a chicken and an egg thing; if the chargers are deployed then perhaps the demand for their cars will go up in those areas too. If they can do it “for free” then it will only serve to improve their network with no real downside. Just as an example, imagine tmobile or any other carrier got free govt money to install their towers in some remote areas - it would help them and their customs, but even other carriers who have roaming agreements with tmo.

I could certainly imagine an agreement to supply existing chargers with adapters, as well as build additional chargers during the award period. With and up front payout and agreement.

 

[Tim S]

@Artful Dodger
Has your conviction on the split announcement this month changed one way or the other?

In 2020, initially the Annual shareholder meeting was scheduled for July-07, later moved due to Covid.
Anyone has any idea on when the notification was published on the initial schedule in July? How many days prior to July-07-2020 was it published?

The 2020 Annual Shareholder Meeting was originally announced May 28th, 40 days before July 7th.

Tesla will hold its 2020 Annual Shareholder Meeting on July 7

Tesla (NASDAQ: TSLA) will hold its 2020 Annual Shareholder Meeting on Tuesday, July 7, 2020, at 2:30 p.m. Pacific Time, at the Computer History Museum in Mountain View, CA. In a Securities and Exchange Commission proxy statement filing on May 28, Tesla outlined seven key agenda items to be voted...

www.teslarati.com

 

[ls7corvete]

Green has found (since mid 2020)- and IIRC James Douma confirmed, that Tesla had run out of compute on node A in HW3, and had to spill stuff over (increasingly so since then) to node B.

Which is fine for L2 ADAS since you don't need full node redundancy- you have a human as your backup.

But it's a non-starter for L4+ where a human can't be required to even be in the vehicle.

Thus (barring some future complete re-write that vastly cuts down what needs to run, by a lot- which seems the opposite of the direction they've been going in running more, and more complex and larger, NNs over time) HW3 is insufficient for L4+ FSD.

Thus they're going to need to upgrade everyone (who fully purchased FSD anyway) to HW4.

I find the fact HW upgrades are NOT included for subscription customers to be significant here.... (the # of HW2.x folks impacted by this is relatively small... the # of HW3 folks that'd be impacted is much, much larger)


That said- they can certainly delay such upgrades until the point it's matters---which is either when they feel confident of having real L4, or if they flat out run out of compute for even the more advanced L2 stuff on HW3.... they did that with the free 2.x->3 upgrades too, delaying the upgrades until it mattered at all.

What are the odds that hw4 is smart enough to "learn and teach" a neural net for hw3 to perform at least L2/L3?

 

[Chunky Jr.]

more price changes...

https://twitter.com/x/status/1422761696288862208

 

[Ohmster]

Green has found (since mid 2020)- and IIRC James Douma confirmed, that Tesla had run out of compute on node A in HW3, and had to spill stuff over (increasingly so since then) to node B.

Which is fine for L2 ADAS since you don't need full node redundancy- you have a human as your backup.

But it's a non-starter for L4+ where a human can't be required to even be in the vehicle.

Thus (barring some future complete re-write that vastly cuts down what needs to run, by a lot- which seems the opposite of the direction they've been going in running more, and more complex and larger, NNs over time) HW3 is insufficient for L4+ FSD.

Thus they're going to need to upgrade everyone (who fully purchased FSD anyway) to HW4.

I find the fact HW upgrades are NOT included for subscription customers to be significant here.... (the # of HW2.x folks impacted by this is relatively small... the # of HW3 folks that'd be impacted is much, much larger)


That said- they can certainly delay such upgrades until the point it's matters---which is either when they feel confident of having real L4, or if they flat out run out of compute for even the more advanced L2 stuff on HW3.... they did that with the free 2.x->3 upgrades too, delaying the upgrades until it mattered at all.

Would removal of radar have a substantial reduction of load on Node A?

 

[Chunky Jr.]

Interesting info from Rob Maurer about Tesla AI day invitation...

Tesla Begins Sending Out AI Day Invitations

Tesla has started sending out invitations for their upcoming "AI Day" event, complete with an interesting teaser...

www.thestreet.com

One sentence sticks out as particularly noteworthy... "They’ll also get an inside-look at what’s next for AI at Tesla beyond our vehicle fleet."

We may get a glimpse of what Elon meant when he said "long term, people will think of Tesla as much as an AI robotics company as we are a car company, or an energy company"

 

[UnknownSoldier]

Green has found (since mid 2020)- and IIRC James Douma confirmed, that Tesla had run out of compute on node A in HW3, and had to spill stuff over (increasingly so since then) to node B.

Which is fine for L2 ADAS since you don't need full node redundancy- you have a human as your backup.

But it's a non-starter for L4+ where a human can't be required to even be in the vehicle.

Thus (barring some future complete re-write that vastly cuts down what needs to run, by a lot- which seems the opposite of the direction they've been going in running more, and more complex and larger, NNs over time) HW3 is insufficient for L4+ FSD.

Thus they're going to need to upgrade everyone (who fully purchased FSD anyway) to HW4.

I find the fact HW upgrades are NOT included for subscription customers to be significant here.... (the # of HW2.x folks impacted by this is relatively small... the # of HW3 folks that'd be impacted is much, much larger)


That said- they can certainly delay such upgrades until the point it's matters---which is either when they feel confident of having real L4, or if they flat out run out of compute for even the more advanced L2 stuff on HW3.... they did that with the free 2.x->3 upgrades too, delaying the upgrades until it mattered at all.

I wonder how much power they will spec for HW4 to try and prevent this from happening again.

I'm increasingly glad I paid for FSD during that last fire sale last year. I was upgraded from HW2.5 to HW3, and I'm covered for a future upgrade to HW4.

Hopefully they have some kind of general plan in place now based on their expected growth of compute power required relative to hardware they want to release to replace HW3 with HW4.

 

[mrdoubleb]

Would be market moving if true:

It appears that Tesla China is making good progress in its development of the highly-anticipated $25k compact electric car. If recent rumors from China are any indication, it appears that the upcoming affordable EV is still planned for trial production for the end of 2021.

The update about Tesla China’s newest electric vehicle was shared by noted automotive segment leaker 不是郑小康, who has proven to be accurate in the past. According to the industry leaker, the prototype for Tesla’s $25K car has already been completed, and most suppliers for the vehicle’s production have already been lined up.

Tesla $25K compact electric car prototype has been completed: rumor

It appears that Tesla China is making good progress in its development of the highly-anticipated $25k compact electric car. If recent rumors from China are any indication, it appears that the upcoming affordable EV is still planned for trial production for the end of 2021. The update about...

www.teslarati.com

I have a hard time believing this rumor. Shanghai Phase 1, Part B has just been completed and this was built for Model Y. Phase 2 is being planned and was thought to be for Model 2/Q but construction has not started yet. So unless Phase 1/B has half the facility empty and somehow they have managed to keep that secret, where would the new car be built at the end of this year?

I can imagine test mules, alpha builds, but production? Nope.

 

[t153]

Just started - Everyday Astronaut (Tim Dodd ) is doing the best journalistic job for SpaceX and Elon - (wish someone of this caliber would coverTesla) note there are only Tesla cars parked outside .. he's reiterating how the design part is like negligible compared to designing the manufacturing of these rockets.
Looking forward to similarly entertaining and instructive videos on Tesla (Gigafactories, Plaid ..etc) - Tesla shorts, FUDsters and the frivolous lawsuits.

Starbase Factory Tour with Elon Musk [Part 1]

View attachment 691706

Random pearls
"All designs are wrong, its just a matter of how wrong" - Elon.
"If you're digging your grave, don't dig it faster. Stop digging the grave."
"Everyone's wrong, no matter who you are, everyone's wrong some of the time"

Cliff notes there: https://everydayastronaut.com/starbase-tour-and-interview-with-elon-musk/

" Interview Part 1

On July 30, 2021 Tim Dodd received a tour of Starbase with SpaceX CEO, CTO, and founder Elon Musk. This article includes key takeaways from the first 50 minutes of the over 2 hour long interview.

Manufacturing:

Musk once again emphasized that manufacturing is underrated and design is overrated; developing a production system is 10-100 times harder than designing the product, which according to Musk is especially true with Raptor. As volume production increases the amount of effort that goes into design rounds down to 0.

For an example, Musk stated that designing a closed cycle engine is easy, but the extremely hard part is the cost per tonne of thrust under $1,000. With each Raptor producing 230 tonnes of thrust this means that each engine must cost less than $230,000 to produce. Musk continued stating that the cost per tonne to orbit and the cost per tonne to the surface of Mars is several orders of magnitude too high on current launch vehicles. This is why it is so important to move as much mass and complexity as possible to the ground.

Super Heavy Design:

Musk noted that Super Heavy’s dry mass should be under 200 tonnes, although the dry mass is a moving target. The engines, including mounting mass, are 2 tonnes; the fuel tank and the liquid oxygen (LOx) ... "

Musk’s Engineering Philosophy:​

Musk overviewed his five step engineering process, which must be completed in order:

1. Make the requirements less dumb. The requirements are definitely dumb; it does not matter who gave them to you. He notes that it’s particularly dangerous if someone who is smart gives them the requirements, as one may not question the requirements enough. “Everyone’s wrong. No matter who you are, everyone is wrong some of the time.” He further notes that “all designs are wrong, it’s just a matter of how wrong.”
2. Try very hard to delete the part or process. If parts are not being added back into the design at least 10% of the time, not enough parts are being deleted. Musk noted that the bias tends to be very strongly toward “let’s add this part or process step in case we need it.” Additionally, each required part and process must come from a name, not a department, as a department cannot be asked why a requirement exists, but a person can.
3. Simplify and optimize the design. This is step three as the most common error of a smart engineer is to optimize something that should not exist.
4. Accelerate cycle time. Musk states “you’re moving too slowly, go faster! But don’t go faster until you’ve worked on the other three things first.”
5. Automate. An important part of this is to remove in-process testing after the problems have been diagnosed; if a product is reaching the end of a production line with a high acceptance rate, there is no need for in-process testing. ... "

Which reminds me of Elon's comments as to how the car parts design followed organizational designs, which introduced unnecessary parts.

Some fun details on the early Tesla battery packs, mentioning how they were initially cured the wrong way .. but also how some guy in Pakistan was using his Tesla as a boat essentially, during a flood, moving around using his spinning wheels for propulsion (anyone seen that video?)

It was Kazakstan

Here's China

https://twitter.com/x/status/1417449054011678721