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

[GOVA]

OT weekend confession.

Just picked up my 1st baby - the MY, my first TESLA. I know, I am kind of late to the party especially in this crowd here.

Anyway, the car is really something else. I did expect a lot, but today I was born again. This thing is absolutely stunning in many aspects of being a 'car'

'Regular' MY, LR, with Gemini wheels.

Wife, kids and I are still trying to pick up our jaws from the floor. Mind blown.
Lord, now I know for sure that legacy ICE is fracked bad.

Last thing this evening, we pulled by Ice Cream Joint. 10-y old kids started yelling TEEESLA!!!
Which concluded the day.

Had to floss this out here. Hope you understand.

Love Tesla, Love Elon. Peace.

 

[M|S|M MYP]

Def demand problem ... car's just sitting in parking lots:

 

[ZeApelido]

Restating above, reaching 1,000 per day means annualized ~ 360k a year just of Model Y

https://twitter.com/x/status/1428917964233797639

 

[uujjj2]

I wish they had shown the robot next year… after the CT, semi, Roadster and 4680s were rolling out of Austin and Berlin.
The FSD and dojo stuff should have been enough for recruiting. The robot seems like a bit of distraction.

I kinda agree. The robot's really cool, but we know Elon and we can surmise that robot's pretty far away from shipping. On the other hand it does show Tesla has a lot of really long-term projects going, not just near-term self-driving car stuff.

 

[uujjj2]

Once Upon a Time: GM had a vision of the future. We're still waiting.

GM Headquarters reception 1965...

I don't have any behind-the-scenes knowledge, but just judging by the recruiting e-mails I've gotten I'm pretty sure GM also has an in-house self-driving computer under development.

 

[Chenkers]

Restating above, reaching 1,000 per week means annualized ~ 500k a year just of Model Y

https://twitter.com/x/status/1428917964233797639

Actually it's 1,000 Model Y per day. Assuming say 340 days production that's 340,000 per year. Although actual run rate would be less than peak production rate.

Here's the description from the video:
After four days of overhaul at Tesla's Shanghai plant last week, the production capacity of Phase II Modle Y has reached 1,000 units per day. This daily capacity has exceeded the number of Modle 3’s 800 units per day. It can be said that the Y-type production has entered High-yield stage. As previously planned by Tesla, the main production of Q3 is for export.

I expect we'll see front and rear casting Model Y's out of Shanghai once the newly installed casting machines are operational. We should also get a structural battery pack once LG or CATL start producing them. Based on recent rumours the latter could come from the soon to be built CATL Shanghai factory. This may be the longer range Model Y Tesla recently registered with Chinese authorities.

 

[SPadival]

Def demand problem ... car's just sitting in parking lots:

No sign of chip shortage there .. I hope demand is low in so that I can get mine in

In other news, does anyone here remember Chris ? I can’t recall what was his anonymous user id here.

EDIT: I found him!! maybe against forum rules to link the two.. he’s posted here recently

https://twitter.com/x/status/1428840022967529473

 

[22522]

If prosthetics don’t need it, the bot won’t.

This is great! Thank you. It reminded me of simpler leg replacements used in India. The foot was passive but elastic. I do not think they were doing natural frequency matching with the other leg yet, if I recall the conversation correctly. Jaipur foot…

Homepage New - Jaipurfoot

BMVSS: A charity organization since 1975, Jaipur Foot has provided 2M free limbs and devices to the disabled across 23 branches in major Indian cities.

www.jaipurfoot.org

The reason this is in the investment forum is to show that the Tesla robot can exist quickly if a decision is made for speed and cost.

 

[Nolimits]

OT weekend confession.

Just picked up my 1st baby - the MY, my first TESLA. I know, I am kind of late to the party especially in this crowd here.

Anyway, the car is really something else. I did expect a lot, but today I was born again. This thing is absolutely stunning in many aspects of being a 'car'

'Regular' MY, LR, with Gemini wheels.

Wife, kids and I are still trying to pick up our jaws from the floor. Mind blown.
Lord, now I know for sure that legacy ICE is fracked bad.

Last thing this evening, we pulled by Ice Cream Joint. 10-y old kids started yelling TEEESLA!!!
Which concluded the day.

Had to floss this out here. Hope you understand.

Love Tesla, Love Elon. Peace.

Congrats!
The conviction builds quickly once you drive it and experience the convenient ownership.
And again sells itself…. Bought my M3 years ago and now MY and CT on order…

The legacy manufacturers r history.

 

[CLK350]

And the beat goes on .. The cars keep going faster all the time ..

Steve Jurvetson (early Tesla Investor, board member) on Twitter
"Tesla now holds the mantle of Moore’s Law. Just as NVIDIA took leadership from Intel a decade ago.
While the substrate has shifted several times, humanity's capacity to compute has compounded for 122 years as if on rails.
It's surreal.

From Steve Jurveston's Flickr

122 Years of Moore's Law + Tesla AI Update

Tesla now holds the mantle of Moore’s Law, with the D1 chip introduced last night for the DOJO supercomputer (video, news summary).

This should not be a surprise, as Intel ceded leadership to NVIDIA a decade ago, and further handoffs were inevitable. The computational frontier has shifted across many technology substrates over the past 120 years, most recently from the CPU to the GPU to ASICs optimized for neural networks (the majority of new compute cycles).

Of all of the depictions of Moore’s Law, this is the one I find to be most useful, as it captures what customers actually value — computation per $ spent — on a log scale, so a straight line is an exponential.

Humanity’s capacity to compute has compounded for as long as we can measure it, exogenous to the economy, and starting long before Intel co-founder Gordon Moore noticed a refraction of the longer-term trend in the belly of the fledgling semiconductor industry in 1965.

In the modern era of accelerating change, it is hard to find even five-year trends with any predictive value, let alone trends that span the centuries. I would go further and assert that this is the most important graph ever conceived (my earlier blog post on its origins and importance).

Why the transition within the integrated circuit era? Intel lost to NVIDIA for neural networks because the fine-grained parallel compute architecture of a GPU maps better to the needs of deep learning. There is a poetic beauty to the computational similarity of a processor optimized for graphics processing and the computational needs of a sensory cortex, as commonly seen in neural networks today. A custom chip (like the Tesla D1 ASIC) optimized for neural networks extends that trend to its inevitable future in the digital domain. Further advances are possible in analog in-memory compute, an even closer biomimicry of the human cortex. The best business planning assumption is that Moore’s Law, as depicted here, will continue for the next 20 years as it has for the past 120.

 

[NoGasNoBrakes]

I wish they had shown the robot next year… after the CT, semi, Roadster and 4680s were rolling out of Austin and Berlin.
The FSD and dojo stuff should have been enough for recruiting. The robot seems like a bit of distraction.

The main reason they announced the robot is because it directly relates to FSD, since the robot is an application of the tech being developed by Tesla for FSD. The same neural net, sensors, cameras, and image processing used by FSD will also be used by the robot. Although the idea is still early-on, a prototype by next year is no small feat, and I think many investors and engineers are extremely excited about the robot as it shows that Tesla is really planning on using the tech they've developed for much more than just autonomous vehicles.

 

[ZeApelido]

Actually it's 1,000 Model Y per day. Assuming say 340 days production that's 340,000 per year. Although actual run rate would be less than peak production rate.

Here's the description from the video:
After four days of overhaul at Tesla's Shanghai plant last week, the production capacity of Phase II Modle Y has reached 1,000 units per day. This daily capacity has exceeded the number of Modle 3’s 800 units per day. It can be said that the Y-type production has entered High-yield stage. As previously planned by Tesla, the main production of Q3 is for export.

I expect we'll see front and rear casting Model Y's out of Shanghai once the newly installed casting machines are operational. We should also get a structural battery pack once LG or CATL start producing them. Based on recent rumours the latter could come from the soon to be built CATL Shanghai factory. This may be the longer range Model Y Tesla recently registered with Chinese authorities.

My wife was talking to me so I failed.

 

[wipster]

<snip>
When we consider where the Robot might be deployed:-

• Delivery - different situations encountered, training is similar to FSD.

The dumb "task oriented" Robot, is mostly what we need.

It can certainly do a better job than my mail delivery person...

 

[JustMe]

Gordon Johnson said Tesla’s AI event showed they use simulated data for their neural net training and lied about using real world data from the fleet.

Steven Mark Ryan offered this gentle correction:

 

[Singuy]

And the beat goes on .. The cars keep going faster all the time ..

Steve Jurvetson (early Tesla Investor, board member) on Twitter
"Tesla now holds the mantle of Moore’s Law. Just as NVIDIA took leadership from Intel a decade ago.
While the substrate has shifted several times, humanity's capacity to compute has compounded for 122 years as if on rails.
It's surreal.

View attachment 699371


From Steve Jurveston's Flickr

122 Years of Moore's Law + Tesla AI Update

Tesla now holds the mantle of Moore’s Law, with the D1 chip introduced last night for the DOJO supercomputer (video, news summary).

This should not be a surprise, as Intel ceded leadership to NVIDIA a decade ago, and further handoffs were inevitable. The computational frontier has shifted across many technology substrates over the past 120 years, most recently from the CPU to the GPU to ASICs optimized for neural networks (the majority of new compute cycles).

Of all of the depictions of Moore’s Law, this is the one I find to be most useful, as it captures what customers actually value — computation per $ spent — on a log scale, so a straight line is an exponential.

Humanity’s capacity to compute has compounded for as long as we can measure it, exogenous to the economy, and starting long before Intel co-founder Gordon Moore noticed a refraction of the longer-term trend in the belly of the fledgling semiconductor industry in 1965.

In the modern era of accelerating change, it is hard to find even five-year trends with any predictive value, let alone trends that span the centuries. I would go further and assert that this is the most important graph ever conceived (my earlier blog post on its origins and importance).

Why the transition within the integrated circuit era? Intel lost to NVIDIA for neural networks because the fine-grained parallel compute architecture of a GPU maps better to the needs of deep learning. There is a poetic beauty to the computational similarity of a processor optimized for graphics processing and the computational needs of a sensory cortex, as commonly seen in neural networks today. A custom chip (like the Tesla D1 ASIC) optimized for neural networks extends that trend to its inevitable future in the digital domain. Further advances are possible in analog in-memory compute, an even closer biomimicry of the human cortex. The best business planning assumption is that Moore’s Law, as depicted here, will continue for the next 20 years as it has for the past 120.

The graph shows the Dojo beating a Nvidia Titan X...that came out in 2015 without any tensor cores. I get people are bullish but graphs like these really brings down credibility of bull's research.

 

[mltv]

https://twitter.com/x/status/1428703824601026562

I think I know the reason Waymo is refer as Wamo : GJ can’t reason because he missing the Y

 

[Chunky Jr.]

Lisa Simpson : Better to remain silent and to be thought a fool than to open your mouth and remove all doubt
Gordon Johnson (thinking) : What does that mean? I better say something or they'll think I'm stupid
Gordon Johnson : AI day proves Tesla uses a simulator and collects no real world data!
Gordon Johnson (thinking) : Swish!

 

[CLK350]

The graph shows the Dojo beating a Nvidia Titan X...that came out in 2015 without any tensor cores. I get people are bullish but graphs like these really brings down credibility of bull's research.

From Wikipedia: " . Moore's law is the observation that the number of transistors in a dense integrated circuit (IC) doubles about every two years ."

Nvidia's TitanX card ca 2015 has 8 Billion transistors, Tesla's Dojo D1 chip ca 2021 has 50 Billion transistors - 2021 - 2015 = 6 years, doubling ea 2 years means 8X increase, so ok 50 isn't 64 but close enough. Besides, the Dojo Clip has 25 D1 chips onboard ..

 

[Singuy]

From Wikipedia: " . Moore's law is the observation that the number of transistors in a dense integrated circuit (IC) doubles about every two years ."

Nvidia's TitanX card ca 2015 has 8 Billion transistors, Tesla's Dojo D1 chip ca 2021 has 50 Billion transistors - 2021 - 2015 = 6 years, doubling ea 2 years means 8X increase, so ok 50 isn't 64 but close enough. Besides, the Dojo Clip has 25 D1 chips onboard ..

Nvidia's A100 has 54.2billion transistors, more than D1. And we don't count 25 d1 chips as that's a package using off chip interconnect.

 

[CLK350]

Nvidia's A100 has 54.2billion transistors, more than D1. And we don't count 25 d1 chips as that's a package using off chip interconnect.

Agreed, good points, thanks - will leave it up to you to chastise Steve J. on his Twitter feed (or Flickr stream) for being unfairly partial to Tesla by not also mentioning Nvidia'a Ampere A100 ; )

Edit: ChananBos already did on Twitter