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

[KBF]

Tesla VP responds to US Senator’s misinformed take on Autopilot recall

As per the Tesla VP, the electric vehicle maker is looking forward to working even with its most aggressive critics.

www.teslarati.com

I find it encouraging that Patel is speaking publicly for Tesla, and is addressing FUD from politicians. That bodes well for the future of TSLA communications, IMO.

 

[AudubonB]

Some field report on the EV state here in Brazil

Was in São Paulo for two days, the amount of EVs I saw dwarfed my previous visit, lots of BYD “taxi”, some other electric delivery contraptions and a reasonable number of private owners vehicle

But the BYD taxis was a lot, to the point that not even looking for them I saw everywhere

Obviously I took the opportunity of doing a test drive in both the BYD Dolphin and the GWM Ora, both for the equivalent of $30k, pretty nice, but expensive for what it is, but the big fail and what tells me Tesla has nothing to fear, the software in both was absolutely garbage, confusing mess of menus, bad screen and the biggest turn of, there is no built in route planing with charging, none, zip, nada. You are on your own using third party apps to figure it out, no estimate SoC on arrival also even without considering chargers

Both would be ok for urban and short highway trip, but that combined with a low range, doesn’t make it a compelling purchase, which is a shame, because I absolutely loved the Ora and might get one in the future if Tesla takes too long to offer something here

Both can’t do a 250 km trip without charging, the GT version of the Ora can but it’s considerably more expensive, and that only in optimal conditions, bring a little cold, wind and rain and range drops a lot

Sadly I couldn’t find any Teslas, I knew there were some used ones for sale nearby, but they were in the same stores that sell Ferraris, Lamborghinis, Rolls Roxie and similar, and my lack of shame only goes so far

Adding to all that, BYD store was packed, GWM only me there

A brother-in-law owns a BYD Dolphin; I'll try to learn from my sister what his response to your critique is (she owns a Model S ).

 

[Iain]

Some field report on the EV state here in Brazil

Was in São Paulo for two days, the amount of EVs I saw dwarfed my previous visit, lots of BYD “taxi”, some other electric delivery contraptions and a reasonable number of private owners vehicle

But the BYD taxis was a lot, to the point that not even looking for them I saw everywhere

Obviously I took the opportunity of doing a test drive in both the BYD Dolphin and the GWM Ora, both for the equivalent of $30k, pretty nice, but expensive for what it is, but the big fail and what tells me Tesla has nothing to fear, the software in both was absolutely garbage, confusing mess of menus, bad screen and the biggest turn of, there is no built in route planing with charging, none, zip, nada. You are on your own using third party apps to figure it out, no estimate SoC on arrival also even without considering chargers

Both would be ok for urban and short highway trip, but that combined with a low range, doesn’t make it a compelling purchase, which is a shame, because I absolutely loved the Ora and might get one in the future if Tesla takes too long to offer something here

Both can’t do a 250 km trip without charging, the GT version of the Ora can but it’s considerably more expensive, and that only in optimal conditions, bring a little cold, wind and rain and range drops a lot

Sadly I couldn’t find any Teslas, I knew there were some used ones for sale nearby, but they were in the same stores that sell Ferraris, Lamborghinis, Rolls Roxie and similar, and my lack of shame only goes so far

Adding to all that, BYD store was packed, GWM only me there

I’m landing in Campinas, Brazil in a few days. Then driving passed Sao Paulo down to Santos for ten days. Now I’ll have something to do on the drive ( count Teslas)

 

[Gigapress]

https://twitter.com/x/status/1737246827047538777

1m CTs / annum seems a little high

I think 1 million Cybertruck/year is about right. Might even be too low. I think the best way to frame this question is: will the Cybertruck be yet another category-killer like S, X, 3 and Y have been? Every time Tesla has launched a new vehicle model, it has eventually dominated its market segment, taking circa 30% market share.

Last year in the US, the Ford F-series lineup led pickup truck sales with 653k units, while the market total was 2.7M (GoodCarBadCar). And 2022 was a down year. The peak was 2019 with 897k F-series and 3.1M total (link).

So if Tesla's Cybertruck outsells the F-series, then we are talking about annual volume of 1M+ in America alone, not even counting Canada and elsewhere.

Bear in mind, on the one hand, that the F-series is a family of trucks in different shapes and sizes. CT's comparatively limited selection may limit its appeal relative to the entire F-series. On the other hand, the CT is also drawing in many people who previously didn't even want a pickup truck. Also, CT checks a lot of boxes for people who would otherwise buy an SUV like a Chevy Suburban. Soccer moms, etc.

In the US overall, the "light trucks" category, which includes both pickups and SUVs, is seeing volume of 12.5 M/yr these days, and the long-term trend is increasing popularity. A few of the main limiting factors on pickups being even more popular are fuel and maintenance costs and environmental concerns. Cybertruck eliminates these drawbacks.

Motor Vehicle Retail Sales: Light Weight Trucks

Graph and download economic data for Motor Vehicle Retail Sales: Light Weight Trucks (LTRUCKSA) from Jan 1976 to Mar 2024 about light weight, trucks, vehicles, retail trade, new, sales, retail, and USA.

fred.stlouisfed.org

 

[elasalle]

https://twitter.com/x/status/1737259411448500552

 

[Knightshade]

I think 1 million Cybertruck/year is about right. Might even be too low. I think the best way to frame this question is: will the Cybertruck be yet another category-killer like S, X, 3 and Y have been? Every time Tesla has launched a new vehicle model, it has eventually dominated its market segment, taking circa 30% market share.

I know S got at least to mid-30% share in large luxury sedans, but I don't think X ever did in a similar SUV segment?

Likewise I know Model 3 was the best selling small or mid-sized premium sedan but did it actually get to 30% share? The #s I can find don't put it that high but it's a bit all over the place.

And Model Y sells like crazy, but how narrow a category gets you to 30% share, given there's over 30 million SUVs sold annually?

 

[GhostSkater]

I’m landing in Campinas, Brazil in a few days. Then driving passed Sao Paulo down to Santos for ten days. Now I’ll have something to do on the drive ( count Teslas)

Sadly I doubt you will see one, I’ve never seen one on the road, saw year ago in a event and recently in Curitiba, friends that live in São Paulo saw maybe two or three times

Just adding a bit more to the comparison, I simulated a bunch of routes that I would like to do with the Ora GT, and threw a Model 3 RWD for fun, turns out that on average, the Model 3, being a bigger car, with more features, more space, way more storage, trunk etc, is 50% more efficient

Ora GT does on the average of all simulated trips 174 Wh/km, while the Model 3 does 116 Wh/km

This is where Tesla wins, good engineering everywhere, but specially where it matters, I’m sorry, I know the Ora starts in China way cheaper than Model 3, $16k vs $35k, but that isn’t a excuse to such bad efficiency

 

[navguy12]

Just read the article. Unfortunately it seems to lean more “true” than “FUD”.

Clear documented evidence that tesla indeed knew they were shipping cars with sub-optimal suspension components with a high failure rate leading to accidents, and internal communications confirming that they had finally fixed the issue with a redesigned part in 2022. They were fixing the part for free in China, but in the US service staff told to blame the drivers behavior for the part failing. NHTSA also confirms they are investigating similar issue with power steering.

Wouldn't surprise me in the least if this leads to an actual physical recall of all pre-2022 cars. (One wonders how service centers will be able to deal with that.)

Also large class action lawsuit seems inevitable.

Would be happy to hear from others as to why this wont happen (but please no “because Reuters is dumb” responses please.)

They were fixing the part for free in China, but in the US service staff told to blame the drivers behavior for the part failing.

This is the one item that caught my attention, the way it was handled in China versus outside of China.

 

[Gigapress]

I know S got at least to mid-30% share in large luxury sedans, but I don't think X ever did in a similar SUV segment?

Likewise I know Model 3 was the best selling small or mid-sized premium sedan but did it actually get to 30% share? The #s I can find don't put it that high but it's a bit all over the place.

And Model Y sells like crazy, but how narrow a category gets you to 30% share, given there's over 30 million SUVs sold annually?

30% is very rough. The general point is that Tesla has a history of introducing outlier products that win their market segments by a wide margin, and if the Cybertruck repeats the feat then we are looking at sales of 1M/year or more.

I could be remembering wrong for X; I haven't checked the numbers in a couple years.

For the US market, Model Y fits best in the small premium SUV segment. Based on GoodCarBarCar.com data for this segment, I calculate 39% US market share for 2023 for Model Y.

Likewise, I get 37% share for Model 3 in the small premium car segment.

Spoiler: Model Y vs competition table

Model	2023 Q1 thru Q3	% Share
Tesla Model Y	93,999​	39%​
Audi Q5	19,365​	8%​
Lexus NX	17,647​	7%​
BMW X3	14,670​	6%​
Mercedes-Benz GLC/GLK-Class	11,504​	5%​
Volvo XC60	9,318​	4%​
Lincoln Corsair / MKC	7,852​	3.3%​
Porsche Macan	7,525​	3.1%​
Volvo XC40	7,478​	3.1%​
BMW X1	5,700​	2.4%​
BMW iX	5,541​	2.3%​
Cadillac XT4	5,301​	2.2%​
Mercedes-Benz GLB	5,111​	2.1%​
Audi Q3	3,708​	1.5%​
Mercedes-Benz EQB	3,270​	1.4%​
Cadillac Lyriq	3,018​	1.3%​
Lexus UX	2,572​	1.1%​
Mercedes-Benz GLA-Class	2,529​	1.1%​
Audi Q4 e-tron	2,446​	1.0%​
Infiniti QX50	2,435​	1.0%​
Infiniti QX55	1,419​	0.6%​
Lexus RZ	1,394​	0.6%​
Alfa Romeo Stelvio	1,280​	0.5%​
Volvo C40	1,253​	0.5%​
Genesis GV60	898​	0.4%​
Audi Q4 e-tron Sportback	834​	0.3%​
Land Rover Range Rover Evoque	677​	0.3%​
Land Rover Discovery Sport	508​	0.2%​
Jaguar E-Pace	84​	0.0%​
BMW X2	35​	0.0%​
Audi eTron	15​	0.0%​

Spoiler: Model 3 vs competition table

Model	2023 Q1 thru Q3	% Share
Tesla Model 3	52,000​	37.1%​
BMW 4-Series	13,668​	9.8%​
BMW 3-Series	8,852​	6.3%​
Acura Integra	8,320​	6%​
Mercedes-Benz C-Class	8,058​	5.7%​
Lexus IS	6,054​	4.3%​
Audi A5	6,037​	4.3%​
Genesis G70	4,455​	3.2%​
Cadillac CT5	4,408​	3.1%​
Acura TLX	4,091​	3%​
Volvo 60-Series	3,167​	2.3%​
BMW 2-Series	3,135​	2.2%​
Audi A3	2,808​	2%​
Mercedes-Benz CLA-Class	2,785​	2.0%​
Audi A4	2,521​	2%​
Cadillac CT4	1,903​	1.4%​
Infiniti Q50	1,827​	1.3%​
Audi e-Tron Sportback	1,755​	1.3%​
Alfa Romeo Giulia	864​	1%​
Infiniti Q60	834​	0.6%​
Audi Q4 e-tron Sportback	834​	0.6%​
Land Rover Range Rover Evoque	677​	0.5%​
Land Rover Discovery Sport	508​	0.4%​
Lexus RC	472​	0.3%​
Jaguar E-Pace	84​	0.1%​
BMW X2	35​	0.0%​
Audi eTron	15​	0.0%​
Mercedes-Benz A-Class	9​	0.0%​
Jaguar XE	0​	0.0%​
BMW i3	0​	0.0%​
Acura ILX	0​	0%​

 

[M|S|M MYP]

Jim Cramer says it's time for investors to take some profit in this hot market

"It's almost impossible to call the top in an individual stock, so what you need to do is sell your winners gradually on the way up," CNBC's Jim Cramer said.

www.cnbc.com

$TSLA = $300 SP confirmed



MERRY CHRISTMAS EVERYONE!!!

Just yesterday this said to take some profit and today he says rally will continue. What a

 

[Musskiah]

I think 1 million Cybertruck/year is about right. Might even be too low. I think the best way to frame this question is: will the Cybertruck be yet another category-killer like S, X, 3 and Y have been? Every time Tesla has launched a new vehicle model, it has eventually dominated its market segment, taking circa 30% market share.

Last year in the US, the Ford F-series lineup led pickup truck sales with 653k units, while the market total was 2.7M (GoodCarBadCar). And 2022 was a down year. The peak was 2019 with 897k F-series and 3.1M total (link).

So if Tesla's Cybertruck outsells the F-series, then we are talking about annual volume of 1M+ in America alone, not even counting Canada and elsewhere.

Bear in mind, on the one hand, that the F-series is a family of trucks in different shapes and sizes. CT's comparatively limited selection may limit its appeal relative to the entire F-series. On the other hand, the CT is also drawing in many people who previously didn't even want a pickup truck. Also, CT checks a lot of boxes for people who would otherwise buy an SUV like a Chevy Suburban. Soccer moms, etc.

In the US overall, the "light trucks" category, which includes both pickups and SUVs, is seeing volume of 12.5 M/yr these days, and the long-term trend is increasing popularity. A few of the main limiting factors on pickups being even more popular are fuel and maintenance costs and environmental concerns. Cybertruck eliminates these drawbacks.

View attachment 1001556

Motor Vehicle Retail Sales: Light Weight Trucks

Graph and download economic data for Motor Vehicle Retail Sales: Light Weight Trucks (LTRUCKSA) from Jan 1976 to Mar 2024 about light weight, trucks, vehicles, retail trade, new, sales, retail, and USA.

fred.stlouisfed.org

Great deep dive! The Cybertruck may define a class unto itself. It will likely prove better at "picking up" (double entendre intended) than a pickup truck class and certainly knocks the "sport" and "utility" out of the SUV class. We should find it converts many former SUV owners into first time truck owners - such as meself.

 

[TN Mtn Man]

Firstly, pretty much any BEV sold from this point on will eventually (<2 years) be compatible with NACS. That help considerably with this issue.

Secondly, If the government can stick to this commitment (why do I feel certain they'll be kicking this down the road at some point), the beauty of a free-market economy is if anyone can figure out how to profit from providing charging infrastructure, they'll build it. Having some certainty that by 2035 100% of new vehicles sold in Canada will be electric (along with measurable intermediate goals) does give some certainty to an enterprise that there will be some demand for their charging network, assuming it's competitive, i.e. convenient, reliable and affordable. Without that commitment from government, it becomes a much more risky investment.

If we sit and complain about how bad the charging is in some areas but no clear direction is given, the problem will never be resolved as no-one will build the vehicles as there's a lack of charging infrastructure and no-one will build out the charging infrastructure as there will be no demand for charging services if there are no EVs to charge. This is a necessary step. By the government setting these targets it gives manufacturers a reason to hit these targets along with defined penalties for not doing so. It also gives charging networks some certainty that there will be some demand for their services. It also sends a clear message to consumers about the direction the transportation industry is headed. Now everyone can go about solving these problems.

I understand your point, but disagree with your conclusion, at least to some extent. EV adoption is where ICE adoption was around 100-110 years ago. Auto owners were still "early adopters", generally owned by the more affluent buyers (the intro of the Model T in '08 started to change that, but it's price actually dropped over the next decade). There weren't gas stations on every corner. Gas was bought at pharmacies and hardware stores in 1-5 gallon containers. Yet still, without government intervention, the market expanded. Vehicles that people wanted were built, private industry and the free market responded with the fueling infrastructure needed. Government intervention tends to introduce uncertainty, companies hold off investing to see what handouts the government might offer-and they change from year to year (if not month to month) and can be "reinterpreted" at any time-as we see with the IRA tax credits. Companies that have to make commitments that might take years to pay off have to do so with government intervention that changes on the whim of some bureaucrat. EVs, particularly Teslas, sell because they are the best tool for a lot of people. And will continue to do so as they become more affordable-and as the DCFC infrastructure in particular continues to build out.

 

[MikeC]

Just yesterday this said to take some profit and today he says rally will continue. What a

View attachment 1001573

To be fair - I don't see that tweet on his feed so it may be a fake.

 

[Artful Dodger]

Giga Nevada workers are set to receive a 10% wage increase on Jan 1st. Anybody want to guess that 2170 pack output will increase by more than 10%?

SR+ started with 2170 cells... jus' sayin'

And what's that new chemistry switch at Giga Texas? Freeing up cathode supplies for Sparks 2170s?

I highly doubt Tesla doesn't react to the loss of the $7,500 IRA credit on certain lower-range products. This messes up the price ladder, and has to be addressed.

 

[2daMoon]

Great deep dive! The Cybertruck may define a class unto itself. It will likely prove better at "picking up" (double entendre intended) than a pickup truck class and certainly knocks the "sport" and "utility" out of the SUV class. We should find it converts many former SUV owners into first time truck owners - such as meself.

I like to think of the CT as the ultimate urban assault hatchback for the zombie apocalypse.

 

[GhostSkater]

Some field report on the EV state here in Brazil

Was in São Paulo for two days, the amount of EVs I saw dwarfed my previous visit, lots of BYD “taxi”, some other electric delivery contraptions and a reasonable number of private owners vehicle

But the BYD taxis was a lot, to the point that not even looking for them I saw everywhere

Obviously I took the opportunity of doing a test drive in both the BYD Dolphin and the GWM Ora, both for the equivalent of $30k, pretty nice, but expensive for what it is, but the big fail and what tells me Tesla has nothing to fear, the software in both was absolutely garbage, confusing mess of menus, bad screen and the biggest turn of, there is no built in route planing with charging, none, zip, nada. You are on your own using third party apps to figure it out, no estimate SoC on arrival also even without considering chargers

Both would be ok for urban and short highway trip, but that combined with a low range, doesn’t make it a compelling purchase, which is a shame, because I absolutely loved the Ora and might get one in the future if Tesla takes too long to offer something here

Both can’t do a 250 km trip without charging, the GT version of the Ora can but it’s considerably more expensive, and that only in optimal conditions, bring a little cold, wind and rain and range drops a lot

Sadly I couldn’t find any Teslas, I knew there were some used ones for sale nearby, but they were in the same stores that sell Ferraris, Lamborghinis, Rolls Roxie and similar, and my lack of shame only goes so far

Adding to all that, BYD store was packed, GWM only me there

Rolls Roxie, yeah Apple, this is exactly what I meant

If there is one player we don’t have to worry about competing with Tesla in the AI space, it’s them. Having an AI is capable of understand the world and correcting mistakes, awesome. Having an AI that takes what was right and makes it wrong? lol

 

[dhanson865]

I watched the whole thing.

... snip

6. Matt mentioned Edward Neidermeyer at least once (I think as a source). You know, the guy with the Tesla death clock website.

He what? Did you pay attention to a small mention I missed in a 2+ hour video or did you confuse Ralph Nader and Unsafe At Any Speed with Edward Neidermeyer?

 

[CaliBear1]

Day 06: Six geese a-laying | Megapack To Excel​

Part of 12 Days of Christmas - Tesla Edition a series (c) by the Artful Dodger, Dec 2023

Over this Yuletide season, I will post a daily installment focusing on Tesla products, past, present, and future (please note that I will express major themes as short-hand bullet points, or I will run out of Yule before the tide peaks). Here's the series so far:

Day 01: A Partridge in a Pear Tree | Roadster Proof of Concept
Day 02: 2 Turtle Doves | S/X Fraternal Twins go Mainstream
Day 03: 3 French Hens | Model 3 Bets the Company
Day 04: 4 Calling Birds | Model Y Built at Four Factories
Day 05: Five Golden Rings | Semi Breaks Physiks

Intro to Part 6: Megapack To Excel​

Megapack is the answer to the question everybody was afraid to ask: if renewables are cheaper, why doesn't everybody do it? (hint: based on it's supply'n'demand economics).

1. Getting those Ducks in a Row​

• Folks love to say the problem with renewables is that the sun doesn't always shine, and the wind doesn't always blow. This is incorrect
• it's about the timing of peak power availability vs peak demand, since on the traditional grid electricity must be consumed within moments of its generation (or the grid becomes unstable)
• Renewables and the 'Duck curve'
  • the real achilles heel of solar energy is that peak demand occurs in the evening when people get home from work, often after the Sun has set, or its after peak insolation
  • the solution is to pair solar power with battery storage, thus flattening the 'duck curve', and creating arbitrage opportunities (c.f. Tesla Autobidder +Virtual Powerplants)
  • Wind is less variable on a hourly basis, and tends to be deployed in commercial wind farms, where geographic distribution (site planning) and added storage make it work
  • Read about the first large deployment of Telsa Megapacks near Adelaide Australia, called the Horndale Power Reserve. This $90M project (which was done in 90 days in 2017 on a dare from Aussie Billionaire Mike Cannon-Brooke) paid for itself within 2 years, and single-handedly saved the South Australian grid several times. Key to future success is the number of even larger such 'big battery' projects now planned or already deployed in Australia - Elon once said 'a continent is just a big island.'
• Whoops, sat on a Duck
  • futher big-battery projects came rapidly after the Aussie success story became widely known, notably the islands of Hawaii and Puerto Rico (huricane-damaged grid)
  • Perhaps the most famous example is the Moss Landing project near San Francisco, which was Tesla's 1st ever 1GWh storage deployment, one so large its capable of running the entire City of San Francisco for 4 hrs - this makes daytime solar extremely valuable (charge), and overnight wind extremely useful (recharge)
  • the economics of arbitrage were now clear: Megapacks let you buy low and sell high
  • even more important you don't need to repair natural gas 'peaker plans' which are now quickly being replaced by grid-scale energy storage projects

Lesson 1: Build the missing piece of the puzzle - Storage infrastructure

2. The Right Stuff: Megapack 2 XL​

• Tesla Megapacks were mostly on the back-burner during the Model 3/Y ramp, the pandemic, and the supply shortage due to lack of battery cells (Giga Nevada used a nickel-based Li-Ion chemistry, and made 10x more cells for the auto business than the energy business)
• Finally, Tesla arrived at the optimal design for their new energy storage product, based on the size and weight of the ubiquitous shipping container: the Megapack 2 XL (4 MWh/pack)
• this form-factor speeds logistics, making it readily adaptable to existing transport infrastructure
• Megapack 2 XL also saw the switch to much cheaper LFP bty cells based on $/KWh*lifecycle (LFP is also longer lasting, more supply available)
• Of course, this new flagship TE product would be built at a new dedicated Tesla factory in Lathrop, CA (about 30 miles N. of the Fremont factory) which is also the lead factory in an upcoming fleet of 'Megafactories'.

Lesson 2: Go big, then go wide (and deep)

3. Megafactories: "The Business model"

• Also coined the 'Megapacktory', this consists of two 20 GWh/yr modules that are designed to be cloned quickly to increase production as required
• Tesla Energy large-scale storage solutions have been called 'recession proof' because their main customers are large utility operators, who's capital budgets are set years in advance and are largely unaffected by interest rates (thanks, Public Utilities Boards, who set the tariffs to ensure minimun profitability of large utility companies regardless of the wider economy).
• Quasi-infinte demand:
  • Electrical grid and generation companies have on the order of 1,200 existing gas and coal fired generators is the USA alone. Each one of these sites typically includes 'peaker plants' which are about twice as expensive to start and operate for the few hrs per day needed
  • when faced with replacing an end-of-life peaker plant (4 years / $200M), engineers and MBA's look favorably at a 1 year / $100M Megapack solution. So do CEO's.
• Flattening the TSLAQs
  • Megapack has a unique business model within Tesla
  • Upfront payments from Customers at the time of order largely cover the initial costs to build their megapacks (typically 20% paid upfront), plus a further stream of payments based on achieving milestones throughout deployment, test, and commissioning phases
  • Each Megapack deployment also earns an annual maintenance fee (which is often conducted remotely via computer) and contributes continuing revenue over the lifetime of the project
  • Due this back-loaded income, Tesla rides a tsunami of steadily-building revenue from its Energy division as the size of the fleet increases year after year (at 12.5 GWh now per Tesla)
    • Giga Shanghai right now is in the proces of building it's first 40GWh/yr Megapack factory, conveniently located just 3 km from the newly-commissioned 80 GWh/yr CATL LFP battery factory)
    • because the Megafactories are modular, I look for the Next Gen to also be in China: a factory to build build more Megafactories
    • for Tesla to make good on their intention to create 1TWh/yr of capacity for Megapack, that's 1,000/40 = 25 Factories the size of Lathrop needed (which is also 4/yr for 6 years <- they definately need a factory to build those factories)

Lesson 3: If you build it, they will buy it

Conclusion:​

Megapack embodies the low-hanging fruit in the transition to renewable energy, bolstering the electrical grid while achieving outstanding ROI. Tesla made the obvious move by creating a new product line focused on the grid. So obvious, that no other manufacturer (esp. Automakers) have been able to follow their lead. Why? Simply because others don't have the Right Stuff, the Secret Sauce, and the chutzpah to create a new sub-industry. Tesla has this right stuff.

Tesla leveraged their experience and expertise in power electronics, manufacturing, and their supply chain to build a recession-proof business line that is already changing the world, and stands to become even larger than Tesla's auto business. Megapack is the Goose that layed the Golden Egg (and just keeps on laying).

Next: Cybertruck: 7 new technologies debut in a world-beating new flagship truck - to da Moon!

Tomorrow's Topic:

Day 07: Seven swans a-swimming | Cybertourdeforce

Dodger, you are amazing! Happy Holidays to you, the indefatigable Mods and all the stalwarts here at TMC. The future is indeed bright!

 

[Artful Dodger]

Day 09: Nine Ladies Dancing | Dojo as a Service​

Part of 12 Days of Christmas - Tesla Edition a series (c) by the Artful Dodger, Dec 2023

Over this Yuletide season, I will post a daily installment focusing on Tesla products, past, present, and future (please note that I will express major themes as short-hand bullet points as I Yule tidy-up the loose ends). Here's the series so far:

Day 01: A Partridge in a Pear Tree | Roadster Proof of Concept
Day 02: 2 Turtle Doves | S/X Fraternal Twins go Mainstream
Day 03: 3 French Hens | Model 3 Bets the Company
Day 04: 4 Calling Birds | Model Y Built at Four Factories
Day 05: Five Golden Rings | Semi Breaks Physiks
Day 06: Six geese a-laying | Megapack To Excel
Day 07: Seven Swans a-Swimming | Cybertourdeforce
Day 08: Eight Maids A-Milking | Model 2 World Car

Intro to Part 09: Nine Ladies Dancing | Dojo as a Service​

Back at Tesla's first AI Day on Aug 19, 2021, the stage was set for the unexpected debut of the DOJO chip. This turned out to be the basic functional unit of a much larger supercluster consisting of 'training tiles' of these chips, run in a massively parallel group. This chip pron revealed the key to Tesla's secret plan to create their own neural net (AI) training system, and promised to become one of the world's largest supercomputers (designed and built in-house by Tesla engineers).

1. I/O Bottleneck ​

• The problem with hand-coding FSD is that, after writing 300K lines of C++ code, the original problem is not important (and still buggy, and bulky, and over-budget)
• Artificial Intelligence in the form of multiple Neural Nets (NN) changes the paradigm by providing literally billions of tunable parameters at the expense of not having direct control over individual weights (indeed, attributing 'causality' in a NN requires an AI psychologist)
• So we trade the limitations of human programmers for the uncertainty of statistics, generated by a NN training system which processes vast amounts of data to create its end product: a NN intended to be used to make decisions in realtime by a much more modest agent
• Project Dojo was designed to make that training regime more efficient. It was built to move data. Really, really large amounts of data. Like a fleet of VW Buses full of data. Quickly.
• What previously took 30 days to create with Tesla's nVidia A100 CPU cluster may take just 8 hrs on DOJO. Why is that important? Because the Autopilot team is experimenting with various techniques to see what type and amount of data, and what level of architecture vs self-learning is most effective. This is empirical science and engineering; there is no wrong path, only paths that lead nowhere. The 'Fail Fast' mantra of Silicon Valley applies to Tesla AI/Robotics.
• In 2023, Tesla spent Billions on nVidia GPU hardware to support its insatiable NN trg needs, and will spend more next year. Eventually though, Telsa will be the vendor of superfast NN trg equipment, shirley one of the fastest growing sectors of the economy over the next 20-30 yrs

Lesson 1: Move it or Lose it

2. A New Metric ​

• In 2016, Tesla began its own CPU design project to create the FSD chip used in HD3 (in service since Apr 2019). It was 20x more powerful than the MobilEye CPU it replaced, but consumed <100W of power. Thus a new mantra was coined: "Useful Compute per KWh"
• Tesla bought an "NN trimming" company as an 'aqui-hire' for their 12 engineers. This small group demonstrated skill in greatly reducing the size of a NN w/o dramatically affecting performance. Indeed the entire Tesla Autopilot team is only a few hundred engineers.
• As always, there is a delicate dance between the cost of training the NN and the size of the NN which can be used by the agent (ie: FSD Computer). This is the art in engineering, and it takes time and effort to reach the right balance. Right now, Tesla is throwing more resources at the training hdw side, believing that is the low-hanging fruit. We should know in a year.

Lesson 2: Make it Fast; Make it Efficiently

3. AI training as a Service​

• Dojo v1 is optimized for processing video. Except for other automakers who may decide to license Tesla FSD (and possibly Optimus Robot applications) there is a relatively small market for video training (as we speak, but this may change)
• Dojo v2 has hardware optimized for general AI training, such as large language models (LLMs) and generative AI (ie: deep fakes!) and thus has a much broader TAM
• It is this new generation of Dojo training services which I believe Tesla intends to commercialize (their are job posting out today for a Data Center API Programmer). Who knows where this will go, but I am certain TSLA shareholders will 'get their beaks wet' in this AI fountain
• Lastly, the race to AGI is a National Security issue, with both China and Russia staking out claims in this latest land rush. The U.S.A. will not be left behind (kicking and screaming).

Lesson 3: Build a Better Mouse Trap (and they'll build a better mouse)

Conclusion:​

Tesla Dojo will be the 'Watt steam engine' that powers AI agents (ranging from FSD to Optimus), and increasingly, some 3rd-party AI applications, until the dawn of true AGI. Right now, the agent (whether a car or humanoid) has only a limited "inference" CPU, meaning it can not learn directly, it can only apply a set of connections created in the Data Center training environment. It is my belief we won't see true AGI until the agent can dream (reprocess experiences internally, conduct what-if trials, and update it's own NN matrix). Whether Tesla that makes that leap, or some other entity remains to be seen. And whether that day is 4 yrs in the future, or 40, does not matter a wit if we humans can not thrive in our new enviroment. The time is now to stake a claim.

Next up: Humanoid robots in the Factory (and Healthcare, and gasp, the Post Office)

Tomorrow's Topic:

Day 10: Ten Lords A-Leaping | Teslabot FTW​

​

 

[Nolimits]

“ Also, CT checks a lot of boxes for people who would otherwise buy an SUV like a Chevy Suburban. Soccer moms, etc.”

Nailed it. That’s how I’ve been thinking CT these days. Larger suburban. For those who don’t need a 3rd row but more space. That Vault on the CT is a game changer