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

[lklundin]

But I remember one of Tesla-Björns earlier videos where he and his pal Jörgen turned the Millennium Falcon into a sauna using the cabin heater and posing more or less undressed. So a lack of sauna can't be the reason, unless Finns find it lacking. Which is of course a distinct possibility. Anyway, one contrary factual argument.

OK, in the absence of posts purporting to be from Finnish people, I will just throw in a random (and confirmed several times) anecdote:

Q: Why is Vodka in Finland sold in 50cl bottles, as opposed to 70cl ones?
A: Because you drink one bottle every day, so 70cl would be too much.

PS. Also, Finland is the country that gave is Linus Torvalds, thus Linux, the Operating System of the Tesla.

 

[Tslynk67]

Speaking of cult-like behavior, it doesn’t help the cause when a bunch of Twitter accounts portray Musk as the Messiah, even photoshopping old Christian paintings and superimposing Musk’s face in them. There’s an increasing amount of that on Twitter and I sure wish these folks would cut it out and focus on critical thinking and facts to combat the fud/misinformation.

Dude, get a SOH FFS!

 

[Fact Checking]

Really good exclusive article just got released about a Fremont factory visit CleanTechnica performed this month, with a lot of new technical details:

A Look Inside Tesla's Fremont Automotive Factory — #CleanTechnica Exclusive | CleanTechnica

"Tesla’s team installed and commissioned the Model 3 servo press in ~25% less time than the fastest installation Schuler had ever done before. And that was just the starting point."

"Changes like this and more allowed Tesla to increase the rate of the press from 12 strokes per minute to 14 strokes per minute, a 16% improvement. That translates to more throughput from an extremely expensive piece of equipment and less capital required to scale up production. Putting that into context, Tesla can now produce body panels and stampings from the Schuler press at twice the rate as its Model X stamping press line and a mind boggling four times the rate of the older Model S production lines."

"Tesla’s Manufacturing Operating System was built completely in-house and has evolved over time as the company grew. It currently supports nearly all of the company’s manufacturing equipment. The custom-built operating system has allowed Tesla to fine-tune its equipment and processes."​

Impressive: I'm not aware of any car factory in the world with a unified control OS for its equipment.

Most European car manufacturers haven't even managed to unify the car's own platform/OS - let alone the factory equipment which is orders of magnitude more complex and diverse...

And to get a feeling for the capex and installation complexity of a car factory:

"Tesla told us that the Model 3 body line is 90% automated and has over 1,000 robots"​

This is how Tesla now has better panel placement quality than most of the German competitors:

"As the Model 3 bodies are assembled, a team of 47 robots are hard at work at a dozen inline scanning stations that measure 1,900 points in every auto body to ensure that the cars coming off the line meet Tesla’s exacting standards. Take that, panel gaps."​

Impressive! Reading the article is well worth the time, IMHO.

 

[NicoV]

At current factory production levels Tesla would have to bring the equivalent of around 7 factories online and up to full speed. I hope alien dreadnought kicks in before that.

Or let the Chinese build it.

 

[EVNow]

We now see FSD computer picking up street lights which was not present 7 months ago.

Interestingly, traffic sign recognition was one of the first competitions to be solved using NN in 2011.

https://www.sciencedirect.com/science/article/pii/S0893608012000524

From wiki,

Ciresan and colleagues won pattern recognition contests, including the IJCNN 2011 Traffic Sign Recognition Competition,[40] the ISBI 2012 Segmentation of Neuronal Structures in Electron Microscopy Stacks challenge[41] and others. Their neural networks were the first pattern recognizers to achieve human-competitive or even superhuman performance[42] on benchmarks such as traffic sign recognition (IJCNN 2012), or the MNIST handwritten digits problem.

 

[AndyH]

Stupid question: where will this monthly registrations data be posted?

The page bilimp.dk has registration numbers. Jan-Feb looks suspiciously low. The politicians have been fooling around with the subsidizes recently which probably has affected Tesla's presence in Denmark. But I've seen a few really positive reviews of the Model 3 in Danish and subsidizes are good now so sales are bound to improve.

Article from April 2018 about earlier failure of sales because of removed subsidizes:
Bloomberg - Are you a robot?

 

[ZsoZso]

That is possible if the ICE industry collapses because of a recession. Tesla just can't scale up that fast, otherwise.

If the current car volumes don't change much, I expect Tesla to be producing about 10M a year by 3035 and #1.

If the market contracts so that Tesla will be #1 with 5M/yr volume, that can happen earlier - though it would still be 10x the current volume and thus will need 10 more giga factories to be built … in 10 years.

Wow, thats both very optimistic and very pessimistic view:

. Its optimistic that you think Tesla will be in business and still #1 in over a hundred thousand years

. Its pessimistic that you think it will take them more than 1000 years to scale to 10M production per year.

edit: oops thousand, not hundred!

 

[SebastianR]

Really good exclusive article just got released about a Fremont factory visit CleanTechnica performed this month, with a lot of new technical details:

A Look Inside Tesla's Fremont Automotive Factory — #CleanTechnica Exclusive | CleanTechnica

"Tesla’s team installed and commissioned the Model 3 servo press in ~25% less time than the fastest installation Schuler had ever done before. And that was just the starting point."

"Changes like this and more allowed Tesla to increase the rate of the press from 12 strokes per minute to 14 strokes per minute, a 16% improvement. That translates to more throughput from an extremely expensive piece of equipment and less capital required to scale up production. Putting that into context, Tesla can now produce body panels and stampings from the Schuler press at twice the rate as its Model X stamping press line and a mind boggling four times the rate of the older Model S production lines."

"Tesla’s Manufacturing Operating System was built completely in-house and has evolved over time as the company grew. It currently supports nearly all of the company’s manufacturing equipment. The custom-built operating system has allowed Tesla to fine-tune its equipment and processes."​

Impressive: I'm not aware of any car factory in the world with a unified control OS for its equipment.

Most European car manufacturers haven't even managed to unify the car's own platform/OS - let alone the factory equipment which orders of magnitude more complex and diverse...

This is what that VW/Amazon partnership is about: trying to get a handle on their plants and their manufacturing Volkswagen to team up with Amazon in drive to boost productivity | DW | 27.03.2019

After all these years of FUD around "if only the big guys wanted to, they can out-produce Tesla no problem" seems to be even less true than I thought.

 

[KarenRei]

We are actually AI created by neurolink to defend Tesla. Some work better than others. Factchecking is actually an early version of AI because we are more human like mimicking average intelligence. Factchecking however is blowing our cover because the poster is obviously not human.

 

[Lessmog]

Wow, thats both very optimistic and very pessimistic view:

. Its optimistic that you think Tesla will be in business and still #1 in over a hundred years

. Its pessimistic that you think it will take them more than 100 years to scale to 10M production per year.

Actually, more than 1,000 years to 3035 ...

 

[wipster]

If we take the assumption that 50% growth rate were to be maintained, there would be a year where they grow from 6.7m production to 10m. At current factory production levels Tesla would have to bring the equivalent of around 7 factories online and up to full speed. I hope alien dreadnought kicks in before that.

You mean the GF that builds the GF? Heck, they may be modular by then...

 

[Fact Checking]

Interestingly, traffic sign recognition was one of the first competitions to be solved using NN in 2011.

Note that while traffic signs are standardized, traffic lights are surprisingly difficult: they are relatively small, can be next to the sun, change color and come in very diverse forms:

Traffic light recognition quality and reliability will be the bellwether of Tesla FSD progress, IMO.

 

[EVNow]

Not due to a recession; due to the sales suppression effect of people deciding that their next vehicle will be an EV.

Also, I'd never underestimate Tesla's scaleup potential. Remember, for example, the reason for the Maxwell deal - to shrink electrode material lines from ~20m long vacuum ovens with solvent recovery systems, to ~2m long hot presses

People deciding their next vehicle is an EV and waiting will affect 1 to 2% of sales currently. A recession can kill 50% of sales.

I'm not even considering battery issues. Just normal car assembly & paint. Most big factories are about 500k in capacity. The biggest in US is about 700k (Nissan's TN plant).

The path of least resistance (and also most likely) is a recession along with high oil prices as in 2008. This resulted in huge dip in car volume - as well as a spike in efficient car (Prius) sales. We can see something like that again. Tesla can pickup some existing plants for a song (may be Emo G) and as the economy picks up, people start buying EVs instead of ICE.

 

[EVNow]

Or let the Chinese build it.

Will EU folks buy cars made in China ? In US - the answer is "no", so far. EVs are themselves a hard sell, let alone Chinese made EVs.

I think the easiest path would be to buy up some closed plants and retrofitting, which should be much faster than a greenfield plant.

 

[Fact Checking]

People deciding their next vehicle is an EV and waiting will affect 1 to 2% of sales currently.

I think deferred purchases will quickly reach a much higher percentage than that, according to a recent U.S. survey:

Poll Finds Americans Feel Electric Vehicles Are The Future of Driving

'Three in four drivers even say that “EVs are the future of driving (74%).” Among EV drivers, this number rises to 79%.'​

Deferred purchases might be even more probable in the premium and luxury cars market, where buyers don't want to buy and be seen using obsolete, unfashionable products.

BlackBerry revenue cratered quickly, it fell to half of its peak value in ~2011 within two years:

BlackBerry Revenue 2006-2018 | BB​

Two years is roughly the max life cycle of a premium business phone - the luxury car equivalent is around 5 years.

Note how the peak in 2011 was ~4 years after the introduction of the iPhone: inertia and brand value, but the rot started in 2007.

Next ~5 years are going to be painful to BMW and Audi.

 

[KarenRei]

People deciding their next vehicle is an EV and waiting will affect 1 to 2% of sales currently.

If I remember the last poll correctly, the actual number for the US is around 20%. And that number will only grow, fast. The "my next car will be an EV" population is always significantly larger / further along the S curve than the "I'm buying an EV today" population.

That said, a recession certainly can help create a perfect storm for traditional automakers. Most have way too much debt, and the value of their assets is highly risky.

 

[Fact Checking]

If I remember the last poll correctly, the actual number for the US is around 20%. And that number will only grow, fast. The "my next car will be an EV" population is always significantly larger / further along the S curve than the "I'm buying an EV today" population.

That said, a recession certainly can help create a perfect storm for traditional automakers. Most have way too much debt, and the value of their assets is highly risky.

Yes:

AAA: 1-in-5 U.S. Drivers Want an Electric Vehicle | AAA NewsRoom

"A new AAA survey shows that 20 percent or 50 million Americans will likely go electric for their next vehicle purchase, up from 15 percent in 2017."​

33% YoY increase in 2018, those numbers are probably above 25% now - but I'd not be surprised about 30%+.

New poll results from 2019 expected in ~1.5 months.

 

[EVNow]

If I remember the last poll correctly, the actual number for the US is around 20%. And that number will only grow, fast. The "my next car will be an EV" population is always significantly larger / further along the S curve than the "I'm buying an EV today" population.

That said, a recession certainly can help create a perfect storm for traditional automakers. Most have way too much debt, and the value of their assets is highly risky.

I'm talking about actually wanting to buy an EV & waiting i.e. they would have normally bought a car this year (because their car has become old) but are waiting to get an EV. That is a small number. Like we have been postponing replacing wife's aging CUV waiting for an ideal EV for past 7 years. She'll pick up Y in 2 years - we would have waited for 9 (starting with C-Max Energi in 2012) !

 

[BioSehnsucht]

So, are we all going Chinese? Cant be worse than American system and don't get me started on the Canadian confusion....

It's not so much the Chinese way as they appear to be the only ones sane enough (or perhaps, simply their government is strong enough for force the issue despite whatever they might have used prior) to standardize on it.

ISO 8601 is the sanest method for dates because you can do a simple left to right sort and they will sort correctly. To sort any other format, you have to separate out the year, month, date, etc from wherever they are in the order of numbers, order them something like ISO 8601 to sort them (or group and sort many times) then convert back.

Or you can just use ISO 8601 like a sane person. Not only is it easier for computers, it's easier for humans too. Left to right is most to least significant. It also makes it less likely for people from different countries to confuse March 8th for August 3rd or vice versa, not to mention the potential for confusion if you're still using 2 digit years.

 

[EVNow]

Deferred purchases might be even more probable in the premium and luxury cars market, where buyers don't want to buy and be seen using obsolete, unfashionable products.

BlackBerry revenue cratered quickly, it fell to half of its peak value in ~2011 within two years:

BlackBerry Revenue 2006-2018 | BB
Two years is roughly the max life cycle of a premium business phone - the luxury car equivalent is around 5 years.

Yes, in premium cars the turn around could be smaller - but people don't wait.

Basically, f you are leasing, you just lease when you need to (when the old lease ends, you need to do something - you can't just wait) but it is only 3 years, instead of 5+ if you buy.