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

[ZeApelido]

• Creating the neural network topology and weights themselves is less than 10% of the work in creating FSD, IMO: the NN layers you can use are mainly a function of your input space and your hardware capabilities.
• The GoogleNet NNs in the latest firmware are probably just a placeholder - there's no trace of the 'real' v10 neural network.

The real deal is the training data and the process that creates and maintains the training data.

Copying the FSD neural network is a bit like stealing a Tesla by taking a video of it: yes, it's more than having no car at all, but it's not the real thing.

So you don't believe transfer learning is applicable to Tesla NNs?

 

[JRP3]

What is wrong with this stock? You can’t eat breakfast and take a long dump without it doing a 180 on you.

It hit 180? I missed out on that

For a look into the heart of darkness Stanphyl Capital put out their year end letter Long rambling section on TSLA of course, including links to what seems to be every article with the word "electric" in it, showcasing all the "competition"

https://seekingalpha.com/article/4232055-stanphyl-capital-letter-december-2018

 

[tinm]

What is wrong with this stock? You can’t eat breakfast and take a long dump without it doing a 180 on you.

Might wanna try something different for breakfast.

 

[bdy0627]

What is wrong with this stock? You can’t eat breakfast and take a long dump without it doing a 180 on you.

Maybe it's time to move your trading desk into the bathroom, you know, like the shorts.

 

[KarenRei]

Maybe it's time to move your trading desk into the bathroom, you know, like the shorts.

That picture will never cease being funny.

 

[ZeApelido]

Really detailed article on TSLA. Please take a look.























This article must have the record for amount of links in it, maybe 100? How could you possibly go against his advice?

 

[ZeApelido]

I am sooo late.

 

[Remus]

"Greater China" includes, for example Taiwan.

Also, China is a planned economy. They decide what industries they want, where, and build to those plans.

oversimplification. Chinese economy is heavily influenced by the government, by policies, financial support and sometimes even executive orders, comparing to developed countries such as U.S. and Germany. But it is nowhere near the level of former Soviet Union.

 

[KarenRei]

oversimplification. Chinese economy is heavily influenced by the government, by policies, financial support and sometimes even executive orders, comparing to developed countries such as U.S. and Germany. But it is nowhere near the level of former Soviet Union.

Nor did I say "it is the same as the Soviet Union". But it is still a planned economy. They designate specific cities or areas for specific industries - say, auto production - and build them to facilitate this. They build whole cities from scratch when they want to build new industries in a certain area.

Lingang New City in Shanghai - the outskirts of which are hosting GF3 - was a ghost town five years ago. Tons of construction, no people. Suddenly, by three years ago, it was populated and operational. By 2020 it's supposed to have half a million people. This is highly planned growth.

 

[Nocturnal]

Sold a few at $315 to move some money around a couple days ago. I figured I'd lose a bit on that. (well, still broke even on those shares) I'm still crazily overexposed so that's not a big deal.

Really don't feel like being at work today. Can TSLA just please pop to 4k a share so I can quit and go home?

 

[Fact Checking]

So you don't believe transfer learning is applicable to Tesla NNs?

Yes, transfer learning is not applicable to any sort of NN that is improving with a significant gradient - which FSD will do for the next 5-10 years.

Just take a look at how long it took for Tesla to reach feature parity with HW1 MobilEye (about 2 years!), and they had ready access to their neural networks, which were tailored to Tesla's sensors.

What Tesla didn't have access to is the hardware and (probably) the underlying training and development infrastructure - that had to be created from scratch.

Tesla's real "competitive moat" is the speed at which they are moving, never standing still in the quicksand:

 

[JustAaronnoraA]

335.46 Resistance is super strong today. Consolidations?

 

[Remus]

Nor did I say "it is the same as the Soviet Union". But it is still a planned economy. They designate specific cities for specific industries - say, auto production - and build them to facilitate this. They build whole cities from scratch when they want to build new industries in a certain area.

in the past twenty years things changed dramatically, private owned enterprises become the growth leader. I saw less and less government influences on them. For instance, the city of Hangzhou, home of Alibaba and Geely, was never planed to be a city of cars and IT. And the area between Hangzhou and Shanghai becomes a manufacturing hub without guidance given by the central government.

 

[Fact Checking]

in the past twenty years things changed dramatically, private owned enterprises become the growth leader. I saw less and less government influences on them. For instance, the city of Hangzhou, home of Alibaba and Geely, was never planed to be a city of cars and IT. And the area between Hangzhou and Shanghai becomes a manufacturing hub without guidance given by the central government.

The point @KarenRei is trying to make I think (and I made similar points previously) is that while China's economy has a fair amount of autonomy, there's a number of key differences:

• The Chinese government sets key control parameters WAY beyond that of any western economy (for example in China all banks are state-owned, so new credit/loans are 100% controlled by the government),
• Also, when the Chinese government explicitly supports you at every level, things start moving really, really fast.

Tesla's Shanghai Gigafactory project got various boosts from the Chinese government already:

• 1-2 billion dollars of loans got approved in record time at favorable conditions,
• Tesla had no counter-bids for the land it purchased,
• Tesla got an electric grid connection three times faster than the average,
• one of the largest state owned Chinese construction firms is building the Gigafactory,
  
• local government officials are making various public statements of support for Tesla's project, which has an effect on contractors and suppliers.

To state that the Shanghai Gigafactory is a "prestige project" both for the Shanghai government (fastest growing city-state of China) and the Beijing central government is probably the understatement of the year. Attracting Tesla to mainland China was a major coup and is a template for future high-tech investments in China, which is strategic to China on many levels.

 

[pz1975]

Just be careful to hedge your risk or position size for the day it breaks the bounds. IEf: mostly if that short squeeze shows up later this year. I like the strategy though...I'm looking into doing more day trading soon...I've been mostly doing months out swing call options.

Agree. I only sell a small number of puts/calls compared to my overall position. I also only sell calls at a level that I would fine selling the equivalent number of LEAPs (that I hold) at. As for the puts, I only sell at a level and amount that I would be fine buying the stock. This way even if there is a black/white swan event I am fine. If I end up red on either end, I almost always just roll out a new position 1-2 weeks away to get back to break even. I have been green 90-95% of weeks doing this for months now and even when red on a certain week, I always eventually get back to break even.

 

[alloverx]

Dont hear much about the Tesla network anymore.
Plus Tesla keeps changing their FSD strategy every 2 years, so is the miles/data from the prior versions really any value?

 

[Lessmog]

Off topic:

While Charles Mwangi says his new, interesting employer still is under the radar he does provide us with one enticing clue.

His good friend, former and now new colleague Shen Jackson is developing a process for physical vapor deposition of the semi-conductor copper indium gallium selenide onto glass, something surely applicable to the production of photovoltaic cells.

So I hope to hear more from these industrious gentlemen.

Very interesting! midsummer is a Swedish startup selling machines for producing PV cells using a sputtering technique to deposit i.a. CIGS onto various substrates (including flexible stainless steel). They also have small-scale production in-house. I got a few shares last year. In fact, I used to live within walking distance from their site, which is where IBM once brought forward the inkjets that spun-off as Lexmark. Maybe comes with the terroir, eh?

 

[riverFox]

The point @KarenRei is trying to make I think (and I made similar points previously) is that while China's economy has a fair amount of autonomy, there's a number of key differences:

• The Chinese government sets key control parameters WAY beyond that of any western economy (for example in China all banks are state-owned, so new credit/loans are 100% controlled by the government),
• Also, when the Chinese government explicitly supports you at every level, things start moving really, really fast.

Tesla's Shanghai Gigafactory project got various boosts from the Chinese government already:

• 1-2 billion dollars of loans got approved in record time at favorable conditions,
• Tesla had no counter-bids for the land it purchased,
• Tesla got an electric grid connection three times faster than the average,
• one of the largest state owned Chinese construction firms is building the Gigafactory,
  
• local government officials are making various public statements of support for Tesla's project, which has an effect on contractors and suppliers.

To state that the Shanghai Gigafactory is a "prestige project" both for the Shanghai government (fastest growing city-state of China) and the Beijing central government is probably the understatement of the year. Attracting Tesla to mainland China was a major coup and is a template for future high-tech investments in China, which is strategic to China on many levels.

I suspect China figures helping Tesla grow faster hurts Trump and his fossil fuel allies.

 

[Fact Checking]

Dont hear much about the Tesla network anymore.

I believe one of the reasons is that Tesla is so close to having functional FSD that they are hiding/downplaying their Tesla Network plans. Once HW3 based FSD is out a whole new era will begin.

The other reason is that Tesla doesn't depend on external Wall Street financing anymore, it has a strong cash flow that self-finances its growth, so it has no motivation to hype future plans - in fact it has an interest in flying under the radar as much as possible.

For example the Tesla AI chip (HW3) came as an (almost) complete surprise back a couple of months ago.

 

[AlexS]

I wish TSLA to report earnings before January monthly options expiration, so before 18th. The shorts will go crazy if report is any good as we expect. Otherwise next week can be big drag down and then these crazy analysts has more time to manipulate up their earning targets.

I think Elon kinda owes us that, for his 420 tweet.

But I have a feeling that it will happen week or two later.