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

[Krugerrand]

Cost reductions often come from energy density improvements. It costs roughly the same to make a high-energy cell, but you need fewer of them in a pack. So count on both energy density changes and manufacturing efficiencies to be discussed. I sharply increased my Tesla holdings after CyberTruck was announced, the pricing of which implied breakthrough battery costs.

Maybe. Maybe not.

There definitely will be cost breakthroughs in a) not building dies sets, adding a new press line and all the costs including time and manpower associated with that, b) reducing build and assembly time and costs, and c) reducing time and costs associated with paint.

Possibly a cost breakthrough in producing their own material, the logistics and time associated with that.

 

[mrdoubleb]

For context, Exxon-Mobile 2019 revenues: $275 billion. Acceptable business expenses?

Exxon's profits decreased by 50% YoY in Q3 (they report Q4 today) to 3.1Bn. I think they have better ways of spending their money than shorting Tesla.
I think there are enough idiots out there who think shorting Tesla is a good move without Big Oil having to foot the bill so directly.

 

[Derek Vinyard]

GF4 (Germany) - project could be delayed by 9 months

there is always some food for the shorties somewhere outside, indeed...

Under German environmental regulations, the project in the town of Gruenheide could be delayed by nine months unless work begins before the breeding period for local wildlife this spring, Joerg Steinbach said in an interview with the Handelsblatt newspaper published Friday./QUOTE]

Bloomberg - Are you a robot?

 

[LJS22]

Cramer on his redemption tour just blasted oil and gas stocks. Basically said everyone is devesting from owning these, managers are tired of being told what is right or wrong with the world and hedge funds are putting their money into clean energy. Called them tobacco stocks and said it’s the other side of Tesla, young people are betting on Tesla and have denounced oil and gas it’s happening faster than you think, this is the nail going into the coffin.

Becky on Squawk Box was shocked that he’d say that. She said “well you mean the stocks are dead, we will be on natural gas for a while” and Cramer just said “eh it’s happening faster than you think”

 

[elasalle]

Gang, I had a bright idea today, which I posted in the Options Strategy thread, bit given that most of you would not venture there I post a link and reproduce below (please answer in the thread though if you have advice): Options trading strategy/advice

Thanks to options, looks like I'm likely to get around $100k to play with in my trading account, slightly up from the $3k I started with

I was thinking how to grow this quickly and came to the following strategy:
- every week - likely Friday, sell a covered put (just 1x100) slightly below the trading range for the following week = pocket the premium
- using Opricot as a guide, most of the time the option will expire worth less
- if it is exercised, I buy the stock, then sell a covered call with similar parameters = pocket the premium
- when exercised, switch back to put

Seems foolproof to me? I cannot lose.

Convince me I'm wrong...

If Foolproof, everybody would be doing it, and they wouldn't be any more Fools around. cheers.
Every Strategy works until it doesn't

However, selling PUTS is one way to diversify your option strategy.

Success is all dependent on how correct/lucky you are in guestimating Market antics.

 

[ReflexFunds]

Is there a new podcast or did I miss this in the first one ?

E2E deep learning is ofcourse the ultimate objective. Ofcourse this statement by Musk raises more questions.
- When did they start the re-write ? Is it the rumored/infamous AK-9 ?
- Is this what DoJo was supposed to do ? Musk also talked about video labeling in ER ...
- If paint the path & label in 3D is the next part, what exactly is the "planning" that is almost complete ?

It’s not so clear from Elon’s comments if the foundational re-write that’s nearly complete includes the full 8 camera video end to end neural nets, or if this will be a later step once they have got Project Dojo up and running. I’m not sure Dojo is necessarily required for this - its just going to make it much quicker and cheaper and allow faster progress.
It could be that the foundational re-write is just a step along the path towards these end to end video neural networks. Perhaps its more similar to the AKnet neural net, using several labelled frames from 8 cameras in one neural net, and using this multi task network to also do path planning, but not using full video and full 3D path modelling.

My interpretation of this is maybe a bit less rosy than some here. They're learning how to do this thing. Training many models for specific tasks with their respective curated training sets is not exactly "vacation mode". They now have what probably can be described as "driving situations" as training sets that include video feeds from all cameras (and probably radar) and they simply train for desired outcomes. That's why he's talking about AP being able to drive like it's a high speed chase. It'll be able to do stuff that even highly trained humans won't do, and nobody will be teaching it how, it'll learn by example and then by trial (probably in a sim first, then in some controlled reality). Watch this video knowing that nobody taught this thing how to do what it does. It was just given the goal and gobs of simulated learning time. That didn't work well enough so they had to give it some more simulated time with modified environments like different gravity or tied up fingers. And then it started reliably working. Solving Rubik’s Cube with a Robot Hand

So yes it definitely is a substantial breakthrough, they got the hardware and a system that can learn how to drive instead of a bunch of little individual subsystems that learn small tasks and then you still have to make them all work together to solve the overall problem. That doesn't mean they're almost done. That means they built a much better tool, now they still have to do the training, they still have to make up training datasets etc. How much of what they had for training datasets before can they re-use in the new system? How much more will they need to get some minimally working FSD? Is the current hardware powerful enough to run what is needed without being rat-holed into some crazy optimizations to get acceptable performance? All that remains to be seen.

Tesla have been collecting full video and sensor data from the fleet for years for specific driving situations, i’m not sure if they also collected driver actions (steering, acceleration etc) but I’d be surprised if they didn't. I’m sure much of this old data can be relabelled on a 3D video basis rather than on a frame basis, but its unclear when they started this. I wouldn’t be surprised if they’d been working on infrastructure towards this behind the scenes since soon after Karpathy joined though - AKnet first surfaced in early 2018.

I would expect to see video collection across the fleet to step up now. I expect they use the fleet to identify a list of priority problems or edge cases - using shadow mode deviations and Autopilot disengagements to identify situations where Autopilot most often fails. From this they can launch campaigns to get the fleet to identify situations similar to these until they have enough data to train and solve the problem.

The most exciting part will be when they get accuracy so high and disengagement rates so low that they can ask the fleet to send full sensor info, driver actions and video every single time a car deviates from shadow mode or for every disengagement.

I do think it will take sometime to continue to optimise the new software base though.

 

[BornToFly]

Gang, I had a bright idea today, which I posted in the Options Strategy thread, bit given that most of you would not venture there I post a link and reproduce below (please answer in the thread though if you have advice): Options trading strategy/advice

Thanks to options, looks like I'm likely to get around $100k to play with in my trading account, slightly up from the $3k I started with

I was thinking how to grow this quickly and came to the following strategy:
- every week - likely Friday, sell a covered put (just 1x100) slightly below the trading range for the following week = pocket the premium
- using Opricot as a guide, most of the time the option will expire worth less
- if it is exercised, I buy the stock, then sell a covered call with similar parameters = pocket the premium
- when exercised, switch back to put

Seems foolproof to me? I cannot lose.

Convince me I'm wrong...

I nearly lost it all selling Puts last year. My OTM 320 SP Puts went into the money and got assigned before the expiration date and before I could roll them. I thought no problem, I'll sell covered calls. The SP kept dropping so selling covered calls for the stock price I paid became worthless, and I didn't want to sell the stock at a big loss. Ended up with margin calls every week. Your plan works if the SP doesn't drop quickly, which is definitely NOT guaranteed with TSLA. Be careful.

 

[Derek Vinyard]

here's the BBG article - full version

Tesla’s German Plant Faces Deadline Over Breeding Wildlife
Stefan Nicola

The forest near the proposed location for a Tesla Inc. Gigafactory, in Gruenheide, Germany.Photographer: Alex Kraus/Bloomberg
Sign up to our Next Africa newsletter and follow Bloomberg Africa on Twitter

Tesla Inc.’s plans to build a factory outside Berlin could be under threat if construction work doesn’t begin by mid-March, according to the economy minister for the Brandenburg region where the site is located.

Under German environmental regulations, the project in the town of Gruenheide could be delayed by nine months unless work begins before the breeding period for local wildlife this spring, Joerg Steinbach said in an interview with the Handelsblatt newspaper published Friday.

“That would be a situation in which I would be much more skeptical about whether we could get Tesla to stick with it,” Steinbach said, adding that the U.S. manufacturer needs to have a convincing proposal to meet local environmental demands and gain approval for the project.

Tesla Chief Executive Officer Elon Musk recently tried to ease local concerns about water usage for the plant, which is located in a forest bordering on a nature reserve. That’s one of the bureaucratic challenges for the proposed plant, which represents the electric-car maker’s most daring attack on the German auto establishment.

An excavator and a tractor drawn road roller sit by the site of the Tesla Inc. Gigafactory in Gruenheide.

Photographer: Krisztian Bocsi/Bloomberg

If the project clears Germany’s red tape and secures funding, the site could churn out as many as 500,000 cars a year, employ 12,000 people and pose a serious challenge to the likes of Volkswagen AG, Daimler AG and BMW AG.

Read more: Tesla Descends on German Town in Attack on Auto Establishment

Workers have already combed through about 150 soccer fields of forest and removed most of the World War II ammunition found there, according to Gruenheide’s mayor Arne Christiani. If Tesla files proper paperwork in the coming days, then logging could start in mid-February -- enough time to beat the breeding deadline, Christiani said on Friday.

“There are rules that everyone has to follow,” Christiani said in a phone interview. “But we’re still optimistic and are very much within our desired time plan.”

Tesla still has to jump through a number of hoops, including scaring off or relocating wolves, hibernating bats as well as snakes and lizards until construction is over. Residents still have the chance to raise objections, and some have bemoaned that they’ve seen little from Tesla since its blockbuster announcement.

To quell some of the more immediate concerns, the U.S. carmaker earlier this month dispatched consultants to an office in the center of Gruenheide to answer questions about the project.

Tesla is working closely with environmental groups to minimize the damage to wildlife, Christiani said, adding that he’s not surprised over some of the resistance to the project.

“There would have been opposition even if we had built a chocolate factory on the site,” he said.

(Updates with comments from Gruenheide’s mayor from fifth paragraph.)
/QUOTE]

 

[Fobble]

Wall Street Investor vs Retail InvestorView attachment 506297

The only thing they agree on is wearing black framed glasses ....

 

[davepsilon]

So one future negative to think about is with the acknowledged delay in FSD schedule and more generally in the large uncertainty in timing what is Tesla going to do with their leases that have no customer buyout option? The “turn them into Robotaxi” always seemed like too much of a timing bet to me anyway. And Tesla has a factory if they need Robotaxis just divert some of the new production, complete with Robotaxi specific tweaks, or even a dedicated design.

So it’s probably no change to their internal plan which I suppose was always to sell the lease returns as used cars. This could even be a positive for financials, if value is higher than what they used for original lease residual value then Tesla can keep the difference. Used values are currently sky high but should come down further along the s-curve of Model 3 output.

Tesla’s online sales method for new cars is great. For used cars they could improve it. I hope they do before being flooded with used lease returns to sell.

 

[Artful Dodger]

What’s the white circle in the center?

Utopia Planitia

 

[SoGA Fan Club]

I see Spiegel has gone from constantly tweeting (retweeting) random Tesla/Musk fraud conspiracies to constantly posting coronavirus conspiracies. He's convinced that it is much worse than China is telling and that they are burning bodies to cover it up. I think he has given up on all the Tesla conspiracies and is now praying this virus will do to Tesla what his lies/FUD couldn't. I think more importantly, he hopes he can at least convince people the situation is dire - whether it is or not.

What is up with people who see these weird conspiracies and negativity in EVERYTHING? Is is just a personality? Or is it a mental illness? I used to work with a guy JUST like Spiegel. He was constantly warning me of the impending financial collapse and telling me I needed to be stocking up on survival items like dried food and packaged meals. To this day, he is 100% convinced that it's not a matter of if, but when. And, every day, it's surely closer to being almost here. I can't imagine living in that kind of fear.

 

[mrdoubleb]

GF4 (Germany) - project could be delayed by 9 months

there is always some food for the shorties somewhere outside, indeed...

I don't think this is new info in itself - though the interview may be new. However, as Tesla has already started to cut the trees proactively, this may be a moot point. Birds are going to stay away from the sound of falling trees and the field should be cleared by the groundbreaking ceremony in early March. Plus, it seems like protests have ended and environmental groups seem to be more or less happy with Tesla's answers to their questions.

Things may change of course, but if there was a 9 month delay, I don't think it is out of the question for Musk to skip to a back-up site. He did say their deal is contingent on speedy progress.

 

[madodel]

Cramer on his redemption tour just blasted oil and gas stocks. Basically said everyone is devesting from owning these, managers are tired of being told what is right or wrong with the world and hedge funds are putting their money into clean energy. Called them tobacco stocks and said it’s the other side of Tesla, young people are betting on Tesla and have denounced oil and gas it’s happening faster than you think, this is the nail going into the coffin.

Becky on Squawk Box was shocked that he’d say that. She said “well you mean the stocks are dead, we will be on natural gas for a while” and Cramer just said “eh it’s happening faster than you think”

He is now a true believer. He's talking about Elon Musk's vision. And the talking heads aren't attacking. I think we have hit a true turning point. He also revealed he uses Sensodyne toothpaste. Very informative show today.

 

[dl003]

Exxon's profits decreased by 50% YoY in Q3 (they report Q4 today) to 3.1Bn. I think they have better ways of spending their money than shorting Tesla.
I think there are enough idiots out there who think shorting Tesla is a good move without Big Oil having to foot the bill so directly.

There're more of them than we think. The bull market has made many idiots think they're good traders just because their swing trades generally gains. I saved this gif for every time some random Joe has the bright idea of shorting

 

[dw4ngg]

Gang, I had a bright idea today, which I posted in the Options Strategy thread, bit given that most of you would not venture there I post a link and reproduce below (please answer in the thread though if you have advice): Options trading strategy/advice

Thanks to options, looks like I'm likely to get around $100k to play with in my trading account, slightly up from the $3k I started with

I was thinking how to grow this quickly and came to the following strategy:
- every week - likely Friday, sell a covered put (just 1x100) slightly below the trading range for the following week = pocket the premium
- using Opricot as a guide, most of the time the option will expire worth less
- if it is exercised, I buy the stock, then sell a covered call with similar parameters = pocket the premium
- when exercised, switch back to put

Seems foolproof to me? I cannot lose.

Convince me I'm wrong...

Works if you don’t mind getting stuck w 100 shares

 

[SOULPEDL]

Waymo had an accident last night in Tempe Az (same city as the Uber accident actually... about 2 miles from my house). Of course the cause is unknown and not life threatening. Big question tho... Autopilot? Seems it's only on local news still... too early?

1 injured after Waymo van-involved crash in Tempe

At first glance, I thought the video showed a Tesla... just because I'm expecting something like this any day now.

 

[ReflexFunds]

So one future negative to think about is with the acknowledged delay in FSD schedule and more generally in the large uncertainty in timing what is Tesla going to do with their leases that have no customer buyout option? The “turn them into Robotaxi” always seemed like too much of a timing bet too many anyway. And Tesla has a factory if you need Robotaxi just divert some of the new production, complete with Robotaxi specific tweaks or even dedicated design.

So it’s probably no change to their internal plan which I suppose was always to sell the lease returns as used cars. This could even be a positive for financials, if value is higher than what they used for original lease residual value then Tesla can keep the difference.

Tesla’s online sales method for new cars is great. For used cars they could improve it. I hope they do before being flooded with used lease returns to sell.

I do think Model 3 residual values will be higher than used in their lease modelling even without FSD, so little to worry about on the lease returns. They can also be used in a pre Robotaxi human driven ride hailing network.

Where do you get an acknowledged delay in FSD schedule though?
Elon said the feature complete driver assist features are likely to be ready in a couple of months and that the foundational re-write of the Autopilot software is nearly complete.
How long it takes to get from feature complete to a level of accuracy/disengagements rates that can convince regulators to allow Robotaxis without driver monitoring is just as uncertain as ever.
I don't think there will necessarily be any Robotaxi design tweaks. Model 3 was designed from day one in 2015 to be a Robotaxi from the bottom up. Future cars may however move to the million mile battery which is primarily an optimisation for Robotaxis.

 

[mickificki]

John D. on Twitter

 

[canoemore]

I wonder how flexible Tesla's neural net training setup is.

Is it relatively constrained to solving the Full Self Driving problem? Or is it possible they could sell neural net training as a stand-alone service product that people could use for all sorts of applications?

I'm imagining something akin to Amazon Web Services, but for neural nets.