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

[jw934]

I got the FSD option, NOT for free, four years ago. I don’t suppose we should be optimistic enough to think that maybe, possibly,......

Naah. He didn’t say WHEN those three months would be.

For those that don't know what this is referring to, Elon tweeted at 5:16PM today: "All Tesla cars delivered in the final three days of the year will get three months of the Full Self-Driving option for free."

https://twitter.com/elonmusk/status/1344044584712728581?s=20

 

[Prunesquallor]

For those that don't know what this is referring to, Elon tweeted at 5:16PM today: "All Tesla cars delivered in the final three days of the year will get three months of the Full Self-Driving option for free."

https://twitter.com/elonmusk/status/1344044584712728581?s=20

And already on Twitter, people who got their cars a few days ago are b****ing and calling up Tesla. I’m with @Krugerrand - people are *sugar*.

 

[heltok]

Imo it would make sense to give ALL(capable) Tesla cars 1 month of FSD for free. Should increase take rate and would cost the company very little.

 

[wipster]

It wasn't so long ago that everyone received 2 weeks of EAP free to test it out.

Ugh, hard pass on that. I can do roller coasters and boats just fine, but stick me in something that spins and it's barf city. VR works fine as long as it doesn't move in weird ways.

But, and I realize this is OT, wouldn't it be cool if home videos that were taken today or even old films could be converted into VR (not that difficult with today's computing power, just look at 3D in Google Earth) and you could view things that have been filmed as if you were in the video? And going beyond that, the computer had the capacity to "learn" and recreate your friends or relatives so you could talk with and to them?

I know this is far out, but give it 10 years and it will be possible. Hope we can survive that long as a species...

 

[Featsbeyond50]

Exactly. One could make the argument if the cars came standard with a free demo of FSD they’d sell more FSD, so it’s almost less of an incentive and more of a sneak attack.

True, but offering demo FSD also comes across a little like a greaseball sales move. Making it look like your just trying to make year-end numbers, on the other hand, looks a lot more virtuous.

 

[StarFoxisDown!]

And already on Twitter, people who got their cars a few days ago are b****ing and calling up Tesla. I’m with @Krugerrand - people are *sugar*.

Man I don't get people sometimes. It's a 3 month FSD trial. It's not like these people are getting FSD permanently for free.

 

[MP3Mike]

Man I don't get people sometimes. It's a 3 month FSD trial. It's not like these people are getting FSD permanently for free.

Yeah, I estimate that this offer is valued at ~$750. While the one year of free Supercharging that people have been getting up to this point is more like ~$300 for most people. (Way more for some people and $0 for others.)

 

[heltok]

InsideEVs - 1.5 hours ago: Tesla Model Y Spotted Driving With Strange Sensor Rack Up Top

Excerpt:

These videos and images raise a whole lot of questions. What is this rack? Will Tesla's Full Self-Driving vehicles have additional equipment and sensors? Will Tesla actually use LiIDAR? Is this specific to future Robotaxis?

While it could be anything really, and may certainly be related to Tesla's Full Self-Driving capability, it could be as simple as a mapping or imaging excursion. However, it looks like there's LiDAR on the rack.

As people have said, this is likely for being ground truth. But let’s clarify this a bit. Currently Tesla are relying on neural networks, these are trained using pairs of input-output data. Input is camera images, output is BEV(birds eye view) grids etc. These output labels are created by humans aided by computer vision, ie the point cloud that karpathy showed in Autonomy day.

My guess is that Tesla is using this Lidar data to create a test set to see how well the vision system is working and to add tests to their unit test set. After training on data from camera from normal cars, they make sure that the neural network, using camera data from the modified Y will output what the Lidar in the modified Y is saying it should predict.

Maybe it will turn out that the Lidar is measuring the distance to the vehicle in front to be 100m, while the neural network is predicting 90m. Then they will know this and can try to find ways to improve the neural network, for example by gathering more data, cleaning up old data, tweaking the network, finding a bug in the code etc. Maybe a few months later it now predicts 95m and they can say that their algorithm is getting better. So the ground truth is not used for training, it is used to test how good the trained system is.

 

[strago13]

As people have said, this is likely for being ground truth. But let’s clarify this a bit. Currently Tesla are relying on neural networks, these are trained using pairs of input-output data. Input is camera images, output is BEV(birds eye view) grids etc. These output labels are created by humans aided by computer vision, ie the point cloud that karpathy showed in Autonomy day.

My guess is that Tesla is using this Lidar data to create a test set to see how well the vision system is working and to add tests to their unit test set. After training on data from camera from normal cars, they make sure that the neural network, using camera data from the modified Y will output what the Lidar in the modified Y is saying it should predict.

Maybe it will turn out that the Lidar is measuring the distance to the vehicle in front to be 100m, while the neural network is predicting 90m. Then they will know this and can try to find ways to improve the neural network, for example by gathering more data, cleaning up old data, tweaking the network, finding a bug in the code etc. Maybe a few months later it now predicts 95m and they can say that their algorithm is getting better. So the ground truth is not used for training, it is used to test how good the trained system is.

As a software engineer, 100% agree with this assessment.

 

[StealthP3D]

I like that Gordo included "our opinion" with every ridiculous assertion. It's as if he's halfway admitting that "none of these are facts and I might just turn out to be a dumbass." I see that as a sign of progress. Way to go, Gordo!

I took it to mean he might get sued for damages without the disclaimers. But that's a self-aggrandizing delusion because there would only be damages if people actually thought he was credible.

 

[heltok]

This is a good point, but the contract is for the supply of battery grade lithium hydroxide - LiOH, which is ~ 29% Lithium by mass (7/24).

$630-880 million lithium hydroxide at current spot price of $7,600/ton would yield 100,000 tons of LiOH. 29% of 100 million kg is 29 million kg of Li. At 10 kg/car that yields 2.9 million cars, or 600,000 car batteries per year. With such a big contract, Tesla must be getting more Li/$, so this is definitely a lower limit.

Lithium price spotlight - Weekly price updates | Fastmarkets

Thanks for doing the math!

I would assume that most of this if for the Chinese market. So Tesla aims to produce their own batteries in China for ~3M cars in the next 5 years. Some would go to storage ofc. But also they will buy a lot from current suppliers(LG, CATL, Panasonic etc) during the same time period.

Let’s assume that in 2021 Tesla starts building a battery factory in China. It is starts to make batteries early 2022 and is close to fully ramped in the end of 2022. 2023-2025 they use most of this lithium. Then we are talking batteries for 1M cars/year for these three years plus whatever they are getting from suppliers.

Currently they have capacity for production of ~550k vehicles per year which is covered by suppliers. And suppliers are ramping.

My guess is that Tesla is gonna Tesla Energy(storage) production in China greatly, add Semi production in China and/or start producing Model 2 made for China with production start early 2022. Otherwise this large investment doesn’t make much sense...

 

[StealthP3D]

Cause he's a clown funded by companies that want to see Tesla fail...he may have gotten away with it in the first year or two..but it is so blatantly obvious now he is funded to spread this FUD.

He would probably gain more exposure and more invites on mainstream media if he put an Alka-Seltzer tablet under his tongue before each appearance. That way, when he went on one of his mandatory anti-Tesla angry rants, there would actually be foam coming out of the sides of his mouth. He would be more convincing too.

 

[CHGolferJim]

Well I hope TSLA has not become boring with the S & P addition.

ARKK was way more volatile than TSLA today. Down more than 4%. Unusual for ARKK to move more than TSLA.

Yes, what’s up with that? ARKG (genomics) also down similarly. Some kind of after-effect of the settlement?

 

[ammulder]

Do you have a link?

No. It's an article called "Minding the Gap" on page 26 of the December 2020 issue of Car and Driver. I can't find it on their Web site. I don't feel like I should post a pic of the whole article, but here's a snippet with the Tesla related bit (see text on the top right, Model 3 measurements on the middle left, and the top couple automaker rankings on the lower left):

 

[Linus_2]

I wonder if other had a similar situation.

I placed an sell order in the morning of Friday 18 at $694 but the order was never executed, closing price that day was $695. I asked my bank (Societe Generale) but they have not got back to me yet.
How is it possible that an order below the closing price is not executed when there is a such big imbalance?

I have been building a Tesla position since 2015, only sell some when the stars are aligned to buy back after, just to stay fit mentally.
Since it is my first post, I like to thanks TMC, the quantity/quality of posts are really helpful.

 

[ggr]

Ah, those that shall not be written because of the fear of MOD(s). I'm aware, but they shant be generated from my keyboard ;-)

The closing price proves that the market hates numerology as much as the moderators do.

 

[dakh]

I wonder if other had a similar situation.

I placed an sell order in the morning of Friday 18 at $694 but the order was never executed, closing price that day was $695. I asked my bank (Societe Generale) but they have not got back to me yet.
How is it possible that an order below the closing price is not executed when there is a such big imbalance?

I have been building a Tesla position since 2015, only sell some when the stars are aligned to buy back after, just to stay fit mentally.
Since it is my first post, I like to thanks TMC, the quantity/quality of posts are really helpful.

That would make no sense, the imbalance was towards lack of sellers so if you wanted to sell you would definitely have had your order filled. I think you need to place either an "at close" (limit or not) or "good till cancelled" order to get into the closing cross though.

 

[wipster]

The closing price proves that the market hates numerology as much as the moderators do.

I'll give you 2 funnies for that, but you'll have to do the second one...

 

[dakh]

So unless there's some advanced bluffing involved, we know they haven't gotten to 500K yet and it's a tight race. Either way would be interesting to see the price action once the numbers are out. I have a feeling that unless it's a substantial beat, it'll be the usual sell the news deal, just the question of how much sell... but this time there's the possible support from the fund managers looking to get in at a discount.

 

[PeterJA]

I do some things in VR and I've seen many different people in VR headsets. Best way to make people to throw up is to put VR headsets on while driving them. Some rollercoasters do that with a great success (by this I mean people throw up a lot).

Motion sickness is usually caused by incongruence between the body's perception of motion and the eye's perception of non-motion, such as getting tossed around in a ship's cabin with no windows to look out.
What causes seasickness?

Virtual Reality that turns objects outside your car into dinosaurs should be fine, as long as the VR motion is congruent with the car's motion. In fact, it might reduce car sickness if the VR makes your car seem transparent, as if you are riding on a sled.