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

[PhaseWhite]

To expand on this :

Avg crash rate : 4.2 MM - 99.9996%
6 9s, 4x better : 1.05 MM - 99.9999%
7 9s, 40x better : 0.105 MM - 99.99999%

If we assume Tesla is at 99% - and improving exponentially by one 9 every 6 months - they will be about human avg in 1 1/2 years. 4x better in 2 years. 40x better in 2 1/2 years

Achieving each additional 9 is 10x more difficult than the last, so I don't think it makes sense to assume a linear time scale like every 6 months. Getting from 99% to 99.9% is a 10X improvement or like going from 1 failure in 100 attempts to 1 failure in 1000 attempts.
Getting to 99.9999 is a 10,000X time improvement over 99%.

The only way I could believe in exponential progress like that is if the NN training pipeline was 100% automated based something like using fleet data for re-enforcement learning (I'm not a Machine Learning expert), but from the recent autonomy investor day we know they still rely on human labeling. That doesn't scale to 10,000X very easily so I think a new approach will be needed to get there.

As far as where they are today, I think we'll know when they get to 99% because I think they'd want to start releasing the "FSD" city driving features with a hands on requirement like NOA is today. That will enable them to start gathering the data they need to achieve those extra 9's

If you disagree, please make your case...

 

[anthonyj]

Buying the fake premarket dip tomorrow morning. We all know that TSLA does best when sentiment is good and the market is in peril

I hope it’s higher than 20k / month by now.

I started being concerned about demand because of the rate at which Tesla has pulled demand levers this year.

My concern grew when Gene Munster stated that orders have dropped off in the last 6 months. I don’t know what data has led him to believe that, but it would line up with the pulling of demand levers so quickly. Hopefully, if he’s right, it was from earlier and orders have now picked back up.

Gene Munster misspeaks a ton. I remember after Q4 deliveries were released in January, he called it a “huge, huge miss and disappointment” when they only missed by what, 1,000 cars? I don’t think he follows the company as closely as some do

 

[AlMc]

For investors: Things may change dramatically overnight but it may not be pretty on the stock exchanges in Europe and the US tomorrow: As of now. Shanghai Composite-3.54%
Hang Seng-2.83%
SP500 FUTS 2,892.38 -55.12 -1.87%

Since TSLA beats to a different drum many times no predictions from me about direction of TSLA tomorrow.....

 

[Electroman]

I have a bad feeling about this insurance thing.

Insurance works best when offered with a large pool, in the millions or tens of millions. Right now there are around 500K Tesla cars. And if you assume around 20% will take Tesla's insurance, that is only 100K cars. How can you offer an insurance product for such a small clientele ?

 

[azaz]

I have a bad feeling about this insurance thing.

Insurance works best when offered with a large pool, in the millions or tens of millions. Right now there are around 500K Tesla cars. And if you assume around 20% will take Tesla's insurance, that is only 100K cars. How can you offer an insurance product for such a small clientele ?

think it's important to understand they are partnering with someone

 

[Pras]

I have a bad feeling about this insurance thing.

Insurance works best when offered with a large pool, in the millions or tens of millions. Right now there are around 500K Tesla cars. And if you assume around 20% will take Tesla's insurance, that is only 100K cars. How can you offer an insurance product for such a small clientele ?

Pros: better data, lower insurance fraud, and incentive for higher FSD take rate. Also reduce cost by using Tesla body shop.
Cons: lack of underwriting experience, smaller pool will result in lumpy claims behavior.

The hope is that initially both pros and cons offset each other, and over time pros outweighs cons.

 

[Carl Raymond]

think it's important to understand they are partnering with someone

... and it’s going to be almost as simple as “Would you like insurance with that?” tick a box on the car order. If the premium is lowest available, why would anybody not tick? 100K will even out statistical aberrations - well enough.

 

[RobStark]

I have a bad feeling about this insurance thing.

Insurance works best when offered with a large pool, in the millions or tens of millions. Right now there are around 500K Tesla cars. And if you assume around 20% will take Tesla's insurance, that is only 100K cars. How can you offer an insurance product for such a small clientele ?

There is also reinsurance if risk is too high.

But Tesla should have superior data which allows them to price more accurately.

 

[kyzhk]

Tesla Model 3 saved me : teslamotors

According to the driver, he got rear ended when slowing down for an exit. Looks superhuman to me how quick the car swerve to the left after getting rear ended and avoided frontal damage.

At least in this case the superhuman reaction time and split second decision made is arguably better than the best human driver.

 

[SPadival]

Speaking of insurance, Tesla’s insurance policy against a Trump trade war with China until GF3 starts production

Twitter

 

[EVNow]

Achieving each additional 9 is 10x more difficult than the last, so I don't think it makes sense to assume a linear time scale like every 6 months. Getting from 99% to 99.9% is a 10X improvement or like going from 1 failure in 100 attempts to 1 failure in 1000 attempts.
Getting to 99.9999 is a 10,000X time improvement over 99%.

The only way I could believe in exponential progress like that is if the NN training pipeline was 100% automated based something like using fleet data for re-enforcement learning (I'm not a Machine Learning expert), but from the recent autonomy investor day we know they still rely on human labeling. That doesn't scale to 10,000X very easily so I think a new approach will be needed to get there.

As far as where they are today, I think we'll know when they get to 99% because I think they'd want to start releasing the "FSD" city driving features with a hands on requirement like NOA is today. That will enable them to start gathering the data they need to achieve those extra 9's

If you disagree, please make your case...

Pls check a few of my earlier posts on this. I started with your premise but looks like exponential improvement is possible under certain circumstances and assumptions.

 

[oneday]

Speaking of insurance, Tesla’s insurance policy against a Trump trade war with China until GF3 starts production

Twitter

Maybe Adam Jonas was right after all. If all those cars become self aware at the same time this poses a major national security risk.

 

[gene]

Tesla Model 3 saved me : teslamotors

According to the driver, he got rear ended when slowing down for an exit. Looks superhuman to me how quick the car swerve to the left after getting rear ended and avoided frontal damage.

At least in this case the superhuman reaction time and split second decision made is arguably better than the best human driver.

I have had a few experiences where the car took over properly. It made, in my opinion, accurate decisions. Last experience was on a city street where the car thought I was too close to a bicycle rider, as I passed the bike my Model 3 moved over a foot or two to the left and then brought me back to center of lane.--this was without AP.

A dear family friend on a bike, riding in a bike lane, was hit and killed by an H2 Hummer. Had it been a Tesla, the young man would still be alive today.

 

[omgwtfbyobbq]

I don't think that's what happened. The margin in Q4 was likely 22%+ and so they lost a couple in absolute terms, and also they had a 200M$ regulatory credit windfall, much much higher than usual per car. I don't think the cost of production moved much at all. You can skip thinking about revenue and just play with the cost numbers to see there's not much room for a lower 3 COGS without taking S/X COGS through the roof. The total mix COGS for the model 3 looks to be about 45k$ - 46k$. I think it might have had some extra upward pressure from international deliveries. I am expecting ASP to drop another 3k-4k$ this year and COGS to drop by less than that. Unless FSD take rate goes up (probably Tesla's theory when talking about 25%), or even more regulatory credits come into the picture (FCA?) I don't see how 25% GM is possible by any reasonable estimation. Shanghai will make cars for less, but also be selling at a low ASP.

I wouldn't discount Tesla's emissions credit sales prematurely. In 2017, Tesla earned $360.6 million from the sale of regulatory credits, and in 2018 they earned $418.6 million. 2017 deliveries were about 103k cars compared to 2018 deliveries at 245k cars, so Tesla may still have a credit surplus from 2018. Last but not least, even if Tesla doesn't increase production over the next few quarters, they will still produce ~310k cars in 2019. My guess is they'll produce around 380k cars, but only time will tell.

https://ir.tesla.com/node/19496/html
https://ir.tesla.com/news-releases/...sla-q4-2017-vehicle-production-and-deliveries
https://ir.tesla.com/node/19496/html

Additionally, we recognized $1.40 billion of additional automotive sales revenue due to the adoption of the new revenue standard and an increase of $58.3 million in sales of regulatory credits to $418.6 million in the year ended December 31, 2018. ZEV credits sales were $103.4 million and non-ZEV regulatory credits sales were $315.2 million in the year ended December 31, 2018, compared to $279.7 million ZEV credit sales and $80.6 million in non-ZEV regulatory credit sales in the year ended December 31, 2017. The growth in non-ZEV regulatory credits year-over-year was generally consistent with the delivery volume growth.

 

[PANN]

So far absolute zero report on Trump tariff threat in Chinese domestic media, must be the Chinese government, IMHO looks more like they are working out the differences than a break-down on trade deal, otherwise I should have seen tons of Trump bashing in Chinese domestic media.

 

[mrdoubleb]

Pros: better data, lower insurance fraud, and incentive for higher FSD take rate. Also reduce cost by using Tesla body shop.
Cons: lack of underwriting experience, smaller pool will result in lumpy claims behavior.

The hope is that initially both pros and cons offset each other, and over time pros outweighs cons.

Agreed. I've been thinking about the same thing and came to a similar conclusion.

First, to me this is like the Superchrager network: Tesla is doing it out of necessity. In my country the first quotes people received for Model 3 were 3x that of a BMW M3. As far as I've read on these forums you can have big swings in the US too in quotes people got.

Second: don't be surprised if you see some unique T&Cs and discounts in the Tesla offer. They will probably reserve the rights to pull data and use it to calculate your premiums - and this will go beyond what anyone else does today. This may be telemetry on speeding, sensors' proximity alerts, instances when AP safety features needed to intervene... They could weed out dangerous or high risk drivers much better than anyone else.

And if it doesn't work out, they can cancel the program: you sign insurance for a year, right? They may only hire or reassign a few dozen back office people for this service and could elect to just let contracts expire if the endeavour appears to be too costly and close this service. It's not like they will have to invest in a national network of branch offices and a new HQ, etc.

 

[azaz]

So far absolute zero report on Trump tariff threat in Chinese domestic media, must be the Chinese government, IMHO looks more like they are working out the differences than a break-down on trade deal, otherwise I should have seen tons of Trump bashing in Chinese domestic media.

Trump probably got word a deal is close and decided to threaten them via twitter so he can take credit after the deal is announced and say his “strong” leadership or whatever saved the day.

 

[EVNow]

Indices futures are down about 2%.

 

[Remus]

Complete speculation.

The goal of Tesla insurance may not be making money rather than make the user experience better. Just imagine the moment you're rear ended, two seconds later your insurance company already obtained the other party's license plate, and submit a claim to his or her insurance with video evidence.

And maybe other car companies want to follow suite if they found it appealing.

What does that mean to Mr. Buffet's insurance business?

 

[Nightshifter]

Good morning,

Secret Meeting of the Tesla Short Sellers - aka “What has Elon Musk Ever Done For US!” #tslaq / #TSLAQ

mods: please delete if already postet.