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

[Fact Checking]

About FSD. It was a kick-starter campaign. I didn't bother with Tesla at those times but I expect they needed client money to cushion finances for the initial research.
Seeing experience of Musk with mass media, I would dig first to find actual Musk quotes in context about FSD.

Agreed.

On the other hand it would be obvious for anybody with basic understanding of data analysis that the last crucial 5% would take 95% of the research-training time.

"The first 90% of a large software project takes about as much time and money to complete as the second 90% of the project" is the conventional wisdom among people with software and high-tech development experience, which conventional wisdom has a lot of truth and hard lessons behind it, but neural networks have radically different emergent properties:

• A NN training dataset, once collected and labelled, defines the capabilities of the network - and a training dataset is growing monotonously, so capabilities are never 'lost'. With conventional software the speed and cost of development scales with project size: the larger the project, the harder it is to expand it. This is not true of neural networks - they only depend on the training data set and available (off-site) training computing capacity.
  
• Tesla is expanding their training dataset by collecting AP disengagement events from beta testers and from the wider owner fleet as well. This means that their ability to expand the training dataset scales with fleet size, which is growing super-linearly at the moment. Also the 'experience' from past disengagement events never goes away: it will be part of every single future neural network trained.
• If the training dataset is large enough neural networks can be 'tested' off site (on large clusters and in simulation environments) with a high degree of confidence. I estimate Tesla's labeled training dataset to potentially be in the 'hundreds of millions of images' order of magnitude, maybe even a billion images, which is unheard of sophistication in academic NN circles.
  
• Tesla's current bottleneck appears to be not NN quality but HW performance - see this comment of mine.

I.e. in a highly counter-intuitive fashion the typical learning curve of NNs is slow progress for some time, then almost miraculously a singularity event and "magic" happens: like AlphaGo which used deep learning to become a Go world champion (most complex major board game known, much more complex than chess), in just 3 days from first principles, by playing against itself and learning. Similarly OpenAI trained itself to beat the best Dota players, in just a few days.

While the "90% rule of software/hardware engineering" still applies to Tesla's NN training, deployment and in-car platform and infrastructure, it doesn't apply to the neural networks themselves. This is how Tesla was able to surpass AP1 quality early last year, introduce Navigate on Autopilot last fall and Elon's Model 3 has a development version that can do stop signs, traffic lights, left and right turns as well - that's an almost full Level 4 autonomy feature set.

 

[CorneliusXX]

The teslamotors subreddit has quite a few interesting topics which have not yet been mentioned here. A couple of the best are linked below.

Tesla closing 45 stores this week:
Tesla will close 15 stores tomorrow and an additional 30 on Monday : teslamotors

Autopilot is relentlessly adding features - it can now recognise stop lines (i believe this is from @verygreen):
Autopilot now can recognize stoplines : teslamotors

 

[Antares Nebula]

@Papafox posted a secondary analysis on the recent action (beyond his daily updates), which some may find informative:

Daily TSLA Trading Charts

 

[Antares Nebula]

Dragon docking live now. Nice detailed visual of the dragon capsule from ISS. Pretty cool.

 

[Antares Nebula]

When can I buy SpaceX stock?

 

[Sean Wagner]

Reading through the latest Tesla conference call provided me with essential information, and I
a) feel much better about the recent / impending changes
b) wish the call had been rapidly published in its entirety
c) really like that they gathered a diverse and often knowledgeable group of media people
d) appreciated the enforced, focussed style

Regarding the Model 3 SR, I was afraid a metal roof might dent its allure and make it appear much more like other EVs.
Very happy Elon explicitly recognized the glass top's unique value [quelle surprise], and that manufacturing costs / efficiencies enabled the decision.

Conversely, the inclusion of full self-driving hardware is rather amazing in my opinion.
I'm somewhat surprised regarding the current "status of autonomy" - expected basic signs & signalling to be recognized.
Still think the latest Cruise video shows a remarkably advanced system, albeit running on prototype hardware.
[And yes, keeping in mind the Tesla approach to learning is different.]

Finally, and quoting Elon: "the car is out".

 

[mrdoubleb]

The teslamotors subreddit has quite a few interesting topics which have not yet been mentioned here. A couple of the best are linked below.

Tesla closing 45 stores this week:
Tesla will close 15 stores tomorrow and an additional 30 on Monday : teslamotors

Autopilot is relentlessly adding features - it can now recognise stop lines (i believe this is from @verygreen):
Autopilot now can recognize stoplines : teslamotors

Right. Especially on the first one there are some interesting comments from self identified Tesla sales staff and ex sales staff. Some of that sheds light to the usually bad Tesla communication both inside and outside the company and we have a the bankruptcy FUD back again.
One interesting tidbit is that the company was projecting to hit only 55% of Q1 US sales target this quarter. This is hard to verify, but would certainly explain why SR launch was pulled ahead and why we saw those significant price cuts across the board.

I must admit I am still somehow skeptical about the closing down of all the stores. Tesla needs to explain how many galleries will remain and where and how people would still get access to test drives other than the 7 day scheme (which, as others explained is not gonna work for those who take out a big loan for the purchase).

I am seriously considering an SR+ or MR towards the end of the year myself, but given that this will be my biggest purchase ever outside my flat and will require me to get a big loan for ~50% of the cost, I would not buy without sitting in the car and taking it out for a spin.

Here is the OP from the first link.
"Moving much quicker than expected, Tesla will close 15 stores tomorrow and 30 more on Monday. There are likely to be more the following week as well. A few of these stores sent employees home with a weeks worth of pay asking them not to come back, which will likely be the case across the board for these closures.

Things are moving at a much faster pace than anticipated. The call to teams on Friday had included a “transition period over the next month or two.” Clearly that is not what is happening and stores with low sales are being immediately shut down.

Edit: no list yet as to which stores are closing outside of the ones already notified and given a weeks worth of pay, which I’m told is ~3 stores. Employees are being kept completely in the dark.

Edit 2: I got a question asking why and I didn’t put the answer or reason in the OP because I know the response will be “FUD!” Hopefully after tomorrow and Monday you will see I’m not trying to spread misinformation at all. I’ll just relay what I was told went down on the Friday phone call:

The question was asked to the management at Fremont why the commission pay was going away and what the incentive was for sales to continue past today (Friday 3/1) to help sell vehicles. The reply was that “there is not enough money to continue to operate. Tesla is closer to bankruptcy than it has ever been and these are the things we need to do to keep the company going.”

Edit 3: Sales in North America were forecasting at 55% of target for the quarter at end of day 2/28."

 

[Dan Detweiler]

Right. Especially on the first one there are some interesting comments from self identified Tesla sales staff and ex sales staff. Some of that sheds light to the usually bad Tesla communication both inside and outside the company and we have a the bankruptcy FUD back again.
One interesting tidbit is that the company was projecting to hit only 55% of Q1 US sales target this quarter. this is hard to verify, but would certainly explain why SR launch was pulled ahead and why we saw those significant price cuts across the board.

I must admit I am still somehow skeptical with the closing down of all the stores. Tesla needs to explain how many galleries will remain and where and how people would still get access to test drives other ta the 7 day scheme (which, as others explained is not gonna work for those who take out a big loan for the purchase).

I am seriously considering an SR+ or MR towards the end of the year myself, but given that this will be my biggest purchase ever outside my flat and will require me to get a big loan for ~50% of the cost, I would not buy without sitting in the car and taking it out for a spin.

Here is the OP from the first link.
"Moving much quicker than expected, Tesla will close 15 stores tomorrow and 30 more on Monday. There are likely to be more the following week as well. A few of these stores sent employees home with a weeks worth of pay asking them not to come back, which will likely be the case across the board for these closures.

Things are moving at a much faster pace than anticipated. The call to teams on Friday had included a “transition period over the next month or two.” Clearly that is not what is happening and stores with low sales are being immediately shut down.

Edit: no list yet as to which stores are closing outside of the ones already notified and given a weeks worth of pay, which I’m told is ~3 stores. Employees are being kept completely in the dark.

Edit 2: I got a question asking why and I didn’t put the answer or reason in the OP because I know the response will be “FUD!” Hopefully after tomorrow and Monday you will see I’m not trying to spread misinformation at all. I’ll just relay what I was told went down on the Friday phone call:

The question was asked to the management at Fremont why the commission pay was going away and what the incentive was for sales to continue past today (Friday 3/1) to help sell vehicles. The reply was that “there is not enough money to continue to operate. Tesla is closer to bankruptcy than it has ever been and these are the things we need to do to keep the company going.”

Edit 3: Sales in North America were forecasting at 55% of target for the quarter at end of day 2/28."

No problem. When do you want me to come over to let you drive my car?

Dan

 

[SebastianR]

One interesting tidbit is that the company was projecting to hit only 55% of Q1 US sales target this quarter. This is hard to verify, but would certainly explain why SR launch was pulled ahead and why we saw those significant price cuts across the board.

Is this bad though? Taking Model S sales from InsideEVs for 2018 (just to not compare to the production constraint Model 3 a year ago) we get this:

January: 800
February: 1125
March: 3375
They did more than 50% of their sales in March

Monthly Plug-In EV Sales Scorecard: Historical Charts

I.e. if they are already(!) at 55% of the target for the quarter, they are doing better today than Model S sales last year. They only way this could be construed as bad is, if they didn't ship as many cars to Europe as planned (in reality I suspect they just managed to make more cars than expected...)

"Moving much quicker than expected, Tesla will close 15 stores tomorrow and 30 more on Monday. There are likely to be more the following week as well. A few of these stores sent employees home with a weeks worth of pay asking them not to come back, which will likely be the case across the board for these closures."

Makes sense. If you cut, cut fast. Then tell the rest it is over. Don't wait for weeks and months. Sucks for everyone involved. Don't like it myself. But no other way to do it.

 

[RobStark]

@kyzhk Re price cut: I think people need to stop complaining. It's one thing if it was Apple, which is rolling in cash. Tesla is still basically a startup trying to figure out this new EV/autonomous car market as it goes. For early adopters it's not just about the car, but the mission. Can't think that oh I could have got it cheaper. That money went to supporting the mission. It's actually a paradox: if they didn't buy it back then at that price, it's possible Tesla would not be around today to lower the prices in the first place.

Be proud that you were among the early Tesla adopters. Back in the day, all my friends and family were talking smack when I got my first Mac. They were all window users. Now almost all have Macs and iPhones. Early adopters got street cred and bragging rights. How much is that worth?

As far as if it was the right move for Tesla, I say rip-off the bandage, unleash the Kraken, and see where things land. This is it. Tesla going full tilt. For investors, its boom or bust. That's the way I roll folks!

In Elon we trust. So say we all!

 

[Buckminster]

Thanks for all the other info as well!

That Tesla has made the right decision to keep the 18,650 cells in S/X is the accepted consensus here on TMC too meanwhile (after a short 2170 craze):

• 18,650 cell mass density has improved together with 2170 mass density improvements. Teardowns confirmed this.
• 2170 has better volumetric density, which matters on the smaller Model 3, but doesn't matter nearly as much on the S/X which are large enough to host the cells.
• Both Tesla and Panasonic have significant capital invested in the ~8 GWh/year 18,650 manufacturing capacity and pack assembly lines, which they should not and will not throw away.
• I.e. the S/X is using the highest grade chemistry available, 18,650 cells are future-proof and there's no trade-off for customers.

FC - this post you at your very best. Concise but leaves no stone unturned. I bought into the 2170 craze - you have won me over - perhaps belatedly....

 

[mrdoubleb]

No problem. When do you want me to come over to let you drive my car?

Dan

I know we have an amazing Tesla community worldwide who are happy to show off the car and let others try it as well. But seriously, Tesla's strategy cannot be to shift all of the physical, real-life demo ("sit-ins") and test drives to enthusiasts. And I am sure it isn't.

I think this could still be a viable approach if they committed to having a gallery in at least all major cities and capitals in the markets they operate and offer test-drives from service centers as well. Especially once you are selling to mainstream customers, they are not going to try and find a friend at TMC to test drive...

 

[JRP3]

I'd like to see them use the blue goo in the S/X or whatever seems to have made the Model 3 pack nearly fireproof. Tired of the burning Tesla headlines.

 

[mrdoubleb]

Is this bad though? Taking Model S sales from InsideEVs for 2018 (just to not compare to the production constraint Model 3 a year ago) we get this:

January: 800
February: 1125
March: 3375
They did more than 50% of their sales in March

Monthly Plug-In EV Sales Scorecard: Historical Charts

I.e. if they are already(!) at 55% of the target for the quarter, they are doing better today than Model S sales last year. They only way this could be construed as bad is, if they didn't ship as many cars to Europe as planned (in reality I suspect they just managed to make more cars than expected...)



Makes sense. If you cut, cut fast. Then tell the rest it is over. Don't wait for weeks and months. Sucks for everyone involved. Don't like it myself. But no other way to do it.

Depends on how you interpret that sentence. Maybe native speakers can help, but remember this is, at best, second or third hand information typed into an internet forum. So were they already at 55% of the target for Q1 on 2/28, or were they projecting to be at 55% at the end of the quarter (unless they do something, hence SR)?

 

[bhtooefr]

Did I miss something -- is it confirmed that test drives are a thing of the past? Or are we speculating based on store closures? If the latter, test drives could still be given via service centers, for instance.

I know we have an amazing Tesla community worldwide who are happy to show off the car and let others try it as well. But seriously, Tesla's strategy cannot be to shift all of the physical, real-life demo ("sit-ins") and test drives to enthusiasts. And I am sure it isn't.

I think this could still be a viable approach if they committed to having a gallery in at least all major cities and capitals in the markets they operate and offer test-drives from service centers as well. Especially once you are selling to mainstream customers, they are not going to try and find a friend at TMC to test drive...

Additionally, as I've suggested before, you can have traveling test drive events (with a dedicated, better-trained team), as Tesla has done for the S and X, and you can also book space at auto shows for sit-in events. This would have much lower cost than maintaining galleries with the downside that you can't just go at any time to do a test drive or sit in a Tesla.

I'm telling ya, head first entry is the back and knee saver. (that doesn't sound very good does it?) Seriously, I started using this method when I drove a TR6 and have used it ever since.

Yeah, I figured this out at 6'1" with a Miata, although I don't use it in my Prius.

 

[SebastianR]

Depends on how you interpret that sentence. Maybe native speakers can help, but remember this is, at best, second or third hand information typed into an internet forum. So were they already at 55% of the target for Q1 on 2/28, or were they projecting to be at 55% at the end of the quarter (unless they do something, hence SR)?

To me Q1 target is an absolute target for Q1.

A YTD Q1 target is the - 'year to date' i.e. today's target to reach the overall Q1 target at the end of Q1.

There was no mention of a YTD target in the post. Either way - this is all hearsay. And I have seen managers use the most "creative" language possible when firing employees myself (incl. one of the managers in my organisation kicking out an unwanted employee for "budget" reasons in an absolute record year where we were swimming in money).

 

[Cobbler]

Any opinions about pricecuts for Powerwall?
In Belgium you pay 7000€ for 14kW or 500€/kWh which is too expensive to have any return as an investment and prices haven't dropped a long time.

Considering a much more complex (but I admit on a larger scale built) batterypack for their car, costs less than 200€/kWh.

 

[mongo]

Makes sense. If you cut, cut fast. Then tell the rest it is over. Don't wait for weeks and months. Sucks for everyone involved. Don't like it myself. But no other way to do it.

Agreed,

The longer you wait to fire someone, the longer it has been since you should have fired them

-Elon Musk (as quite my Ashlee Vance)

In the 2018 thread there was a link to an article regarding a peace time general and a war time general. I thought that very fitting.

There is a difference between someone cuting off your arm out of indifference (or worse to cause you pain) and someone cutting off your arm due to gangrene.

 

[CorneliusXX]

Depends on how you interpret that sentence. Maybe native speakers can help, but remember this is, at best, second or third hand information typed into an internet forum. So were they already at 55% of the target for Q1 on 2/28, or were they projecting to be at 55% at the end of the quarter (unless they do something, hence SR)?

In my opinion they are describing the latter. However, it's important to reiterate that this information is completely unverified.

 

[KarenRei]

Is there a scale shown for the GAAP profit/loss?

Right-hand side. All dollars are on the same scale.

BTW: AP now appears to be detecting stop lines at intersections in shadow mode:

And it's no-fun days over in Ford:

Anxiety builds among Ford employees. Hackett says that's fair, but he's confident