General Discussion: 2018 Investor Roundtable

[RobStark]

 

[sundaymorning]

View attachment 294363

Rob, do you find this site to be accurate?

 

[RobStark]

Rob, do you find this site to be accurate?

I think the consensus at TMC is that the Bloomberg tracker is a trailing indicator. I agree with that.

For Q1 2018 the Bloomberg tracker was within 5%.

Now they added the "trend" feature trying to predict what the current production is now, that may be confirmed in a week or so.

 

[vigleik]

All of the factory videos in one place Check Out All 8 Videos Of Tesla CEO Musk On CBS This Morning

Does anyone know if the videos are available to an international audience anywhere? Or do I just have to mess around with web proxies? (I hate the “the uploader has not made this video available in your location” message.)

 

[EVMeister]

I live in the UK and could see them, no problems or proxies.

 

[avoigt]

View attachment 294363

The interesting part is that in the Video when they went into the meeting room where he sleeps there is a white board where its written that a production line rate of 425 units has been achieved. Maybe a burst rate for just one day

Extrapolating to a week that would make it 3045 cars.

Pretty consistent with the Bloomberg tracker.

Two independent data points in the same range make me feel confident that 3k is in reach.

https://i.redd.it/ovo8sd7eo0s01.jpg

Putting my Bull cap on: With 2k S+X and 3k 3 (bust rate) they are at 5k/w or 250,000 cars annualized these days. Thats a 150% increased versus 2017 after just 4 months.

That number, the 435/day may very well have been triggering the higher than usual VIN registration # we have seen as automated system have be alerted and did their duty. Would expect that we have been falling lower in the meantime though.

 

[Reciprocity]

I am mostly worried about the gross margin right now. Elon said clearly that certain systems didn't work and they need(ed) a lot more manual labor. This is not only a loss on their investment (both faulty systems and stagnant capacities) but there are also additional costs for labor. When Elon announced a 25% gross margin target he still thought that the machine that builds the machine is going to work a lot better than it did.

With it's Alien Dreadnought plans Tesla viewed manufacturing mainly as a highly technical problem. The only humans in the factory were supposed to maintain the machines (end goal). The obvious advantage was that machines can physically move a lot faster than humans can and therefore the whole production line could be a lot faster. But there are disadvantages. It's not only very challenging to make these incredibly complex systems work, but they are also expensive and if certain sub-systems fail the whole line stands still. You are only as fast as your weakest/slowest part. While Toyota's lean production system and human labor in general is more flexible a failing technical sub-system can be disastrous.
After having thought about it a bit i now believe that EM comment "Yes, excessive automation at Tesla was a mistake. To be precise, my mistake. Humans are underrated." indicates a major strategic move away from pure Alien Dreadnought. It seems that EM is now valuing the flexibility of humans that saved the production in the Gigafactory while it took Tesla&Grohman half a year to built a new technical system/line that is not even proven at high volumes yet.
There are going to be fast systems like the new Grohman line (if it eventually works) and human labor for systems that didn't work, bottlenecks, downtime and so on. It's also going to be challenging to coordinate human labor in parallel so that it can match the speed of the fastest machines. There is no point of having machines that are crazy fast if you can not feed them fast enough. The conveyer belt system therefore needs to be replaced with something that is equally fast but EM seems confident. After all the whole production line needs to be smooth without idle running or congestions.

Given that those observations are true i now wonder how many more humans Tesla needs short term and how expensive this is going to be. Can anyone tell me how much a median worker earns at Tesla? What do you think are the absolut worst case scenarios? Something like 5.000 workers more earning 10k? (I'm just putting out numbers) Therefore absolut worst case additional expenses of approx. half a billion a year?
If the lines are less automated, they will certainly need more space for slower human labor and therefore the output per square meter is lower. I think there is no way to get capacity to 15-20k a week. As far as i remember EM said in the last earnings call that he estimates 600.000 Model 3 100.000 Model S/X max because pushing it further would make no economic sense.

Margin Shouldn't be an issue. We are not talking about thousands of people and machines are not free to run either. They just don't call in sick and don't require training. But they are also not as flexible and adaptable as people.

Remember the engineer who was close to Tesla that said the marginal cost for a model S was something ridiculous like $23,000. With more pricing per due to volume, 20% smaller and 30% cheaper battery pack, the model 3s marginal cost could be in the mid teens, say $16,000. I know there is a lot more to getting a car Delivered then marginal costs, but it's the starting point where you start to add people's time and the other somewhat more variable costs. Also remember, Tesla could be building most of these cars in China in next few years and Fremont could be mostly S/X, roadster and semi at some point. A new US plant could also be more cost effective if it's not in California where it is mostly illegal to manufacture anything except trees.

So I'm a China scenario, Tesla would copy what they built for 3 and Y in China. No major ramp headaches, just buy an enormous building and get started. Same goes for the Gigafactory 1&2. But only if they feel they can sell solar tiles in China because they can source solar that at government dumping prices, you can't compete with that. The point is, it's a copy of the machinery. The labor will actually be a bigger issue in the ramp the more labor you need because of training and finding qualified labor. So more robots is good but having some humans is not going to kill Tesla's long term plans to automate more, just delay them. The cost to margin should be under 1% total, they will need to find that somewhere. Let's not forget something. Elon was extremely conservative at $41k ASP. Like EXTREMELY.. with fsd, there will never be a model 3 sold that doesn't eventually have $8,000 - $10,000 worth of software enabled at some point in its first 5-6 years of life. ASP will be greater than $45k, period. That software is a sunk cost so the margin is essentially 100%. That's because no new Tesla will exist that will not have the cost already in the car to support the software. Current model 3 is close to 80% take rate for eap and 30% for fsd. That alone makes it all but a certainty that the asp will be North of $45k.

On the NUMMI debate. As critical is the plant was, the huge stamping press from Detroit was also a steal. I believe it was something like 4mil for a 50M press. Lots of valuable equipment for cheap is going to become available soon.

 

[vigleik]

I live in the UK and could see them, no problems or proxies.

Just in case anyone else if suffering from region restrictions, I found these:

 

[sdtslafan]

About the CBS interview:

First of all I am delighted to see an interview and videos with a positive minded sentiment not falling into the negative narrative that we all see that often. Congratulations to Elon that he obviously understood that positive PR is a requirement to achieve his goals. The lady is balanced and impressed about him and the factory as well as its obvious that Elon is very proud to show his factory. Furthermore the more details are shown and explained the less likely it is that FUD does stick with media and people and is repeated as truth.

I listened to the videos a couple of times and there is a ton of information in it. This is not only what has been said and shown but body language and mimics tell a strong story too that is worth talking about.

Just some findings:

The production line is highly automated and full to that extend that a human cannot even enter the room. To protect workers its standard procedure to build fences around robots. The over automation he referred to may be linked either to certain parts of the production and/or maybe even to the conveyer belt complexity he referred to. In short, this line is highly automated and can run on a much higher speed for sure once bottle necks are eliminated. The use of workers may be helpful in some parts of where for ands we have seen some of those processes and workers as well. No reason for me to believe that he steps back from AD but they adjust just from fully crazy automation to just crazy automation with no compromise on output. All what they said as I hear it is that at some parts humans at least at this time of the ramp can be superior over robots.

I liked particular how good Elon feels in his factory. Its like he loves to walk around see the machine and what has been accomplished so far , quite remarkable for a person with that pressure from the street. It reminds me of an interview with Boris Becker at the Center Court in Wimbledon where he said "this is my living room". People who feel that good often achieve unknown levels of performance. There is a lot of confidence in his words, gestures and responses. Given that he defines his role where his strength are for instance in the issue solving capability overseeing a complex and long production line and not as an investor or entrepreneur the weight of his words that they will be CF+ in Q3 and Q4 is stronger than just from an investor without deeper insights and understanding of technical issues. He does see the path out of the production hell much better than before and therefore we should take his Tweet seriously and to be much more than and anger Tweet about FUD news.

Elon confirmed the batterie assembly was and is still the bottle neck. IOW although we all have seen the increase above 2k/w the car output is still determined by the speed of Batterie output. The car

I think the consensus at TMC is that the Bloomberg tracker is a trailing indicator. I agree with that.

For Q1 2018 the Bloomberg tracker was within 5%.

Now they added the "trend" feature trying to predict what the current production is now, that may be confirmed in a week or so.

Its 4 in the morning where I live so it's entirely possible I'm dreaming, but are you saying they have increased production from 2000/week to 3000/week in a matter of 12 days?!? Because if that's the case, please don't wake me up until after market opens on Monday.

Couldn't be happier to be a Tesla shareholder right now.

 

[Reciprocity]

Its 4 in the morning where I live so it's entirely possible I'm dreaming, but are you saying they have increased production from 2000/week to 3000/week in a matter of 12 days?!? Because if that's the case, please don't wake me up until after market opens on Monday.

Couldn't be happier to be a Tesla shareholder right now.

Reality doesn't matter. The Bloomberg tracker only mattered when it said 700 and stock tanked. They will just say it's not accurate now because Tesla is pumping it by registering tons of unnecessary VINs and randomly producing cars with high VINs. The next bit of real info comes at earnings. My expectation is at the very least they are maintaining 2000+, which is pretty good considering what we heard about getting rid of the conveyors and Elon's own words reaffirming 5000k+ in Q2. I think he said 3-4 times. Now we don't know what he is multiplying it by to start with. Should have more concise numbers in early May.

 

[sweter]

The interesting part is that in the Video when they went into the meeting room where he sleeps there is a white board where its written that a production line rate of 425 units has been achieved. Maybe a burst rate for just one day

Extrapolating to a week that would make it 3045 cars.

Pretty consistent with the Bloomberg tracker.

Ha, good spot! But it says "Vehicles built in one day", so I think it means S3X total :/

Interestingly they don't have a separate count for the 3 at all.

It might be of no importance at all. In my office we like to put things like that on a whiteboard as well and never ever update them again even if those records are beaten Might explain why there's no Model 3 listed.

 

[elasalle]

View attachment 294363

Per EM CBS interview they produced 2071 M3 in past 7 days - best to stick with this for at least another 2-3 weeks

Looks like the Bloomberg model is adding too much weight to newly registered vins - if this was low the trend would be low

Other observation I see is that all latest vins in chart are in yellow- suggesting that all new vins directly reported to Bloomberg - this really cant be true - I think there is more vins reported to tmc - most likely that spreadsheet still getting used by them

 

[mongo]

Ha, good spot! But it says "Vehicles built in one day", so I think it means S3X total :/

Interestingly they don't have a separate count for the 3 at all.

It might be of no importance at all. In my office we like to put things like that on a whiteboard as well and never ever update them again even if those records are beaten Might explain why there's no Model 3 listed.

It looks like the date was in January whatever that means (the Q1 burst rate number?) Or just someone messing with the peanut gallery...

 

[JRP3]

Per EM CBS interview they produced 2071 M3 in past 7 days - best to stick with this for at least another 2-3 weeks

But when was the interview taped?

 

[Cattledog]

But when was the interview taped?

I believe Tuesday, April 10.

 

[j0hn]

it appears that buffering for export by ship has started again looking at these images :

Compare daily satellite images of Tesla production locations

maybe someone with a drone can go have a look

 

[kelevra]

Margin Shouldn't be an issue. We are not talking about thousands of people and machines are not free to run either. They just don't call in sick and don't require training. But they are also not as flexible and adaptable as people.

Remember the engineer who was close to Tesla that said the marginal cost for a model S was something ridiculous like $23,000. With more pricing per due to volume, 20% smaller and 30% cheaper battery pack, the model 3s marginal cost could be in the mid teens, say $16,000. I know there is a lot more to getting a car Delivered then marginal costs, but it's the starting point where you start to add people's time and the other somewhat more variable costs. Also remember, Tesla could be building most of these cars in China in next few years and Fremont could be mostly S/X, roadster and semi at some point. A new US plant could also be more cost effective if it's not in California where it is mostly illegal to manufacture anything except trees.

So I'm a China scenario, Tesla would copy what they built for 3 and Y in China. No major ramp headaches, just buy an enormous building and get started. Same goes for the Gigafactory 1&2. But only if they feel they can sell solar tiles in China because they can source solar that at government dumping prices, you can't compete with that. The point is, it's a copy of the machinery. The labor will actually be a bigger issue in the ramp the more labor you need because of training and finding qualified labor. So more robots is good but having some humans is not going to kill Tesla's long term plans to automate more, just delay them. The cost to margin should be under 1% total, they will need to find that somewhere. Let's not forget something. Elon was extremely conservative at $41k ASP. Like EXTREMELY.. with fsd, there will never be a model 3 sold that doesn't eventually have $8,000 - $10,000 worth of software enabled at some point in its first 5-6 years of life. ASP will be greater than $45k, period. That software is a sunk cost so the margin is essentially 100%. That's because no new Tesla will exist that will not have the cost already in the car to support the software. Current model 3 is close to 80% take rate for eap and 30% for fsd. That alone makes it all but a certainty that the asp will be North of $45k.

On the NUMMI debate. As critical is the plant was, the huge stamping press from Detroit was also a steal. I believe it was something like 4mil for a 50M press. Lots of valuable equipment for cheap is going to become available soon.

I guess you are referring to fixed cost not marginal cost? Marginal cost is the whole cost of a car getting produced and consists of fix cost/quantity produced + variable cost. Do you have the original source of that engineer saying that?

I always thought that if Tesla succeeds with Alien Dreadnought California could be a location advantage because of the highly skilled talents available and then copying what they built in the whole world for better logistics, tariffs exactly like you said. There are certainly going to be improvements with automation but if it turns out that the cost of machines, running those machines, downtimes etc. are higher than the cost of human labor Tesla could significantly walk back on their overall automation plans. I know that for instance Toyota walked back on automation and that their newest generation factory actually is less automated and less high tech than the generation before. How does this effect the planed margin? I don't know. But i think it could (not saying that it will) be way more than just 1%. Maybe up to 5% or maybe even 10% if you consider all costs. (human labor, restructuring of factory, lower density, lower speed) I don't think 10% is likely but i have to calculate certain margins of error/safety (because of my lack of knowledge) into my models.

I think you're right with the ASP. Is there data that confirmes that the eap take rate is 80% and fsd 30%? Can you post the source?
Same question for @ValueAnalyst. Are those numbers (take rates for models, options) your assumptions or is there indicating data?

bloomberg tracker being at 3000/wk MODERATOR DELETED EM said in the interview a few days ago that they produced 2071 last wk.

 

[mongo]

I guess you are referring to fixed cost not marginal cost? Marginal cost is the whole cost of a car getting produced and consists of fix cost/quantity produced + variable cost. Do you have the original source of that engineer saying that?

I always thought that if Tesla succeeds with Alien Dreadnought California could be a location advantage because of the highly skilled talents available and then copying what they built in the whole world for better logistics, tariffs exactly like you said. There are certainly going to be improvements with automation but if it turns at the the cost of machines, running those machines, downtimes etc. are higher than the cost of human labor Tesla could significantly walk back on their overall automation plans. I know that for instance Toyota walked back on automation and that their newest generation factory actually is less automated and less high tech than the generation before. How does this effect the planed margin? I don't know. But i think it could be way more than just 1%. Maybe up to 5% or maybe even 10% if you consider all costs. (human labor, restructuring of fabric, lower density, lower speed) I don't think 10% is likely but i have to calculate certain margins of error (because of my lack of knowledge) into my models.

I think you're right with the ASP. Is there data that confirmes that the eap take rate is 80% and fsd 30%? Can you post the source?
Same question for ValueAnalyst. Are those numbers (take rates for models, options) your assumptions or is there indicating data?

bloomberg tracker being at 3000/wk is bullshit. EM said in the interview a few days ago that they produced 2072 last wk.

@Reciprocity
It was Tom Mueller, Chief Propulsion Technology Officer at SpaceX.
Tesla is going to ‘kill’ the auto industry with Elon Musk’s way of thinking about manufacturing, says SpaceX CTO

Elon: How much do you think it cost to make a Model S?
Tom: I don’t know. About $50,000?
Elon: No. About $30,000 is the marginal cost to produce that car. How much do you think it weighs?
Tom: About 5,000 lbs?
Elon: Right, about 5,000 lbs. And how much a Merlin engine weigh?
Tom: About 1,000 lbs.
Elon: So why the heck does it cost a fraction of $1 million to make a Merlin engine? I’ll give you a factor of 5 since it is not made of aluminum and it’s not stamped. So why is it 20 times the cost?

 

[Starno]

Margin Shouldn't be an issue. We are not talking about thousands of people and machines are not free to run either. They just don't call in sick and don't require training. But they are also not as flexible and adaptable as people.

Yes but it's also about speed :

" the fastest auto manufacturing lines move at 0.2 meters/second, whereas a human walks 1 meter/second (5x faster), and said he saw no reason it couldn’t be 2-3 meters/second. If one company (Tesla) can speed up its output by 5x, whereas another has to build five factories, who’s going to win, he asked? "

 

[goinfraftw]

Yes but it's also about speed :

" the fastest auto manufacturing lines move at 0.2 meters/second, whereas a human walks 1 meter/second (5x faster), and said he saw no reason it couldn’t be 2-3 meters/second. If one company (Tesla) can speed up its output by 5x, whereas another has to build five factories, who’s going to win, he asked? "

Believe you're referencing this video w/ Sam Altman and Elon Musk. Must watch - especially for the cool robotics in the background even back then .