Tesla Q2 2018 earnings thread

[ZachShahan]

And asked the stupidest fuqqin questions to boot. Neither bull nor bear would be happy.

Dude, those were some of the most epic questions in Tesla conference call history. I was laughing so much — hoped my questions would come right after his. :-D

Adam Jonas is a comedic genius. One day he will pop onto a late night talk show and tell us his years of weird questions were a long-term comedic act. :-D

 

[ZachShahan]

Next question is Zach Shahan from CleanTechnica!!! Asks about "conquest sales" -- how much are they pulling from other segments. They haven't got the data for anything but the top five.

Musk ends up saying that the determinant of Tesla's growth will be cell production and the cell supply chain. (An interesting answer. Is he going to buy his own mines?)

Powerwalls/Powerpack are currently limited by cell production (not module production). They're rapidly solving and expecting to ramp up "later this year and early next".

Solar Roof is going through validation stuff (fire testing, etc.).

"Will Model Y use the same reservation process?" No decision.

"Can Musk talk about the technical bottlenecks right now -- can we picture ourselves in the factory?" Musk says he loves Zach's detailed articles but wants to make sure he doesn't generate a bad sound bite. Zach agrees. Biggest constraint right now -- which will be solved in a week or two -- is in body production. Musk says you can tell what the bottleneck is by what *Musk* is personally working on. They've produced 800 bodies in 24 hours recently.

"Don't let the trolls get you down". Elon thanks Zach for his coverage of cleantech.

Galileo: Will you manufacture trucks in Nevada? Musk says location of manufacturing gets so much attention that he cannot answer *any* such questions.

Galileo: Any synergies between Model 3 and Semi? Lots, but we can't be specific.

Galileo: How are you prioritizing batteries between auto and stationary storage? Why isn't storage ramping faster? (This is an *absolutely great* question, BTW.)

Deepak basically says it makes sense to prioritize Model 3, but that they are expanding cell production ASAP.

JB says that they use additional vendors (Samsung and LG) for Powerwall/Powerpack, but not for cars, which does explain why the Gigafactory production is prioritized for cars.

Deepak explains that they're also bottlenecked on installs. Elon says there aren't enough electricians in the US (I can confirm this is true -- I have massive delays on an electrical project because of the electrician shortage), and Tesla is starting their own *electrician training program*. (Now that's interesting!)

Galileo: what didn't we ask about that we should have asked about?
Musk thinks they covered everything.

Galileo: will the new strategy of *no outside capital* slow your growth down?

Musk says that they're now scaling up so fast that spending more money isn't speeding it up any more. (Mythical Man-Month, I guess) but leaves out the option for debt for particular Gigafactory constructions. (GREAT answer.)

----
As a personal note, I really appreciate Zach's and Galileo's questions. I also appreciate when Musk responded to dumb questions by idiot analysts by *meandering off the topic* and telling us something much more interesting. It's a very polite way to give us useful information without bothering the egos of the analysts.

Thanks, man!! Your feedback has been invaluable over the years. Much of our content comes from reader feedback & ideas, but you definitely stand out — can't thank you enough.

Of course, I had a million questions, comments, and jokes on my mind, but focused on the ones I was most curious about + thought would be most useful.

With Model Y, I hope it gave him a good amount of time to think more about this in order to optimize and not have the issues we've had with the Model 3 reservations. I'm not sure of the best solution of course — cash up front is nice, but quite dangerous to have people waiting too long. It's depressing to be in Europe with a Model 3 reservation.

And I took his compliments to mean for the whole team at CleanTechnica.com — I hardly even write these days, and it's the hugely awesome content from several contributors that I think got us on the call.

Cells for storage vs cars: I've just assumed it's obvious that the cars are the face of Tesla, the reservation list is far "too long" for the Model 3 and people are getting impatient (can wait on a roof or energy storage unit, not so easy to wait on a car), and presumably the cars bring in more revenue per product or kWh of battery capacity anyway. Though, the storage production & growth target was further down my list, so it was awesome Elon & Galileo spent good time on it. :-D

"Elon says there aren't enough electricians in the US (I can confirm this is true -- I have massive delays on an electrical project because of the electrician shortage), and Tesla is starting their own *electrician training program*."
— That was crazy interesting. :-D

Wish Elon could have answered some of Galileo's questions that he didn't feel comfortable jumping into.

And Galileo had awesome ending questions.

"I also appreciate when Musk responded to dumb questions by idiot analysts by *meandering off the topic* and telling us something much more interesting."
— That was a superb showing. Loved his professionally, politician-like control of the story. I know many of us were asking for that, but it was better than I was even hoping. And the apologies seemed extremely sincere and humble, which is so cool. The questions were boneheaded, but you don't have to pick on someone for asking dumb questions and can always control the answer. Slick performance. Elon was a rock star.

 

[ZachShahan]

Given what they've learned, I do think they could get 150,000 out of Shanghai in 2020.... during the latter part of the year... if they manage to get battery cell supply lined up.

Think of Intel's "exact copy" procedure for new factories. Tesla isn't doing that, but they are trying to clone with only small changes, which should be faster than a from-scratch rampup. However, battery cell supply isn't lined up yet, as Elon noted. It's gotta be lined up.

(So I kind of think they won't get it done on time.)

Also, I think it wasn't clear, but isn't he talking about a run rate? As in, if they produce 20,000 vehicles in the last week of 2020, they hit the rate.

But, yeah, his comments on this call seem to make it clear that battery cells are a big challenge going forward.

 

[wdolson]

OK, so I liked the Autopilot info. First, they're focusing on on-ramp to off-ramp in the near term (which is *much* more viable than full level 5 everywhere). Second, they've developed an optimized matrix multiplication (tensor processing) chip which is 10 times as fast as other people's. This design is informed by the *structure* of the neural network which they got from their first/second generation Autopilot, so that was useful. It's also a drop-in replacement in the existing computers, so everyone with an older Autopilot version will get a hardware upgrade...

I doubt AP1 will get an upgrade path though it would be nice if it did.

I could see what they did with AP. They worked out how the neuro-net was supposed to work with an off the shelf graphics processor. Once they got the low level bits worked out, they designed custom silicon to do that job. That will always be faster. The chips take the same input format and have the same output format as the existing GPU board, so it's a drop in replacement in the existing AP2 system.

(But frankly, I bet Tesla could sell that tensor processing unit as a product. It's useful for many purposes, particularly if other problems end up with neural networks with similar structures to the Autopilot structure, which is likely. This is making me think of the way Google and Amazon got their fingers into everything.)

Those types of hardware do tend to be designed for one specific purpose, though I suppose they could leverage it for some other high sensor application down the line. Before processors got as powerful as they are today, discrete hardware that did the job was much more common (I did some in the early 90s), but a generic processor running your software is good enough for most applications and it has the added benefit of being much easier to change through software and can be upgraded to some degree with faster processors or more memory.

Discrete hardware specifically done for the job is necessary when the job is too big for a processor to handle in any reasonable time. The number of crunches necessary to mine one Bitcoin has gotten so complex (it was designed to start easy and get more complex over time) that the only people who are doing it profitably are one with chips designed from the ground up for that one purpose.

But then Musk added that more than half of all customers are choosing AWD, which is useful because it confirms something I suspected but didn't have proof of.

There was all the hand wringing in the media about how so few reservation holders for the Model 3 had converted. The answers were obvious to anyone who spent any time here. There were three groups of reservation holders who had not converted: people outside the US where the car is not sold yet, those waiting for the short ranged version, and those waiting for the AWD version (including performance). As soon as they opened the door for the AWD version, those who had been waiting for it jumped.

AWD sales may drop a bit over time, but the AWD version of the S was more popular than the RWD version, though there you also got a range boost and that isn't the case with the Model 3. With the Model 3, the gains from the extra regen are probably offset by the extra weight. Just checked the EPA's site. They report all 3 current versions at 310 miles range, but the RWD is 130 MPGe and the AWDs are 116 MPGe.

Compare Side-by-Side

Then they used the silly question as a riff to say totally different things...
The top 5 non-Tesla trade-ins to Model 3s...
Toyota Prius
BMW 3 series
Honda Accord
Honda Civic
Nissan Leaf

...OK, to me this makes sense. I think Prius sales will collapse; it was "green" but it's not "green" any more. Leaf will probably survive but have to cut their price. And someone was suggesting that the Toyota Camry sales would suffer, but it looks like it's *Honda* who's going to get slaughtered instead.

If you're talking about who was predicting the Camry would take a hit on this forum, it might be me. I have thought for some time the real competitors to the Model 3 were mid-sized (now most are technically full sized by the goofy way these things are calculated) family sedans, not luxury sedans. Somebody who can afford a fully loaded Camry or Accord can probably squeeze the budget into a Model 3.

Tesla only reported the top 5 and they didn't really look any further. We don't know the rest of the list, nor do we know what cars people are moving from which weren't traded in. When someone has a couple of months to wait for their car, they have time to sell their old car, or transfer it to a family member. Somebody needs to do a poll of new Model 3 owners and ask them what they were driving before.

I didn't trade in my old car because it had become a minor collectible (a 24 year old Buick with a 5.7L V-8). Being 24 years old, Tesla would have given me a pittance and sent it to the wreckers. I found a local enthusiast who had one that got totaled. The fact that mine was in perfect running condition with an almost new looking interior were good selling points.

I do find it very interesting that two mainstream ICE are in the top 5.

Second apology to second guy, and at least the second guy asked a reasonable question: how much will the China gigafactory cost? Guessing closer to $2 billion for 250K vehicles/year. He says that number doesn't include cell production and it hasn't been figured out yet. (Includes everything from pack to final assembly.)

JB explains that some of the most capex-expensive areas have been sped up by 20% by figuring out what matters and what doesn't

Whenever anybody does anything new, the capex is going to be higher because you are going to run into some expensive dead ends. Setting up the second Model 3 factory will be much cheaper than the first.

That one didn't jibe for me. The follow-up was "if Fremont can only make 600,000, where do you get the extra capacity?" And Elon said slyly "there's a place called Shanghai..." But they said to save CapEx they won't start building Shanghai until later in 2019, so it's extremely doubtful (another Tesla over-optimistic goal) they could be producing 200,000+ vehicles in Shanghai within a year. Unless they'll build them in a tent?!? Some pieces don't fit there - they can't scale back capital investment to get profitable, and also accelerate production past what existing lines can be optimized for...

The Chinese can move very quickly when they want to. It is possible they are looking to move into an existing building and expand it.

I don't know if Shanghai will be producing any cars by 2020, but it might. On the other hand, the world economy could take a hit soon and everything will slow down. Current political environment aside, the world economy and especially the US economy is in an unusually long growth period and that will have to end eventually. Probably sooner rather than later.