特斯拉 (TSLA) 2015 Q2 法說會逐字稿

Ladies and gentlemen, thank you for standing by and welcome to the Tesla Motors, Inc., second-quarter 2015 financial results Q&A conference call.

(Operator Instructions)

As a reminder, this conference may be recorded. It's now is my pleasure to turn the floor over to Jeff Evanson. Sir, the floor is yours.

Thank you, Hughie, and good afternoon, everyone. Welcome to Tesla's second-quarter Q&A webcast. I'm joined today by Elon Musk, Tesla Chairman and CEO; JB Straubel, our CTO; and Deepak Ahuja, Tesla's CFO. Our Q2 results are announced in the shareholder letter at the same link as this webcast. As usual, the letter includes GAAP and non-GAAP financial information and a reconciliation between the two.

During our call, we will discuss our business outlook and make forward-looking statements, which are based on our predictions and expectations as of today. Actual events or results could differ materially due to a number of risks and uncertainties including those mentioned in our most recent Form 10-Q filed with the SEC.

And now, Hughie, let's go to the first question please.

(Operator Instructions)

John Murphy, Bank of America Merrill Lynch.

Good afternoon. A first question on the pre-owned program, it seems like there was a fair level of success there with $20 million in revenue. I'm just curious if you could dimension how many vehicles were sold through that program? How many units remain in inventory there? And if we think about how that inventory is restocked, what percentage of new unit Model Ss that are purchased are accompanied by a Model S trade in?

One second.

Hi, John, Deepak here. Firstly, we just kicked off the program in April, so it's actually been pretty heartening to see how it's done so far. To keep it at a high level, we are actually selling these cars at a faster rate than we are getting these trade-ins to come in. As this program picks up, it's going to really be a successful program. And it's creating good demand for us on the pre-owned side.

And I would hesitate to make predictions based on by such early history. I think there's room for optimism in the future here. Because these are obviously very low, that we don't have to make the car, so we are essentially getting commission on selling the car, and it's very capital efficient. So I think there's some potential there, but nothing we want to really make predictions on until we have more of a history.

Okay, maybe to think about it in terms of -- I mean the every vehicle that's going into this pre-owned program I would imagine is accompanied with a new Model S sale? I'm just trying to understand how the acquisition process is working.

Absolutely. We accept a trade in only when the customer is buying a new Model S, if that's what you're trying to suggest.

Yes, that's what I was trying to get at, understand the percentages, but I guess it's too small right now to be too material. The second question is, as we think about the referral program, an interesting marketing angle here. Just wondering if you can juxtapose what you think the ultimate cost was?

It looks like it's about $2,000, all in, versus your acquisition cost for a customer. Just trying to understand where you are going to save money and how this makes the customer acquisition process more cost effective.

Well, it's not necessarily making it more cost effective, it's intended to be somewhat of a wash so that if we achieve our $2,000 savings, we essentially pass that on to the end customer. We don't know what that's really going to look like until the program is complete, which is almost 90 days from now. Early indications are quite positive.

But, a bit obviously, for this 90-day period there's going to be some overlap where we essentially incur a dual expense because we still have all of our stores and have the referral class. I don't think it's going to have a big impact on our numbers, but there will be some dual expense because there's no way -- you have to run these experiments on parallel. There is no real way to do it otherwise.

How do you build up to $2,000 acquisition cost? As you mentioned, it seems like it's a lot of fixed cost that wouldn't get taken out that quickly in the test phase. I'm just to understand how you think about that $2,000 number and how you come up with that?

It's not taken out in the test phase. It's to inform our long-term decision in terms of how many stores should we open. A store should be thought of as like a demand-generation item, and in order to understand should we do a lot of stores, small number of stores, somewhere in between, we need to know how effective our referral program is. If you can think of some other way to do this that we are not aware of, we would love to hear about it.

Sure, we will brainstorm on that. Just lastly, as we think about the cash burn in the quarter, but also the set up of the credit lines, it seems like you guys are recognizing that, some point down the line, there might be a need for a capital raise. Would you consider an equity raise in the market or do you think these credit lines are enough to lean on for now before you ultimately have to make a decision on raising capital in the next 12 to 18 months?

We only drew down $50 million on our credit line, so we have sufficient lines available, and that's expandable, too, to $750 million, which gives us some comfort that we can be close to $1 billion as we go through the year. And clearly as X ramps up and gets to steady state, it enables us to generate free cash flow. So we feel pretty comfortable overall on that front. We'll just take it as opportunities come in the future.

So do you think as you go through the launch of the Model X, and ultimately the Model 3, that you'll turn cash flow positive at the right point where you might not need to do a capital raise going forward? Is that how you're thinking about this, with the credit lines on top?

We're comfortable with the cash levels, I will put it that way.

I don't think that there's not a need to raise equity capital. There may be some value of doing so, as a risk-reduction measure, but to be clear, what Deepak is saying is that, given the absence of any additional capital generation activity, we would have on the order of $1 billion through year. Basically that would be our cash position.

I think the risk-reduction function you mentioned is probably the most valuable and that's why we were asking about that. When you look at the cash burn and how the capital market sometimes shift very quickly, it seems like an opportune time to take advantage of what you might need in the future. So that's why we are asking.

Yes, I think we are in the same mind frame as you are.

Great. Thank you very much.

Rod Lache, Deutsche Bank.

Hi, everybody. A couple questions, first, maybe you can elaborate a little bit on the drop in the forecast from 55,000 to 50,000 to 55,000. Is there some aspect of the Model X launch that is not what you projected? It doesn't really sound like this is related at all to the backlog of orders for Model S.

Correct, we do think that it's going to be quite a challenging production ramp on the X. And if we are faced with the choice of deliver-- we only want to deliver great cars. We don't want to drive to a number that's greater than our ability to deliver high-quality vehicles. The nature of the production ramp, which is basically an exponential ramp, that then becomes an S-curve, is basically for every week longer that it takes us to climb up that exponential is about a 800-vehicle reduction of the X.

That's why we do want to emphasize the longer term. Longer term really just meaning Q1 next year type of thing, not super long term. I think everyone gets a better picture of the business, just thinking about that, and that's where we feel really highly confident, at the 1,600 to 1,800 combined production of S and X, and both production and demand.

Okay, was there an adjustment from something like 2,000 units a week to that 1,600 to 1,800? Are we thinking about a 48-week production year, or are you thinking of this in terms of the full calendar year?

That's a good question. That's averaged over the year. That means, in a given week, we might go as high as 2,000 to make up for holidays, factory retooling and that kind of thing.

Okay. Your run rate of gross margin obviously is being affected by a number of things, by launches, by mix, the deferral of autopilot revenue. Can you talk a little bit about what are some of the issues there? What is the issue with autopilot? Would it be reasonable to expect the margins to rise again to the 25% plus level once X is out and Gigafactory begins to ramp?

Yes, definitely. Most of the autopilot deferral, we should be able to recover that later this year, and maybe a small portion going into 2016, but difficult to say at this point. The two big issues that have been affecting us, I'd say, is the dollar, the strong dollar, and then the mix, especially as we start to build 70D and the 70 cars recently.

The dollar has had a huge impact just from Q1 to Q2, 200 basis point out for us roughly. So even after we consider all that, and we look at 2016, to say that we will be at 25% and better, with S and X combined yes, we should be there.

Okay, and just lastly, you've have a few more weeks here since the announcement of the stationary storage business. Do you have any additional thoughts on how that's expected to ramp?

Yes, I do want to preface this with a certain degree of uncertainty because this is quite new. And again, we've got that challenge of exponential ramp. And then, depending upon how you moved the twin of on how the dates fit over the exponential ramp, the actual numbers in a given quarter could be quite different. But the demand has been really crazy.

So it's well in excess of-- if you just take the reservations that have been made thus far, it's for well over $1 billion worth of Powerpacks and Powerwalls. And that's with no marketing, no advertising, no sales force to speak of really. Not trying to sell it, it's basically a presentation and a webcast and [3 minutes] of press Q&A. So there is probably room to improve.

So this is, really, we're basically sold out of what we could make in 2016 at this point, assuming these order are real, which they seem to be. So we're looking at maybe, again, preface with made for uncertainty 40 million to 50 million in stationary storage in Q4 and maybe as much as 10 times that number for next year. So it's 40 million to 50 million this year and 10 times that next year.

And that growth rate is probably going to keep going at quite an high level. It's probably at least a few billion dollars in 2017. That's sort of speculative at this point, but I think that's likely. So it's growing by half order of magnitude to an order of magnitude per year.

Great. Thank you.

Trip Chowdhry, Global Equities Research.

Thank you. A couple of questions regarding the validation units. I was wondering regarding Model X, do we have a general ballpark number in terms of how many validation units we may have to produce before the robots become smart enough?

I'm not sure I totally understand your question.

Like when we had the Model S retooling we created a few Model S validation unit, I think probably around 40 units or 50 units, which were used to train the robots. So before the product really goes into production, I do believe there are some validation units of the car that are produced so as to train the robots? I could be wrong, but that is my understanding.

That is true. It's a little more complicated than that because they are parts of the factory that are much more automated than other parts. So the programming the robots, it varies quite a bit in terms of how much programming there is, how difficult that programming is. We actually have now produced several Model Xs off the Tesla production line. This is a complex machine with several thousand unique components, so there are still a lot of low-volume parts from suppliers on the Model X. But with each week, we keep building more and more Xs off the line with greater and greater part maturity. And we expect to do our first delivery of production Model Xs at the end of next month.

Excellent. I had a question also on the autopilot. I was wondering, are you aware of this research paper from Alex. I think they also call it Alex Net. In few conferences we've ran, they talk a lot about image classification, especially with NVIDIA. If you haven't heard a word, that's fine, but if you've heard about it I was wondering -- are we working with some similar technologies with autopilot? Or we are doing everything in house? In terms of image recognition and auto steering.

The overall system is designed by Tesla, but then there's components from a number of other suppliers. The Autosteer, highway autopilot essentially, is using a combination of the forward camera, forward radar, the side ultrasonics and then the GPS navigation system. So it integrates those four systems in order to do Autosteer on the highway. So that's what we are working out the final details on.

We're targeting release to our early access customers on August 15. Then, depending on how that is received and what issues we encounter in different parts of the world, we expect to go to wide release on highway Autopilot and Autopark in one to two months after that

Last question, I have regarding the plan to year 2016 production. I was wondering, in 2015, this current year, we are having a few retooling and getting the assembly lines, or I should say the production lines of Model X and Model S that are up-and-coming. Do we see any similar disruption happening in 2016 or we are pretty much great? And when you put the Model 3 line in place, it will be completely isolated from Model X and Model S?

There are periodic down periods in order to do equipment maintenance, where that can't be accomplished in, let's say, on a Saturday, Sunday. But we're not anticipating any significant downtime for S and X, it's just, maybe it's one or two weeks out of the year, something like that. Because we do partuity improvements to reduce the production cost and that's just general equipment maintenance.

For the Model 3, we're really doing our best to make sure it does not affect S and X production. We don't currently anticipate it affecting S and X production in 2016, but there may be some affect in 2017

If I may squeeze in one more, you did mention that Model X production is challenging. Like there's a difference between saying durable and undurable and everything is challenging in the world. I was wondering like we have a control over this challenge, because every problem is, we will not enjoy something not challenging. You think it's challenging, like is out-of-control, or challenging because you are enjoying doing it?

Some things are definitely a lot more challenging than others. And the Model X is, I think, a particularly challenging car to build. Maybe the hardest car to build in the world, I'm not sure what would be harder.

Excellent.

But it is an amazing vehicle, and I think it's going to blow people away.

Looking forward to it. Thank you.

All right, thank you.

Adam Jonas, Morgan Stanley.

Hello, Elon and Deepak. First question, Steve Jurvetson was recently quoted saying that Uber CEO Travis Kalanick told him that -- if by 2020 Tesla's cars are autonomous, that he'd want to buy all of them (laughter). Is this a real, I mean, forget the 2020 for a moment, but is this a real business opportunity for Tesla? Supplying cars to ridesharing firms, or does Tesla just cut out the middleman and sell on-demand electric mobility services directly from the Company on its own platform?

That's an insightful question.

You don't have to answer it.

I don't think I should answer it.

Okay, let's move on. Second question is, there's been -- sometimes you can tell more from the non-answer than from the answer. There's been a lot of excitement about mapping technology for autonomous and semi-autonomous applications with the German consortium bidding for here, Nokia's business.

I would love to hear your views, Elon, on how you view Tesla's mapping capabilities. Is this something you need to continue licensing from outside vendors or can you use your own unique connected machine learning suite to build your own mapping capabilities and be self-sufficient? Or would you rather not answer that (laughter)?

The fact of the matter is, there's no publicly available data that is sufficiently accurate for autopilot, as far as navigation data, suite data, it's too coarse. So, it looks like we don't really have much choice but to create our own data set for driving in the long term in order to provide a high quality autopilot experience. It's just the only way we can think of to do it.

Great. Thanks.

Ryan Brinkman, JPMorgan.

Thanks for taking my question. First question is, I'm just curious what the new guidance for a range of delivery is, 50,000 or 55,000 versus 55,000 prior. What does that mean, if anything, for the earlier target of free cash flow breakeven in 4Q? Is that something we should think about maybe more in 1Q 2016 then?

Certainly in 1Q, we will be free cash flow positive. Q4, it's hard to predict given that range. Certainly towards the end, as Model X deliveries are ramping up, we would be. But then if you look at the full quarter, it's somewhat close to call.

Okay. Thanks, that is helpful. And then just last question, it seems like we have seen a number of announcements since your quarter now by utilities and other companies about using your Powerpacks, on -- Amazon, et cetera. I'm curious, how much of the Tesla Energy business do you now expect to generate from corporate or institutional demand versus more retail customers? I'd be interested, too, if this mix, an expectation if it's changed over time at all since you first debuted the product?

Sure, this is JB. We're getting a little bit better sense of it, but it's still early days. I think initially we thought that the majority of the business would be the Powerpack in the commercial or institutional business, and we've actually been a bit surprised at how strong Powerwall, the retail demand is and the interest is. So, again, it's hard to guess at numbers, but I would say that we're perhaps in the, maybe not quite close to 50/50, but I think 70% perhaps Powerpack.

Yes, it's really early days. A big dependency here is when someone orders the Powerpack, how many Powerpacks do they order? Powerpacks are 100 kilowatt hours and industrial and utility customers may order as many as 100 or 200. I'm not sure, what's our biggest one so far is that 200 (multiple speakers). So, our biggest one is about 250 or so. It could be quite a large, in terms of the kilowatt-hours, quite a large amount, going in the direction of Powerpack.

In terms of unit volume, the Powerwall would be the greatest, but the likely thing for Powerwall is somebody's going to order maybe 1 to 3, maybe it's an average of 1.5, or something like that, whereas Powerpack could be an average of 5 to 10.

I think as the market grows, and as we go further into the future, we are going to see more and more total energy demand coming from Powerpack. That's still our expectation.

Yes, exactly, one important thing to understand, and agreeing with growing up just a little bit, that for stationary storage the fundamental economics of cost are always true. Meaning that there is always a cost advantage to someone, to a system-wide implementation of stationary storage, because of the high peak-to-trough electricity usage.

So, if you have buffering, which is what stationary storage allows for, then you only need your power plants to operate at the average energy usage. Which means that you can basically, in principle, shut down half of the world's power plants if you had stationary storage. This is independent of renewable energy. It doesn't matter whether you have solar panels, this is just being able to shut down half your power plants if you have buffering. Because you can then have your power plant output at the average of what is needed by the consumers.

That is very interesting.

Yes, it seems like sometimes people link this too much to renewable energy. Of course, we're refute believers in renewable energy but, that is not the getting function of demand for stationary storage. Stationary storage is really as compared to existing power plants.

Great, thank you.

Depending on the country, that may be represented as a price to the user of the electricity. So, for example, in Germany and Australia, there is time-based cost of electricity, like it costs more some times of the day than others. Whereas in most of the US you just have a meter that's ticking over. So it doesn't differentiate between, say day use of energy or night use of energy. So in places where the price represents the cost, the Powerwall makes economic sense, but the Powerpack makes sense everywhere.

Great.

Patrick Archambault, Goldman Sachs.

Good afternoon, thanks for taking my questions. Just wanted to follow up on the Autopilot, and I'm sorry if you guys have said this. Can we talk a little bit more about the capability that, that is going to bring? Is that kind of a hands off, feet off, kind of product that would be somewhere close to like NHTSA Level 3? Or we just talking about more lane-keep assist and forward collision warning and that sort of stuff that would still be Level 1 to Level 2? Let me just start off with that question.

I'm not too familiar with what the various NHTSA levels mean, but I'll tell you just what translates to in the initial autopilot. We don't want to set the expectation that you can just basically pay no attention to what the car is doing.

We do want to set the expectation that it's much like the Autopilot in a plane, where you turn the Autopilot on in a plane, but the assumed expectation is the pilot will pay attention to what the plane is doing and won't go to sleep or disappear from the cockpit. So we don't want to set that expectation with consumers.

That said, in terms of what the capability of the system is, I think its capability in steering and control of acceleration and braking is excellent when it has a tracking vehicle in front. You basically have high confidence in steering, braking and acceleration, basically when you are in some kind of traffic situation where there is a car on the road in front of you. I think it's pretty good in the absence of that, so if there's just lanes, it's pretty good. And it will get better over time as we refine the software. I would certainly not take the initial version of Autopilot as the final version, it will just get better and better over time.

Got it. That's helpful --

Just with software updates. Just with software updates, to be clear.

And then, I take it there's some kind of human machine interface thing that keeps you focused and beeps if you try to use your Blackberry or something?

That's still something that we're debating and I think we want to see how the early access program goes, that's basically our public beta. And based on that, we'll set the warning levels, yes.

Got it. More of a guidance question, it was touched on earlier that the 1,600 to 1,800 is down from I think what was being floated around as 2,000 for next year's production level. I get that the production overall might be lower based on a slower ramp but is there something structural that's keeping you from hitting that 2,000 on a run rate basis, understanding that you can search to that at certain points? Is there something that in preparing for the launch you've realized that you're just not going to have the capacity that you thought you would have?

Frankly, the main thing is, we don't want to set high expectations and then -- the only way for us to feel good about the future is that if we exceed really high expectations. So, it's sort of like, winning needs to feel like winning, if that makes any sense. So, that's really why we're saying those numbers. Could we do 2,000 aspirationally? Yes. Do we want to commit to that? Ideally not.

Got it, understood. Just setting the bar at achievable levels is something that makes sense. Just last one for me, an accounting issue. I think there was a $0.10 gain or a $0.13 gain on FX revaluation. I don't know, Deepak, I didn't know what that was exactly.

That's driven by our balance sheet currency and receivable evaluations that happen at the quarter-end exchange rates. It's not representative of what happens during the quarter, in the middle of the months, and based on where the exchange rates were we had good news this time from that evaluation. Clearly, as you are aware, Q1 we had significant bad news, and if you are looking at the two out on a full-year basis, it's a small number.

Okay, got it. So it's just the impact of transactional stuff sequentially?

Yes.

Okay. Got it. Thanks a lot guys.

Just to verify, it is translational and transactional in the sense that we have foreign currencies on hand, which we are translating to dollars at quarter end and that impact has to flow through income.

We haven't actually done the exchange.

Yes, it's unrealized. There is some realized portion that happened during the quarter, but most of it is unrealized.

Understood, thank you.

Emmanuel Rosner, CLSA.

Hello, good afternoon, everybody. I wanted to start just with a quick math question. So your comments suggested that if there's any sort of delay towards the end of Q4, this would have an impact of about 800 Model X units per week. And then you're also guiding to about to 1,600 to 1,800 combined volumes of production per week next year. Does that just simplistically mean that you are targeting the mix to be roughly 50/50 between X and S right from the get go in 2016?

It does, although I wouldn't put too much precision on that because what we're going to try to do is to push the production slightly more in the direction of X because people have been waiting a long time for their cars. And then in any given month it could be 60/40 one way or the other. If you look at worldwide demand for SUVs and sedans, it is almost dead even at 50/50.

Some regions, sedan is a favorite, some regions SUV is a favorite. But generally it is, on a worldwide basis, 50/50. It is difficult for us to say exactly what the S/X demand ratio will be until the car is out there, people are experiencing it, and we see what the relative order volume is. But we have so many advance orders on the X that this certainly will be an issue in the early days, and we are going to try to get people their car as fast as we can.

Okay, so I understand the demand aspect, but I guess from a production capacity point of view you think that, as of early 2016, you could theoretically have as many X as Model Ss being produced?

Yes. To put our cards on the table here, we're setting factory capacity to be 1,000 Ss and 1,000 Xs per week. Actual production and capacity are not exactly the same thing. There's always some percentage lower than capacity that is -- or occasionally, it may hit capacity, but it's hard to maintain capacity.

But our goal is, what our internal plan is, I can tell you is that we want the factory to be able to make up to 1,000 Xs, up to 1,000 Ss per week next year in terms of capacity. And then, actual production is affected by real-world issues, so there will maybe some weeks where there is 2,000 produced and some weeks where there is very few produced because we have got factory tooling situations. Hence the 1,600 to 1,800 average over the year that we're predicting.

And it would be fair to say the goal of the factory is to not just produce cars, but it's to produce cars with the right cost and the right quality.

Yes, particularly once we hit steady-state we can certainly do that. It's just difficult during the ramp phase.

Okay, that's very clear. And then, one question on China. Very happy to see that your revised strategy is getting some traction, so what exactly are you doing differently there? I guess besides offering a home charger for the buyers of Model S? Are there any other things you are doing differently in China? And any lessons that you can learn from there in terms of applying that to other regions?

China does have a unique set of challenges. For example, the whole license plate question. To get a license plate to drive a car in a lot of the major Chinese cities is quite difficult. So whether you have electric vehicle exemption in a given city or not makes a big difference because if you don't, then people can buy the car, but they can't drive it.

So it's taken us a while, but we have been successful in getting EV plate exemptions everywhere except Beijing, and we are optimistic about Beijing. We're seeing a Beijing exemption in the hopefully near future. So that's important for China, but not something that one can extend to other parts of the world.

I think it's really that in China, and in most countries, that there is a bit of a slow build of awareness and confidence in Tesla. And depending upon when we went to market in a particular country, that feeling is going to be at a low or a high stage of maturity.

In the US, say particularly California, it's a high stage of maturity. The awareness and comfort with Tesla in California is very high. But that's at a moderate stage of maturity in, say, the Northeast of the United States, and so at the low stage in most of Asia. And the same has been true in Europe.

Basically, it seems like with every country, you have got to build confidence in the brand over time. It just takes time, you can't just have it be immediate. Just because people love it in California does not mean they will automatically love it in other places. You got to build the confidence over time.

We've been in China only slightly over a year.

Yes, just a year.

Understood. And then final one just on the Gigafactory. Can you just give us an update on how things are going there in terms of the preparation? And then there was also, you've been mentioning a bunch of hints throughout the month on potentially wanting to add some space there, I think some capacity. Can you talk more precisely in terms of where would you see capacity go beyond the initial stages?

I'll say a few words and maybe JB can weigh in. Whenever engaging in speculative comments like this, I think it is important just to remember they are speculative and not a prediction that we would have say with very high confidence.

But what we have found is, with the Gigafactory, is that as we spent more and more time on it, we have found we've been able to improve the space efficiency of the production and the overall efficiency by more than our initial expectations. So the net result is that we think in the same volume we can do potentially significantly more output. JB, do you want to elaborate on that?

Yes, I think our plans are still on track and unchanged for the first phases of production, to support Model 3 and to support the Tesla Energy business. But the ultimate production capability of the site is what we believe can go much higher than we maybe initially thought it could. And we do remain on track for construction at the site. We'll have first equipment being installed at the end of this year and planning to start production on Tesla Energy products Q1 2016

And the Gigafactory, we're already in production with Tesla Energy products at Fremont, but that production will transfer to the Gigafactory next year.

And expand and ramp significantly.

Perfect. Great to hear.

Brian Johnson, Barclays.

I've got two questions, one on the Powerpack opportunity and another relating back to Model X. On the Powerpack, I want to get a deeper understanding of where you see your competitive advantage. If I think about simplistically four levels of a stack, in terms of the utility Powerpack solution, with the upper-level being the software to grid interface with the grid and kind of tie-in once it's not needed. Second layer, the battery and management software, and then at the hardware, the inverter and other power electronics, followed by the battery itself.

Where do you see your advantage at each of those levels? How is it important to play at all of those levels? And how do you interface (laughter) with some of the existing middle people, you might call them, or consultants in the industry, or other software providers who might be providing elements of the stack already?

It's a pretty detailed questions. I think one of the -- maybe at a high level, one of the biggest benefits that we offer and where we have a competitive strength is having a system that solves pretty much all those problems together.

I think a lot of other people aspiring to be in this market sell one piece among that entire stack, as you're calling it. And then you'd have to go to different companies to find the other pieces. At Tesla, we're integrating all those pieces together for a very turnkey solution that a utility or a commercial customer can just install.

Basically, plug and play matters even if you're at the megawatt scale.

Yes, exactly. I think, of course, the pricing fundamentals, starting with the battery itself, are really the foundation of that, but we have a lot of expertise and a lot of know-how in power electronics, and software as well, that we've built on the car side of the business for many years.

And how about the software to interface with the grid and determine when to charge and when to discharge?

That is something we are working on now. And we are also working in partnership with many different utilities on this. There's not a universal point of view on exactly where that control and dispatch should live. A lot of utilities want to be very involved in that themselves. So we're basically setting up the tools and the infrastructure so that they can control in a way that is familiar and the most convenient for them.

It's essentially like an API, so that their utilities system can essentially put or call power to the pack. And then they can query the pack for information about its status. But as JB was saying, it does have to interface with quite a heterogeneous set of systems around the world. That's why you have to have basically an API interface to the pack -- so that the utility system can request power or put power to the pack.

Okay.

And maybe just one more comment. At the commercial level, that is something that Tesla is engaged in much more directly, for things like demand management and those type of applications. But it's a bit of a different answer depending on which market we are in here.

How are you getting utilities comfort with the number of cycles in the lifetime of this? Are they taking your word for it or is that something they are seeking to explore through pilots and/or their own high-intensity testing of your battery packs?

We have a lot of data, actually. They don't just have to take our word for it. A lot of these tests have been running for a long time, and we can show them fairly hard cycle data and some lifetime data. So that's been very helpful.

And also, of course, there's all the field experience from the automotive fleet. That's much bigger than what we are deploying in the stationary fleet and will remain so for a number of years. Having the confidence there and how those batteries have aged and done quite well is extremely helpful in building confidence.

Got it. And just final quick question. Model S, you noted some risk to 4Q, depending on what happens with the X ramp up. Earlier in the year, you talked about a 30% gross margin for the Model S part of 4Q, of course that was not for 4Q overall. Where do you stand on that 30% internal goal now?

That 30% goal was before the dollar began to really strengthen at this level. So clearly that has had a fairly large impact on it. And also the mix has had some impact. Our focus also, in the last few quarters, has been much more so on model X.

There are various other cost reduction opportunities that we have that we are hoping to get in Q4 but they could potentially get delayed into Q1. We want to make sure they happen right and we are at the same time focusing on X. So, I would say broadly, if we put aside exchange, we see a trend of improving gross margin despite mix over time.

Okay, but no longer a hard 30% in 4Q?

No, not next quarter basically. But next year, contingent on macroeconomic conditions not going wacky, that seems like a potentially attainable number.

Okay, thanks.

Colin Langan, UBS.

Thanks for taking my question. Can I just follow up on some of the numbers you threw out on stationary storage? The $1 billion, is that orders or reservations? I just want to get a sense of how firm that was. And did you imply that, it sounds like $400 million to $500 million for next year? And is it a few billion for 2017 at the storage market, is that what you said?

Yes, these are reservations, so reservation may be an order or maybe not, but that is certainly what people said that they want. Over 100,000 reservations have been placed for Powerwall and Powerpack. And of that, you can spec that as to what would be, how many Powerwalls, how many Powerpacks, per reservation. It's likely to be more than one -- particularly in the Powerpack case.

Yes, that leads us to think 40 or 50 for Q4, and maybe 10 times that number next year, and then 5 to 10 times that number in 2017. But as I said earlier, as we get further away in time, the numbers are more speculative.

And what would that put you in terms of market share in storage at that point? Will that be most of the market, stationary storage in 2017?

We don't really know.

The market is definitely growing very quickly. I'd say we're speculating on what the actual entire market is going to do and how that is going to grow. So that's --

I don't think you can draw a lot of conclusions from like what is the market -- how much of the stationary storage was sold last year. Just the beginning of electric car production for Tesla, people were trying to say, well, how many electric cars have were sold last year, almost none. Therefore, Tesla will sell almost none. That's to summarize what the vast majority of predictions were about Model S.

Yes, the prices that stationary storage was selling at last year, for instance, are so much higher than where they will be in 2017, that you can't extrapolate those two.

The demand for stationary storage increases at quite an extreme exponential as the cost per kilowatt hour decreases. The utilities used in terms of levelized cost of energy. And depending upon where you are in the world, that number in some places is very high, some places is quite low. As you start to approach the average value, the demand basically scales into multi-terawatt hour range.

Grid parity is the wrong concept here, but it's maybe somewhat of an analogy to think about.

At grid parity, the market is staggeringly gigantic.

And what is your all-in cost? I know it's $250 for the Powerpack. With installation or anything, do you have any estimate of what it actually costs like a commercial or utility, all in?

It depends a lot on the scope and scale and how many other pieces of that installation are bundled together. So we haven't quoted or listed those numbers since they vary so much from one installation to the next. The battery cost is really what matters most in the economics, so it's where we have listed those prices.

Any sense of the high-end and low-end range for installation?

Nothing that we'd be ready to share quite yet.

Okay, and just last question. How should we think about the margin profile over the next few years with this ramp? I believe you said Q4's pretty low. Should that meaningfully improve? And when do you get parity with your gross margin?

We are getting quite speculative about the battery business. In the early days, the factory gross margins that are on the order of 15%. Over time, that could rise to 25% or maybe 30%. We just don't know that quite yet. We have to look at what the price list and the demand is to understand whether -- where we should we be pricing and what's the right gross margin to aim for.

Okay, thank you very much.

Ben Kallo, Robert W Baird.

Thanks for taking my question. As far as the Model X goes, last time, Elon, you talked about configuration in July, and I understand with the slippage, but when can we start thinking about configuring cars who have orders out there, customers have orders? And then with about a month away from deliveries, when do you expect to show it to people?

We're probably into upper configuration two or three weeks of the X. So that should go live on the website before the end of the month. In terms of the initial deliveries of the X, that's consistent with what we predicted on the last call, which is end of September.

Okay, and then as we look out to the target of 500,000 cars in 2020, and some of the things you learned recently, do you still stand behind that number? And what gives you confidence in looking ahead to that number and ramping up production?

Sure, I do remain confident about 0.5 million cars in 2020, and maybe being able to exceed that. It's worth noting that, so 2020, that's five years from now. If you go five years in the past for Tesla, we were producing 600 cars per year. Now we can produce 600 cars in three days. So I think you're going from here to 500,000 cars a year is much more a leap than what we did over the past five years.

Got it. And then when we think about your currency exposure -- where are you guys at on thinking about moving any kind of manufacturing outside of the US, or any parts or manufacturing or additional manufacturing?

It depends on what time frame. In the next few years, we expect to be focused on Fremont and the Gigafactory in Nevada. Long term, again we are getting very speculative here, maybe three- or five-year time frame it's going to make sense for us to think about localizing production in different markets. Having a factory in Asia, a factory in Europe, other factories in North America.

So, in order to go beyond the 500,000 units a year, that's what we would need to do. Our Tesla factory in Fremont and near Reno is that sort of size for the 500,000 level. Might be able to do a little bit more than that, but let's say 500,000 should be very achievable. When it was operating as NUMI, it did roughly 0.5 million vehicles a year, so for us to assume a number is quite reasonable, I think, without adding new factories. So the new factories would be for going past 0.5 million.

And my final question, a headline after this call might be, Tesla is going to raise capital. Just from your comments, I think some people are going to walk away thinking that. I just want to make sure that you guys can set the record straight if that's in the cards in the near term or if it's not?

We can't comment on that specifically. Next question.

Andrea James, Dougherty and Company.

Hello, thanks for taking my questions. What have you learned or discovered between the time when you set that 55,000 unit goal and now when you're saying it's better to have some more breathing room on the ramp? And also, I guess one more, along with that, is there any supplier in particular that is concerning you?

I don't want to name specific suppliers. Our biggest challenges are with the second row seat, it's an amazing seat, like a sculptural work of art, but a very tricky thing to get right. The Falcon Wing door actually seems to probably not be a critical path item. There's some interior components, interior trim. That's our -- probably should be on the critical path.

But it's always hard to say exactly what lies in the critical path because these things tend to play a schedule leap frog, and it's kind of a set of constraints that, one day it's this constraint, then the next day it's another constraint. The pace of progress is really dependent on which supplier is the slowest and least lucky. The supplier has unexpected challenges which can range from force majeure to suddenly having to redo a design because the initial design was wrong.

But when you have a complex product like the Model S that is dependent on thousands of suppliers, you can say that the pace of progress is determined by the 1,000th least lucky and slowest. But if we knew in advance which ones those would be, we would take action. So yes, it's --

That's helpful, I think everybody understands that you're -- an investment in Tesla is a company that's learning as it goes along. It's always interesting to see what you're discovering as you attempt to build.

Can you walk us through what you accomplish during the factory shutdown in the quarter? It looks like you really did a lot of work in one week. Can you help us understand the significance? And then also maybe paint it in terms of what you are doing in the factory this year that's going to actually be repurposed or can even be used toward Model 3 production?

The refueling was both for the X as well as for improved efficiency of S production. And I think we got a lot accomplished there. For Model 3, the biggest single item is the paint shop. The paint shop is size scaled to do 10,000 cars per week, so we've laid the foundation for that rate in the paint shop.

I think there's also room for significant increases in our foundry, in terms of casting, and we've also made a significant investment in stamping and some advanced level sheet forming technology that is not stamping. Anything you want to add to that?

I think in plastics and paint, especially paint, we made modifications there on the existing paint shop. We're getting the new one ready. I think we've essentially added capacity in many different shops ahead of Model X, which required some production interruption to do it right, including getting our production control and inventory management processes much better. You just can't do that when production is running.

What about the investments in the drive unit manufacturing? If any of that small drive unit going to be used for Model 3? It seems like you are really scaling up there.

We certainly learned a great deal going from the original Model S drive units to the small drive unit. It's dramatically easier to build, it's much more automated. That said, I think we will do a further revision for the Model 3 and essentially go to third-generation production technology for the Model 3.

Yes, maybe one thing to keep in mind is the small drive unit capacity is quite an excess even in vehicles because of all-wheel-drive. As that is ticked off, we've increased the capacity of the small drive unit in production that you see as well.

Did you need two of them? And finally, I guess Deepak, I'd love for you to say as long as possible, but I'll just ask this anyway. What is the status of your CFO search?

(Laughter) we're interviewing candidates all the time. Deepak is part of the interview process. Yes, I think we are talking to some interesting candidates. I'm not sure if there is anything more to say about that.

Clearly, I'll stay around to ensure there's a smooth transition. We're continuing to talk to lots of candidates and make sure we find the right person to come in.

Okay, thank you. Actually, do you have any progress on regional heads of sales? And that's my last one, thanks.

We do have a regional head of sales for Asia who came to us from EMC. And then, there was some press about I had a North American sales who came from Burberry, and then just because we have one person from Burberry, then people think we're copying Apple, which is ridiculous. He's great, but just one guy. And then we're continuing to search for our head of Europe sales.

Thanks so much.

George Galliers, Evercore.

Good afternoon, and thanks for taking my question. I have a strategic question for you. It looks like in the next two years we'll see a fist of new and improved actual electric vehicles from your peers, ranging from mainstream models from GM ranging to more premium labels from Audi and Porsche. Strategically, how do you think about future electric vehicles from other ODMs?

A, do you view them as a scratch and competition in a limited EV market? B, do you see them as contributors to the common cause, raising awareness of EVs as more consumer EV adoption increases market size? Or C, given certain compromises on the competitive assets to date, do you actually see them as the opposite? I.e., they form negative pressure to consumers' minds with respect to electric vehicles' costs, practicalities and performance?

I think it was the first multiple-choice I've gotten as a question. If think if you just look at our comments in the past, what we have said is that we're really excited to see other car companies do electric vehicle programs. What's been done to date is not much. They generally have been fairly small programs, and often, just set to achieve a regulatory minimum.

So that hasn't been great thus far, but I'm encouraged by what I see about the future plans. They sound like they are headed in the right direction. And I would be super happy to see the whole industry go electric. We have open-sourced our patent, so that those would not be an impediment, and perhaps could be helpful. And, yes, it would be really great if the whole industry would just go electric as soon as possible. In fact, our view is that the whole industry will go electric eventually, they really won't have much choice. But the sooner they go electric, the better.

Thank you.

Thanks.

I think that was the last question in the queue. Thank you, everyone, for joining us, and we will talk to you next quarter. Goodbye.

Thank you.

Thank you, gentlemen, and thank you, ladies and gentlemen, for joining us today. We hope that you found today's event informative. This will conclude our call. You may now disconnect and have a wonderful day.