ASML Holding N.V. (ASML) Q2 2013 Earnings Report, Transcript and Summary

Ladies and gentlemen, thank you for standing by. Welcome to the ASML 2013 Second Quarter Conference Call on July 17, 2013. [Operator Instructions] I would now like to turn the conference over to Mr. Craig DeYoung. Please go ahead, sir.

Thank you, Kristen, and good afternoon and good morning, ladies and gentlemen. This is Craig DeYoung, Vice President of Investor Relations at ASML. Joining me today from ASML's headquarters in Veldhoven, the Netherlands, is Peter Wennink, ASML's new CEO; and for the final time, Eric Meurice, ASML's outgoing CEO. I'd like to take a brief moment to publicly say to Eric that on behalf of the ASML IR team, it's been our extreme pleasure to have worked with and for you over the last 9.5 years. And we wish you and your family the best in the years to come.

Thank you.

The subject of today's call is ASML's second quarter 2013 results. This call is also being broadcast live over the Internet at asml.com and a replay of the call will be available on our website for approximately 90 days. Before we begin, I'd like to caution listeners that comments made by management during this conference call will include forward-looking statements within the meanings of the federal securities laws. These forward-looking statements involve material risks and uncertainties. For a discussion of risk factors, I encourage you to review the Safe Harbor statement contained in today's press release and presentation, both found on our website at asml.com and in ASML's annual report on Form 20-F and other documents as filed with the Securities and Exchange Commission. The length of the call will be 60 minutes. Now I'd like to turn the call over to Peter Wennink for a brief introduction.

Thank you, Craig, and welcome to everyone. I would like to start by providing an overview and some commentary on the second quarter results and our view going forward. At the end of my introduction, Eric would like to use his -- this opportunity to review a few key issues that he considers important in what will be his last contribution to the results conference calls of ASML. Our second quarter results came in very much as planned. We saw this quarter that foundry system sales equaled memory system sales, with each representing about 45% of total system sales. The quarter shipments were highly skewed to emerging systems, driving the ASP of all systems recognized in the second quarter to EUR 24.1 million versus EUR 23.3 million in the first quarter. Service and field option sales saw an uptick to EUR 271 million. However, that included one month of Cymer's share to sales. Second quarter net bookings came in at EUR 1,065,000,000, which accounted for 38 systems, excluding EUV, with booked average selling prices of EUR 28 million, impacted again by a high percentage of immersion systems, with strength in the logic and the memory sectors. The next several quarters' bookings will reflect this secular ASP trend as well, as we expect increased demand through the second half of the year with the company's leading-edge systems for logic and memory applications. Our system order backlog at the end of the second quarter grew to EUR 1.4 billion, with total systems remaining at 42, which excludes EUV. The backlog profile at quarter's end shifted more towards memory but is still heavily leaning towards logic as a result of the recent bookings strength in this sector. As I mentioned last quarter, we expect more substantial working capital requirements in support of our EUV and 450 millimeter programs over the coming 12 to 18 months. Longer factory cycle times for these new products and investments in facilities and prototypes to deal with the anticipated volume ramp will obviously require additional capital. Short term, we will also see work-in-process increases to support our sales ramp in the second half of this year. Despite the requirement for additional capital, we will keep our promises to return cash to our shareholders. The company paid approximately EUR 210 million in dividends, EUR 0.53 per share in this quarter -- or in the second quarter. And we bought back approximately 1.4 million shares, which was part of the EUR 1 billion buyback program that's announced last quarter. Going forward, we will be guiding the combined ASML and Cymer business as one business. For the third quarter, we expect sales -- total sales to be about EUR 1.3 billion, with a gross margin of about 40%. The combined R&D for the quarter will be about EUR 245 million. And other income, which, again, consists of contributions from the participants of the Customer Co-Investment Program, will amount to EUR 17 million. The combined SG&A is expected to be EUR 91 million, but that includes about EUR 10 million of onetime reorganization charges. As mentioned in our press release, we will incur EUR 60 million of nonrecurring, noncash accounting charges in the second half of the year, resulting from the Cymer purchase price allocation, of which about 75% will relate to the third quarter earnings. Without this purchase price adjustment in the third quarter, gross margin guidance would be between 43% and 44%. As to our business, we continue to see sustained demand from the logic sector as a first foundry customer begins with initial aggressive deliveries in the second half for volume ramp of the 2016 of 40-nanometer combined nodes. In addition and as mentioned last quarter, inquiries for litho system lead times have begun to turn to real orders from a broad base of DRAM makers due to what appears to be strong demand for mobile DRAM. We are also seeing NAND orders later in the second half for fab capacity additions that have been planned for sometime now by a large NAND customer. We now believe that this incremental combined memory spend could drive 2013 sales, excluding Cymer, to a level of up to EUR 5 billion in 2013. On top of this, we expect Cymer sales to contribute approximately EUR 180 million. We see also growing demand for our Holistic Lithography YieldStar metrology solutions, with the demand for our in-line and stand-alone metrology tools that could reach a shipment total of 100 systems this year. We expect more than EUR 350 million in total Holistic Lithography product sales in 2013, divided over system sales and service and field option sales. The buildup of 11 EUV systems continues in the factory and we now plan for the shipment of 5 NXE:3300 systems this year versus our prior plan of 6, which was affected by a 5- to 6-week delay of one of the system modules. The first systems are in the process of shipment and installation at customer sites now. With most systems shipping later this year and with revenue recognition only possible when both scanner and source have been shipped and installed at the customer site, we expect now to recognize about EUR 210 million in revenue for EUV, still allowing us to forecast the 2013 revenues at up to EUR 5 billion, which excludes Cymer, but that's due to the underlying strength of our non-EUV business, as mentioned earlier. And with that, I would like to turn it over to Eric for a review of our immersion and EUV programs and his last comments.

Thank you, Peter. I wanted to focus my last comments on the company trajectory, which both Peter and I supported for all these years and which is based on a very simple strategy executing around a huge R&D investment for technology leadership. This leadership hinges on 4 projects basically at this moment. The first project is a relentless, continuous improvement of our XT and NXT architectures, so basically today, immersion. Now Project 2 is a buildup of an application product suite, enabling and differentiating the NXT products. Project 3 is introduction of the next-generation, game-changer technology, this is the EUV, which will be required by the industry for the next 15 to 20 years. And the Project #4 is the capability to insert costs effectively when the industry requires it, the next wafer size, 450 millimeter, so far planned for production not before 2018. So on Project 1 and 2, as you saw from the press release, we are, again, proving our leadership case with our new NXT:1970, which is to be introduced in H2 of this year. This is a 250-wafer per hour system, with overlay having been achieved down to 1.9 nanometer. And with a suite of applications, product package is valued this year at above EUR 350 million of sales and growing further and fast. With these capabilities, we will be benefiting from current semiconductor process complexities for 20-nanometer logic in DRAM. And these processes are more litho-intensive than ever and from the added complexity of inserting FinFET transistors at the same pitch, which will be called 14-nanometer or 16-nanometer. Even the complexity of vertical NAND which, although on paper, should not require state-of-the-art lithography, in practice, however, requires significant risk-reducing, best overlay and focus capability and requires significant numbers of system as NAND chips volume grows. On Project 3, which is EUV, we made fundamental in the first half of 2013. We have now enough data to confirm the possibility to ramp system in production in 2015 for semiconductor production in 2015 and '16. So I know a question has been asked about the progress of power and et cetera during the quarter, but I want to remind the audience that this quarter was a quarter of transition between the platform 3100, which was limited in power to about 55 watts, to the new architecture, 3300. So this is basically what has happened in the quarter. And within this quarter, we've been able to accumulate certain numbers of data, like that we have now more than 44,000 wafers processed in the field. We have checked further the source potential concept in the lab. We have factory experiments on the new source, a concept called MOPA PrePulse. The new NXE:3300 scanner, which I just talked about, have confirmed 10-nanometer logic and 1x nanometer DRAM overlay and focus capability. More importantly, even the EUV light sources are now running, I would say, routinely at 55 watts, with production-worthy dose control. So this is an equivalent of the machine capable of 40 wafers per hour in the production of NAND. So this performance supports our targeted production throughput for 2014 of 69 wafers per hour, upgradable within 2 years to 125 wafers per hour. This is our commitment to the customers since about a year or 1.5 years. So the current EUV data, the current progress is basically triggering now customer decisions. In logic, 1 or 2 layers of EUV for the so-called 10-nanometer node will deliver already significant die shrinkage compared to any alternative technology inclusive of double patterning or triple patterning. This die shrink will improve the cost of the die, obviously, the power consumption capability and the speed. This justifies -- this value, the shrink factor, justifies EUV as a good technology, even if the throughput of EUV machine is lower even than our 2015 commitment. If, however, the throughput of 69 wafers per hour upgradable to 125, as we committed, does materialize in time as expected, then more EUV layers in logic, up to 7 layers, will deliver process control, yield and cost savings. So first, you have shrink, die shrink, which is 1 or 2 layers. And in addition, you would have your process control, yield and cost saving. And this capability will also enable the technology of EUV to be used on 3 layers in DRAM. So these 2 facets are fundamental to understand, in fact, what will come later. The 3 technology demand drivers that we just mentioned, if we hit the throughput targeted of 125 wafer per hour, we would expect to ship maximum -- our maximum EUV capacity of 25 to 30 tools in 2015. After the current throughput of only 40 wafers per hour, we could still plan to ship 12 to 15 EUV tools, which will be, in fact, used for these 1 or 2 layers of logic, which are compensating emerging tool volume for the layers not handled by EUV. So to finish, I would like to say that it has been my great pleasure to have met many of you over the 9 -- the last 9 years. I would like to thank you for your support of our business model, as measured by your very, very, very detailed interest in lithography. I wish you all the best and I certainly hope that our path will cross again.

Thank you, Eric. And with that, we would be pleased to take your questions. Kristen?

Thanks. Let me just say before we start here that, ladies and gentlemen, the operator will instruct you momentarily on the protocol for the Q&A. [Operator Instructions] So now, Kristen, could we have your instructions and then the first question, please?

[Operator Instructions] Our first question comes from Francois Meunier from Morgan Stanley.

It sounds from your introduction, Mr. Meurice, that you're more bullish on EUV and less bullish on quadruple patterning. Is it really proving that difficult to use quadruple patterning for your customers at this junction and maybe just thinking about using EUV maybe earlier than what they were thinking before?

Let me answer that question, Francois. The discussions we've had with our customers over the last few months have taken another step in intensity, talking about EUV introduction, because you seem to suggest that the difficulties of customers have, when they think about not the 20-nanometer node but the 10-nanometer node, using multiple patterning strategies, those difficulties are quite significant. Just to give you an indication, the multiple patterning strategies, we believe, after consultation with some of our customers, that the shrink factor that can be achieved by using multiple patterning strategies at 10-nanometer is only going to be 1.5 to 1.6, where 2 used to be the more slow rhythm. Doing only 1 or 2 EUV layers, in effect 1, brings that 1.5 up to 1.8. And if you use an additional number of layers when we have a higher throughput, then it goes back to 2. So on top of that, the electro characteristics by using EUV are also significantly better. So yes, I think when you think about the cost advantages, the speed advantages and the power advantages, using the maximum shrink factor by using EUV, it is not a surprise that we have intensified the discussions with our customers. And customers wanted to intensify the discussion on the introduction of deep UV at the 10-nanometer node.

Okay. Can I ask another question about 3G NAND, actually, because -- can you give us maybe a number of what would be the lithography intensity for 3D NAND versus normal NAND?

Yes, for 3D NAND, we believe that the litho intensity is about the same. There is an upside because there are a significant increased of number of layers. And an upside is not particularly for litho, litho will be about the same level of intensity, but will be for other, let's say, process steps that position etch. So from a litho point of view, we don't see a major difference with what we see today also because those 3D NAND fabs are very, very likely going to be greenfield fabs. And in those greenfield fabs, when you buy leading-edge litho, you also buy for the extendibility, although you could argue that for the introduction node, you could have a slightly relaxed immersion litho requirement. It will always be immersion, by the way. And customers choose for the extendibility over the more leading-edge machine. So yes, I would say there's upside for other process equipment makers, but for us, the intensity will be the same.

Our next question comes from Gareth Jenkins from UBS.

A couple of questions around one subject, on YieldStar. So I just wondered if you could give us a sense of how much revenue you expect from YieldStar specifically. And given you talked about 100 tools, what that means in terms of capacity you expect with 30-nanometer coming online next year. And then related to that, I just wondered if you could give us a sense of the margin impact of YieldStar. Is that in line with your Holistic Lithography products historically, so high double digits?

Thank you for asking 3 questions, Gareth. You probably have a follow-up. The question, how much revenue in YieldStar, YieldStar products sell for about $2 million each and we plan to ship about 100 of those this year. But they're part of the total EUR 350 million number that we quoted in the introduction. So YieldStar can be purchased as a separate stand-alone tool. But increasingly and especially for the 20-nanometer node, we see that most advanced applications used as an in-line solution with our other integrated metrology applications. So that means that it is part of our EUR 350 million target for this year. And Eric already stated in his introductory statements that this Holistic Litho application group has a suite of products that is particularly usable for the 20-nanometer node and beyond. So that means that we have very good expectation on the growth rates of that business. On the capacity expansion that you asked, yes, we will do about 100 systems this year. The capacity expansion for YieldStar next year will go to 150. And that is a reflection of what we think the needs in the industry are for the in-line metrology solution. The impact on the gross margin of the total Holistic Lithography group is significant because there's a lot of software in there. So they have margins, without going into detail, that are better than our corporate average margins. And of course, when the total sales of that business unit grows, I think it will have a positive impact on the gross margin. I think it is evident if you think about -- I mentioned in my introductory statement that without these onetime accounting charges on the purchase price allocation of Cymer, we would still do between 43% and 44% gross margin on our EUR 1.3 billion sales number. That is -- given the fact that if you've been here lately, you see we have quite a significant infrastructure for EUV. All that cost is currently also taken up by the fact that we are improving our margin for our base business by driving cost down but also by increasing value for the products that we ship to our customers, of which the Holistic Litho product program is a very important part.

Our next question comes from Simon Schafer from Goldman Sachs.

I actually wanted to ask about sort of how you think about the margin accretion as it relates to the Cymer acquisition now that you've closed and so you're sort of getting an initial feel of what the expenses are. And I understand how in the near term, of course, you've got some of the accounting to take care of and some of the incremental costs are also rising because you're in the that EUV ramp phase. But more broadly, how are you thinking about the medium-term accretion in terms of Cymer? I think last time on the call, you said looking for sort of net accretion, I guess, within a couple of years of closure. Any comment on that would be really helpful.

Yes, I think when you look at the margin accretion for Cymer, you have to split the Cymer business in 2 parts. One part -- and it's basically also the way that we run the businesses now. We have the deep UV business, which is a very good, stable, profitable, cash flow-rich business, which that will be added to the overall corporate margin of ASML. And you can also look at the historical Cymer data of 1 or 2 years ago and you would get a clear indication of what I would mean. Now clearly, Cymer has stepped up, over the last 1.5 years, the expenses for the EUV source development. Stepping up those expenses was a necessity after, I must admit, a couple of years of underinvestment. We have made great progress over the last 9 months, but it was, of course, due to the fact that we were looking at, let's say, a merger of the 2 companies. But note in the last part, it was used -- it was based on the fact that we spent significantly more in research and development. And you could argue that, that has been a catch-up of the underinvestments of the years before. We do expect the next 12 months, 12 to 18 months, we will need to keep doing that. So when you look at the results for the month of June, as we showed in the press release, you see that the operations from Cymer did not contribute yet to the net income of the company. And that is because we are using the cash flow and the profitability of the deep UV division to fund the EUV source, the development cost for the next 12 to 18 months. After that, however, we see that significant integration possibilities are there on the R&D cost. It has to do with the fact that we have been developing an EUV source on 2 sides of the ocean, so to say. We have close to 1,000 people now working in R&D, of which more than half here in Veldhoven and slightly less than half in San Diego. Now you can imagine, if you do that as 2 separate companies, that is not as efficient as if you would do it as 1 integrated company. And that's what we're working on. That will take this 12 to 18 months to really sort that out. Then I do expect a significant efficiency in the R&D cost. And by that time, we're in 2015 and we'll start ramping our EUV systems, which should then start generating the gross margins to cover for those R&D costs. So it is a long answer but, in fact, it actually says we need 12 to 18 months for integration to make sure that we get EUV source to the powers so that we can deliver 125 wafers to our customers. And that will be a short-term impact. And I think that's basically it.

Makes sense, great. My follow-up question, actually, just a quick one, is along the same lines in terms of how that shapes your thinking on capital returns. If I understand correctly, I think in your remarks, you basically said, look, as part of this ramp, of course, your capital consumption, specifically working capital, will probably rise a little bit still. But what sort of outflow do you have in mind and how does that shape your ability to buy back more stock or choose another method of redistribution?

Yes. What we did over the last couple of years, basically, if you look at the numbers, everything that we added above our cash targets, we get back in terms of dividends and of share buybacks. Now part of that cash flow for the next 12 to 18 months, we will have to use for the investments in the production facilities and in prototypes for the EUV ramp and 450-millimeter. But I can assure you that there's money left. There's money left to do the dividends and to do buybacks, so we will continue with that. But if you look at the historical pattern of the last couple of years, where we basically distribute our net income, for the next 12 to 18 months, we will have to use part of -- the part of that to build infrastructure for our earnings power going forward. And I think that is an inevitable fact when you start ramping big new product lines like EUV and like 450. But I don't expect it to last longer than the next 12 to 18 months.

Our next question comes from Weston Twigg from Pacific Crest Securities.

Just one thing I was hoping that maybe you could help us with is if customers need EUV at 10 nanometer, at least for a couple of critical layers, can you give us a feeling for how a customer can rely on ramping a tool that hasn't yet been proven in HVM? And how does that qualification process take place when that ramp is really only 2 years away at the leading logic customer?

That's a very good question. This is why I started the answer to the first question off with the fact that we have taken the level of introduction with our customers to new level of intensity. It basically refers to your question. In order to make sure that we can do that, we have teams on site that make it -- they make it very clear and transparent what the tool can do. And based on the performance of the tool and the ramp of that performance, the customer is going to scale, basically, their activity. So it's going to be a very close coordination and cooperation between ASML engineers and the customer engineers. Also, what is important, by the time that we ship those tools and the customers will calculate how many tools that they need, there will be a certain level of safety buffer in there in terms of what those tools can do now. And under what we believe is a realistic assumption that we will get to the throughput and the reliability numbers that we are currently planning, it is just a temporary buffer investment that has to be shared between the customer and between ASML. So we're just going to hedge the customer together with us on making sure that the output capacity of what they are planning on the 10-nanometer ramp is, in any case, secured.

Okay. That's helpful. And then just as a follow-on, I was wondering if maybe you could also help us understand a little bit about the cost of ownership of an EUV platform beyond just throughput, things like power consumption and consumables and how that's coming into play in terms of customer decisions.

That's a pretty complicated decision that customers need to make. But I can assure you that customers are focusing on the, let's say, the cost benefits on the product level that makes them to decide for EUV. Yes, there are certainly infrastructural changes that need to happen in those fabs. You mentioned things like power consumption. You mentioned things like maintenance. That is all going to happen, but that is included in the cost calculations that our customers do. And that means that with those indications that we gave them, which are many, so I'm not going to go into detail there, they came to the conclusion that the 1 layer -- 1 or 2 layers for the shrink only is absolutely a necessity. And for the further layer growth, let's say 7, it is all driven by the economics. And then part of the running cost, as you mentioned and the throughput are all part of that equation. So for the shrink-only step, I would say, that is less of an issue than when you want to move to more layers because then it is really an economical decision.

Our next question comes from Satya Kumar from Credit Suisse. Satya Kumar - Crédit Suisse AG, Research Division: I was wondering if you could comment a little bit about the DRAM upside that you talked about earlier. Do you now expect to see a higher bit supply growth for DRAM as a result of the upside in orders that you're seeing? Or is there something else also happening in the DRAM in terms of lithography intensity? Because some of your customers are talking about moving to double patterning and we do know mobile DRAM has a high die size. So could you talk a little about the impact on supply growth and whether these orders are for technology or capacity?

It's a very good question. If we think about DRAM in terms of -- if you're asking a question specifically on DRAM, I suppose, then yes, there is definitely a case of higher litho intensity for the next nodes. There are more layers and more immersion-related layers and, also, double patterning comes into play. Now if you combine that with what we have seen over the last 18 to 24 months in a situation where the DRAM market was in overcapacity and the economics in that industry were not very good, we also saw the tendency to basically do rotations out of DRAM into a leading-edge NAND. So if you combine leading-edge litho out of the DRAM into NAND with the introduction of the new products requiring higher litho intensity, then it is not a surprise, when you look at the capacity of wafer starts, that even with the additions that we see today, we do not expect that the wafer starts capacity will grow this year. Actually, as a matter of fact, at best, it will stay flat, but it could even decrease a little. And that is because of the 2 reasons that I just mentioned: litho intensity for the new nodes; and, basically, the rotation of leading-edge immersion from DRAM to NAND. And that is escalated by the fact that, as you mentioned, we believe DRAM mobile chips are a bit bigger. Not that much bigger, it's about 5% bigger. But as you said, that's where the demand is. If you put that all together, it's not a surprise that we currently see the demand from the DRAM customers being that high. Satya Kumar - Crédit Suisse AG, Research Division: Very helpful. And then on the foundry side, is the order send that you're expecting to see -- I mean, especially if I look at the guidance implied for Q4 sales, it seems like it's significantly higher than Q3. Is that concentrated towards one foundry customer, or are you seeing activity from multiple foundry customers? And broadly speaking, what do you see as the current capability of the foundry part of the industry in terms of the readiness to ramp FinFET-based designs?

Yes. When you look at Q4 -- it also has a couple of questions in there. If you look at Q4, definitely, you can do the math. Our guidance was our guidance for Q3. Yes, Q4 is going to be a large quarter. That means that from a logic point of view, there's not been a lot of change. I mean, it is largely the memory customers that have upped their demand on us. And as we all said in the previous call, we have decided to build some buffers at the beginning of the year, so that's going to help deal with that upside in the relatively short time frame. But for logic, I said it in my introduction speech, there is one large foundry customer that is currently aggressively ramping the 20-nanometer node. And clearly, we have discussions with all the customers also. I think it is inevitable that all the customers will also follow in this respect. And your last question was on FinFET. Could you repeat it again, Satya? Satya Kumar - Crédit Suisse AG, Research Division: Yes. Just how do you see readiness of the foundry industry in terms of its ability to ramp FinFET?

Well, we don't have that visibility. You should ask our foundry customers.

Our next question comes from Mehdi Hosseini from SIG.

Eric, best of luck with you at the next endeavor you're pursuing. I'll just follow up to Satya's question. How should we think about sustainability of DRAM spending into next year, especially if the EUV progress were to continue? And I'm under assumption that DRAM is going to be the first segment to adopt to EUV.

Okay. DRAM is sort of in -- sorry, EUV is, on the way that we see it, currently in 2016. So when you talk about the 2014 DRAM ramp, it's not affected by any EUV program, other than that some of our customers will take, indeed, to the first EUV systems, the 3300 systems, to start prototyping and basically preparing for the 2016 DRAM -- sorry, EUV introduction in DRAM. For 2014, like I said earlier, the DRAM wafer out-capacity still stays flat, even with the latest additions. And, I would say, at best, stays flat. So that means that I do expect that also in 2014, we will see some continuation of the pressure on us to buy machines.

Isn't that a material change? Because I was expecting that at least one of the Korean customers were to adopt EUV for DRAM sometime in 2014.

We don't think so. I mean, 2014, they will take tools to start doing the prototyping and to start doing the qualification process. But it doesn't mean in 2014, they'll start producing DRAM using EUV. That is 2016.

Sure. So in other words, in 2016, we should see DRAM and foundry or logic, 1 or 2 layers, putting EUV into production?

Correct. And it could by 2016 that we have mastered the 225 wafers per hour, as we think, and then it will be more in logic than 1 to 2 layers. It will be 5 to 7 layers.

Sure. And then a follow-up just quickly on 60-nanometer. Again, 1 or 2 foundries are putting risk production. As you do your own planning for next year, as you think about your own internal capacity, how are you managing the expectation for 16 -- I'm sorry, for 14-nanometer ramp? Because this is quite different than the Taiwanese customer that has a good solid track record. How do you manage the risks associated with the 14-nanometer ramp that could be significant?

From our point of view, there's no risk. Simple reason, that's where 14-nanometer or 16-nanometer FinFET node we use the same litho machine as for 20. So it's the same pitch. So it's not -- for us, it's not a problem. You could even say it extends that node from a litho point of view from our own point of view with 1 year, which means a better learning curve. So that's okay. So I think the risk for that introduction is with the customer in a sense that, of course, they need to do it. We just need to provide the same litho solution as we did for 20-nanometer.

But for internal planning, there could be 1 customer or 3. Wouldn't that impact your own internal capacity allocation or just planning?

Yes. I mean, when you talk about our internal capacity, our internal capacity is sufficient to do, on a, let's say, stand-alone basis, EUR 1.6 billion to EUR 1.7 billion of sales. So we're not there yet. So let them give us that chance and we'll deal with it.

Our next question comes from Amit Harchandani from Citigroup.

Actually, more clarification from my side, maybe a couple. Firstly, in terms of the EUV demand, if I heard correctly earlier in the call, you said that at the current level of 40 wafers per hour, probably around 12 to 15 tools in 2015 for EUV. And if it goes up to 125 wafers per hour, then clearly, you're looking at 25 to 30. Just wanted to confirm that, that was correct. And secondly, in terms of the non-EUV tool requirement or tool sales, we are probably looking at somewhere, I guess, probably EUR 3.7 billion to EUR 4 billion for this year in terms of non-EUV sales. Do you think that level is maintained as we go towards 2015 and '16, or does EUV start cannibalizing the same?

Okay, good. On your first question, yes, the point that Eric tried to make in his introduction statement is that even if we would be -- if we would stay at where we currently are, 40 wafers per hour, which is, of course, not what our plan is, but even if it would be the case, then it is still economical for our customers to still buy that tool. Why? They can do 1 layer, whereby they can increase their shrink factor from 1.5, 1.6, to 1.8, which is a significant economic benefit, even at that low throughput. That's what he was trying to say. And that would mean that you're going to calculate what would the need for the industry be, in the logic industry, for just this 1 layer, would be 12 to 15 units. Now we expect not to stay at 40. We will go higher. So when we are at 125 wafers or even over 100 wafers, then we will have to use our maximum capacity that we have at Veldhoven, which is between 25 and 30 systems in the year 2015, if the demand could be higher, but we cannot make more. And so in that sense, if that's your understanding, then it is, indeed, correct. On the non-EUV requirement, EUR 3.7 billion to EUR 4 billion in sales, we expect, even with the introduction of significant EUV sales, where we're going to use EUV for the critical layers, there's going to be a significant EUV -- sorry, non-EUV requirement. That has to do with the fact that the complexity arises, the number of layers goes up and what are now critical layers will become noncritical layers. And they will still need immersion machines, some even at a double patterning level. So we don't believe that there's going to be a significant reduction in the non-EUV requirement for our tools. Whether it's going to be EUR 3.7 billion to EUR 4 billion is not known, but when I mean significant, it's definitely not going to have -- it's going to be more than that. So it is too early to give you a direct guidance, but a very significant part of our business will still be immersion.

Our next question comes from Didier Scemama from Merrill Lynch.

My question relates to gross margins. I was just wondering whether you could maybe give us a sense, quantify the headwinds you see on gross margin in the second half and, perhaps, persisting into '14, coming from the increase in capacity for EUV and, in fact, perhaps, even for '15, if that's relevant. And then second, on the split of immersion tools between 28-nanometer logic and 20, whether you could give us a sense of how much is 20 in the second half, how much is still 28, whether it's still shipping, in fact.

Okay. Quantify the headwind, well, like I said, we do have EUV-related costs at this moment because we are building up the capacity and we are increasing the gross margin on our main products. So in any case, that's where we find some level of compensation. It is obvious then where we ship the first EUV tools, the EUV -- while we have a significant increase in gross margin for our non-EUV business, our planning for the EUV gross margins when we start shipping them is in the 20s, in the areas of 20s, while we are clearly above 40 for the non-EUV system. So it's about half. That means that if we have a significant number of EUV tools shipping the next couple of quarters, that means that, that will have an effect on the gross margin of the company and you could actually calculate it. I mean, if we're about half of the gross margin, then it's not that difficult. So it will have an impact. It will have an impact of a couple of percentage points. But that, of course, like we said in previous conference calls, will take some time. We need volume. We need volume not so much in our own factory, which is an element, but more importantly, we need volume in the supply chain. In order to drive cost out in the supply chain, we need a learning curve. And learning curve goes faster when we have a higher volume. So when we follow our volume ramp planned, which actually is using our capacity in 2015, 25 to 30 units and doubling that in 2016, by that time, we should be back at our corporate average gross margins. So it will take some time. But the -- so you will see a gradual improvement as we go until we have reached the corporate average as we see it today. But that is driven by, largely, volume. And yes, we have to take into account the next couple of quarters with some gross margin percentage points impact because of the early EUV shipments.

Can you talk about the mix of 20-nanometer scanners? Is that all you're shipping right now?

No. The mix this year, to give you an indication, I think, of the capacity that we shipped for logic, which was logic of 28 and 40 -- I'm sorry, 28 and 20, about 40% of that shipment capacity was on 20-nanometer, about 60% was on 28. So 28 was largely the first half of the year tailing off in the second half of the year. And 40 -- this was 20-nanometer really ramping in the second half of the year and then accounting for about 40% of our total immersion shipments to the logic industry.

Got it. And my quick follow-up in one word. I think last time or last couple of quarters, you talked to us and you said that you felt like your 20-nanometer capacity at foundries would be, I think, in the range of 40,000 to 60,000 wafer starts per month.

Yes.

Can you update on that? And second, if you could give us your thoughts on how much wafer capacity we could see at logic customers exiting 2014, that would be great.

For 2013, I can answer that question rather specifically. I think with the 40% of our shipments, which I mentioned as an answer to your first question, I think we will add 60,000 wafer starts capacity by the end of the year. Which, of course, customer will take some time to turn that capacity that we shipped into output. But it's about 60k. And that is what we see today. Now for 2014, it's too early because we haven't finalized yet all the shipment plans with our customers for the next 18 months. For the next 12 months, we have, but it only covers the first half of 2014, so I cannot answer your question on the end of 2014 yet.

What is it, then, for mid-2014?

I don't have the data immediately here, but you can follow up with Craig and with Franki. I had the 2013 data in my head.

Our next question comes from Andrew Gardiner from Barclays.

I have a couple more around gross margin dynamics. Peter, you've highlighted the drag at the moment from the Cymer acquisition, from the pressure that you'll see as you ramp EUV from a lower level towards the corporate gross margin. But sort of looking beyond that and thinking about what you've described for Holistic Litho and with YieldStar in that, and I think you guys had said previously that you're aiming for something around EUR 800 million of revenue a few years from now, also there's a positive impact, albeit, perhaps, small, from the Intel sort of part of your Customer Co-Investment Program coming through the revenue and gross margin line. So ultimately, what's to prevent you guys from reporting gross margins into the high 40% range?

Yes, I think it's a very good question. You just hit on, I think, one point. I think the co-investment program impact is more or less an accounting impact. So I'd like to leave that to one side. Yes, I think the impact of the non-EUV product portfolio is going to be more significant because of the fact that the Holistic Litho product portfolio carries a lot of software. So that is going to be added to the gross margin. Now once we are at volume production EUV, which by 2016, if we do 60 units, you could argue that we are there, then we should definitely also see the benefits of that volume coming through the supply chain. But on top of that, we have now insourced the EUV source, which actually means that the gross margin that would normally go to a supplier now goes to the bottom line of ASML. So your high 40s assessment of where we could be is definitely not an unrealistic one.

Our next question comes from Sandeep Deshpande from JPMorgan. Sandeep S. Deshpande - JP Morgan Chase & Co, Research Division: Just, again, actually talking about this gross margin. Your guidance of 40% gross margin, and you said in your introductory remarks that it would be somewhere 43%, 44%, if it was not for the impacts of the Cymer acquisition. So would you quantify how long that impact, the Cymer acquisition, will impact your gross margin so that we could take at it as one-off associated with that and then model beyond that as such? And then I have a follow-up as well. And there is also, I guess, an impact on the gross margin in the short term from your investments in EUV, which are not resulting in revenue.

That's correct. On the latter point, I said in the, let's say the 43% to 44%, is of course the rest of EUV costs in there that are being absorbed by the fact that we have good gross margin on our base business. So that's in there and I said that before. Now the impact in the second half has to do with the fact that in the purchase price -- now I'll give you a small lesson in accounting. I can't help it. But we have to revalue, at the moment of a purchase of a company, we have revalue the assets at fair value. Which, of course, is clear for the intangibles and the earnings power on the deep UV business and the installed base business. But also, you have to do that on inventory. Now while the inventory at Cymer is all on order, it means that the fair value of something that is on order, where you have an order that you cannot cancel, the fair value is then brought up to the sales price. So in fact, the first 6 months that Cymer sells off its inventory, they don't make any money because it's already taken up in, let's say, the opening balance sheet of Cymer. It's been, you could say, jacked up to the sales price. And that goes away in 6 months' time. So this EUR 60 million, about 75% of that is taken in the third quarter and the remainder in the fourth quarter and then it's gone. So it's an accounting adjustment. That's noncash. That has to do with the revaluation of those assets. On the other elements, there is about a EUR 10 million per quarter amortization of intangibles, which also has to do with this revaluation of those assets, which is basically the earnings power, the discounted cash flow of the earnings power of the business that you have bought. That will be amortized over a 15-year period. So it's between EUR 8 million and EUR 10 million per quarter that we will book and that will go into the gross margin line. But I think looking at the size of the company, that's going to be relatively negligible. And as the last element, which is the -- when we acquired Cymer, we had a remeasurement date of the share plans of those Cymer shares that turned into ASML shares. And that, over the next -- until the end of next year, I think first quarter 2015, will create a compensation expense, which we need to book. But that will show up in several lines. It will be in the R&D line because it relates to people, will be in SG&A line because it relates to people, will be in the gross margin line because it relates to people. So that will be spread all over. Those are all accounting adjustments, which are noncash, but it has to -- but it is what it is. So I think that is really the major impact. When you think about gross margins, you have to think about the base business. Now the ASML base business consists of our immersion business and our holistic application business, which are very good gross margins. We have -- the next 2 years, we have the ramp up of EUV, which will have a dilutive effect when we start. It is inevitable, but the numbers won't be that big. So that is ASML. Cymer has very good gross margins because they have an installed base business which is very profitable with very healthy gross margins. And I said earlier, you can refer to the Cymer historical financial statements, so you can see what that is. So from an operational point of view, I think the gross margins are healthy, but they will be impacted by some accounting adjustments, and over the next 2 years by the introduction of EUV. And I think I explained that in the previous answer. Sandeep S. Deshpande - JP Morgan Chase & Co, Research Division: That was very comprehensive. Just my follow-up question is regarding those tools you've talked about in previous questions that potentially, that if you do 40 wafers per hour, you could ship potentially 15 EUV tools in 2015. I mean, does this indicate, based on what you said, that your intensive contacts with some of your customers -- because when you look at the number of fabs some of the customers have, if you're only going to ship 15 tools, it would mean virtually -- if it's 2 customers, you're shipping 1 tool per fab or something like that. Is that how we are looking at the initial rollout of EUV, or is it not going to be -- is it going to be a couple of tools for fab that are required for, say 1 or 2 layers, if you're going to introduce 1- or 2-layer EUV in 2015?

Yes, I think -- and when we think about 12 or 15 units, you talk basically about the way we're planning it now in that time frame, 2015, only 2 logic customers.

Our next question comes from Mahesh Sanganeria from RBC Capital Markets.

Peter, there's a lot of discussion on gross margin, but can you give us an indication of what the OpEx will be for next several quarters? How should we think about modeling OpEx?

I think when you look at OpEx, it consists of R&D and of SG&A. Our R&D spend this year was going to zoom in on EUR 800 million, gross. That was before the contribution of the co-investment program. We said in the previous call that could go to around EUR 900 million next year. Cymer's R&D is just above EUR 100 million. We are now guiding on EUR 245 million per quarter -- for the third quarter. And it is my assessment that we will not go over the EUR 1 billion if you put the 2 companies together. So that is the EUR 900 million next year plus about EUR 100 million of Cymer. And our current plan is to stay within the EUR 1 billion, so that's EUR 250 million per quarter. On SG&A, the run rate is about EUR 65 million for ASML, about EUR 13 million, EUR 14 million for Cymer, so it's just under EUR 80 million run rate. We guided EUR 91 million, but it has EUR 10 million in there as a result of some reorganization charges that we are taking today.

Okay. So one more follow-up. You had a pretty big jump in terms of bookings on the IDM and memory and foundry is flattish in terms of absolute dollar. How do you see that trending in the second half? Is it IDM maybe go down a little bit, memory come up a little bit and a big jump in foundry in the second half? Is that how we should be looking at?

Well, if you listened carefully to the call, I think memory has just started. If you -- I don't think we're at the end of that yet, so I think memory for the next few quarters will probably stay at a good momentum. IDM is pretty predictable, but that basically is a result of the shipment plan and our adherence and their adherence to 6 months order lead time. So that is just going according to plan. And logic, logic will stay strong because we are ramping the 20-nanometer node and there's only one major logic player -- logic foundry that is currently ramping the 20-nanometer node and the rest has to follow. So I think there's going to be a mix whereby logic will stay strong, but also, we will see the return of our memory customers.

Ladies and gentlemen, I think we've run out of time. There was a lot of good questions. Hopefully, they represented the questions that others that weren't able to get on had in mind. If not, I know your history shows you know where to find us. IR is available after this call to manage as many questions and callers as we can. So again, on behalf of the board and management, I'd like to thank you for joining the call today. And if, Kristen, we could have you close the call out formally, we would appreciate it. Thanks.

Thank you, sir. Ladies and gentlemen, this concludes the ASML 2013 Second Quarter Result Conference Call. Thank you for participating. You may now disconnect.