Craig DeYoung - Vice President, Investor Relations Worldwide Peter Wennink - President and Chief Executive Officer Wolfgang Nickl - Executive Vice President and Chief Financial Officer

Kai Korschelt - Bank of America Sandeep Deshpande - JPMorgan Gareth Jenkins - UBS Jérôme Ramel - Exane BNP Paribas C.J. Muse - Evercore ISI Amit Harchandani - Citigroup Mahesh Sanganeria - RBC Capital Mehdi Hosseini - Susquehanna Pierre Ferragu - Bernstein Andrew Gardiner - Barclays Francois Meunier - Morgan Stanley Farhan Ahmad - Credit Suisse Patrick Ho - Stifel Nicolaus Timothy Arcuri - Cowen and Company

Welcome to ASML 2015 fourth quarter and annual results conference call on January 20, 2016. [Operator Instructions] I would now like to turn the conference call over to Mr. Craig DeYoung. Please go ahead, sir.

Thank you, Aaron, and good morning and good afternoon, ladies and gentlemen. This is Craig DeYoung, Vice President of Investor Relations at ASML. Joining me today from our headquarters here in Veldhoven in Netherlands is Mr. Peter Wennink, our CEO; and Wolfgang Nickl, ASML's CFO. The subject of today's call is ASML's 2015 fourth quarter and annual results. Before we start, I'd like to take a brief moment to address some questions from previous calls about our Q&A queue. The process is you'll be advised that the Q&A starts upon the operator's instructions at the opening of the call and not before. Therefore, there is really no value in calling in too long before the call starts in an attempt to get into the queue. And as the operator mentioned, questions will be taken in the order that they are received. The length of the call will be 60 minutes. The call will be broadcast and is 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. Lastly, before we begin, I'd like to caution listeners that comments made by management during the 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 these risk factors, I encourage you to review the Safe Harbor statement today contained in today's press release and presentation 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. With that, I'd like to turn the call over to Peter Wennink for a brief introduction.

Good morning and good afternoon, ladies and gentlemen. And thank you for joining us for our fourth quarter and annual 2015 results conference call. Before we begin the question-and-answer session, Wolfgang and I would like to provide an overview and some commentary on the recent quarter and provide our view of the coming quarters. Wolfgang will start with a review of the Q4 financial performance with added comments on our short-term outlook, and I will complete the introduction with some further comments on the current general business environment and our future business outlook. So Wolfgang, if you will?

Thank you, Peter, and welcome everyone. For Q4 our net sales came in at €1.43 billion. This included system sales of €881 million, of which memory represented 44% and logic represented 56%. Service and field option sales came in strong at €553 million. This part of our business as grows, continues driven by strong demand for holistic lithography option, high value upgrades and the growing installed base. Our gross margin for the quarter came in at 46%, slightly above the guidance. R&D expenses came in at €273 million and SG&A expenses came in at €90 million, essentially as we guided. For the full year, our net sales reached a new record of €6.3 billion, which is up 7% from the prior year and includes over €2 billion for field options and services. Gross margin for 2015 was 46.1%, up from 44.3% in 2014. Our basic earnings per share for 2015 were €3.22, up 18% year-over-year. Turning to the balance sheet. Quarter-over-quarter cash, cash equivalents and short-term investments grew from €2.68 billion to €3.41 billion, driven by strong free cash flow, which was impacted by a significant amount of customer prepayments on orders received. In Q4 we repurchased shares worth €141 million, bringing the total amount for repurchased shares in 2015 to €565 million. Regarding the order book. Q4 bookings came in at €1.2 billion, 31% above our Q3 bookings. Strength in memory bookings continue to be notable with a significant growth in our foundry bookings, leading to a strong and nicely balanced backlog across all industry sectors of approximately €3.2 billion. With that, I would like to turn to our expectations and guidance for the first quarter of 2016. We expect continued healthy memory shipment supported by a strong backlog. Our service and field option sales will be around the €500 million mark again. We expect relatively modest logic shipments in the first quarter of 2016, leading us to guide Q1 revenue at approximately €1.3 billion. As indicated at this time last quarter, we do however expect our logic customer to take shipments in Q2, which will start the ramp of 10-nanometer, and as a result we expect Q2 sales to increase significantly from Q1. Based upon expected customer and product mix, a lower sales volume forecast and lower field option sales, we expect gross margin for Q1 to come in at around 42%. R&D expenses for the first quarter will be about €275 million and SG&A is expected to come in at about €90 million, both roughly the same as the previous quarter. Our annualized tax rate for 2016 is expected to increase to around 13%, based upon a change in tax rules that transfer some tax benefits into R&D credits. Peter will tell more about our 2015 EUV accomplishments and 2016 key performance targets shortly, but I would like to make a few points regarding 2016 EUV shipments and explain a bit further the current and expected situation related to EUV revenue recognition. We completed three EUV shipments in 2015 and started the shipment of a fourth system before yearend. One of the three systems was recognized in revenue during 2015. The other two systems that shipped and the one system, where shipment was started, should lead to a revenue of about €110 million in the middle of 2016 with the balance booked in 2017. For 2016 we expect to ship between six and seven EUV systems. The 2016 shipments will be a combination of NXE:3300, 3350s and 3400, going to both logic and memory customers. As a reminder, we will continue to guide expected revenue timing on additional EUV systems as they ship. And finally, but certainly not without significant importance, ASML paid €302 million in dividends in 2015 and we purchased 6.3 million of our own shares for €565 million, providing a total cash return to shareholders of €867 million during the year. In 2016 we are proposing to our Annual General Meeting of Shareholders on April 29 to increase our dividend by 50% to a level of €1.05 per ordinary share. Today, we have also announced the plan for an additional €1 billion of share repurchases over 2016 and 2017 on top of the remaining €500 million from all prior programs. With that, I would like to turn the call back over to Peter.

Thank you, Wolfgang. As Wolfgang highlighted, we concluded financially a very satisfying year. Expectations for the first quarter sales are approximately €1.3 billion, reflecting continued shipments for 28, 16, 14-nanometer logic capacity additions, albeit at the relatively low level, and shipments to memory customers consisting of a mix of advanced and more mature system types. While first quarter sales are expected to be relatively moderate, we clearly see, as Wolfgang said, a significant increase in combined system and service sales in the second quarter, largely driven by system shipments for the initial ramp of advanced 10-nanometer production processes at our logic customers. While it is still a bit too early to say anything quantitatively about 2016, we do see trends and developments that we believe are worthwhile mentioning. In memory, as evidenced by the fourth quarter bookings and our current backlog strength, our customers have indicated that their system demands will continue throughout the first half of 2016 at levels roughly equivalent to those of Q4. Although we expect that two DRAM fabs will continue to install some additional tools, largely meant to support the next DRAM node, we also believe that 2016 shipments to this application will be down versus a strong 2015. On NAND, we believe that the capacity expansions will be focused on 3D NAND applications. With the announcement of the conversion of a Chinese fab to 3D NAND, that is now largely supported in our current backlog, we expect a flattish level of NAND systems revenues versus 2015. One additional memory development that appears to be worth following is the introduction of the cross-point architecture. While the full opportunity extent of this new memory architecture is still under evaluation, its potential seems significant and could therefore become important to our business, as the advanced processes anticipated in this application are quite litho-intensive. As for logic, it has now become clear to us that introduction of the advanced 10-nanometer node ramp is progressing well, hence the continued and clear customer commitment to ramp this node starting in Q2 2016. The speed and initial size of this ramp can be explained by the value proposition provided by the significant shrink of this node versus the 16 and 14-nanometer node. The ultimate spend levels for logic in 2016 will depend amongst other things, above the level of end-demand and a rate at which our customers will be able to execute their ramp, and is therefore too early to accurately predict this today. For field options and services, we see continued strength in 2016 and this should show growth previously estimated to be in the range of 10%. On the product side, ASML continues to focus R&D spend on lithography tools that are essential to ramp all of the current and advanced processes. With growing litho challenges of complex and costly multi-pass patterning, our recently launched TWINSCAN NXT:1980 DUV immersion scanner, with significant improvements in all key performance metrics, started volume shipment last quarter. And now with its widespread acceptance, it is ramping at a rate greater than any other advanced system in our history. Our holistic lithography products continued to gain acceptance at leading-edge customers. We are using our full suite of immersion process window enhancements and process control solutions to optimize yield at most advanced product nodes. Holistic lithography products are now extending also into EUV processes, with customers evaluating our EUV source-mask optimization software for the development of their 7 and 5-nanometer technologies. And finally on the EUV, as most of you are aware, our 2015 focus has been on improving EUV stability, availability and productivity, the key performance metrics that drive new technology adoption. In several recent public presentations our customers have clearly recognized our EUV progress in these areas. In a four-week customer run manufacturing readiness test at production conditions, we've seen 15,000 wafers exposed with comparable results achieved using the same power configuration at multiple customers. On the raw productivity side, we have a new system configuration, the NXE:3350, that has demonstrated in our factory more than 1,250 wafers exposed in a 24-hour period. Six out of eight systems at customer sites have achieved four-week average system availability of greater than 70%, with one system reaching the 80% volume. However, the worldwide average is currently still lower, indicating that performance stability in the entire installed base needs to be further improved. We believe that our 2016 performance targets of 1,500 wafers per day and 80% total system availability are achievable and that will be aggressively pursued over the course of this calendar year. Now, with that, we will be happy to take your questions.

Thank you, Peter. Ladies and gentlemen, the operator will instruct you momentarily on the protocol for the Q&A session. But beforehand, as I always do, I'd like to ask you to kindly limit yourself to one question with one short follow-up, if necessary. This will allow us to get as many callers on today as possible. Now, operator, can we have your instructions and then the first caller please.

[Operator Instructions] One moment please for the first question, and it comes from Kai Korschelt from Bank of America.

So the first one was just on the second quarter. So I'm just wondering, what level of visibility do you have? What could be the magnitude of, I guess, the snapback? And the reason I'm asking is I think if I just take your bookings in the fourth quarter and if I add maybe €0.5 billion service sales, then it looks like we should be well above €1.6 billion. So I'm just wondering, if that's the right ballpark? And then I have a follow-up.

Well, we don't want to leave you down any particular number, but we chose the word significantly wisely. So we wouldn't do that, if it would be just a little bit up. So we'll leave it at that today. But it's all underpinned by a strong ramp in 10-nanometer, and we also said that that memory will be throughout the first half at Q4 level. So I think you can approximate it from there pretty well.

And then just a quick one on the gross margin. So I think the Q1 guidance is probably below, where most people think it would be, even if we look back at quarters with similar revenue run rate. So I'm just wondering, broadly, what are the reasons? And should we kind of expect that if we do see the recovery in demand and revenues in the second quarter that we should sort of settle back at the 47%, 48% level that I think we've become used to as obviously before EUV?

Yes, Kai, I can address it as well. So first of all, there are no structural changes in a sense that you should be worried that pricing came down or we have any cost issues. This is purely a function of lower volume at one, and then more so a change in mix between products and customers. And then also within the service and field options, the services are a bit higher when compared to the field options, which come at higher margin. And then you can also read that when we say foundry is lower and memory is higher. Foundry tools are usually in the richer configuration going to the customer. Now, if you want to look at Q2, we won't give you an exact number. But when you consider that the volume will go up and the foundry shipments will go up and it's 10-nanometer, it will be pretty nicely configured tools, you'll know in which direction the margin will go.

The next question comes from Sandeep Deshpande. Please state your company name, followed by your question.

I'm Sandeep Deshpande from JPMorgan. My question would be, Peter, on the memory market. You've said that you're going to have a flattish trend in the first half in the memory market. Do you see this into the second half, substantially correcting or you don't have visibility at this point into this segment? Secondly, I have a quick question on EUV as well. I mean, clearly, EUV is progressing much better at this point in terms of throughput and you've given some of the statistics. What timeframe do you see EUV actually being built for production volumes?

On the memory market, like I said in the introductory comments, we have very clear visibility given by our customers, also evidenced by our bookings that we received in the fourth quarter, on what we are going to ship in the first half, which is focused on also the introduction of the new DRAM nodes. So you could say it's clearly a technology transition that is driving the shipments in the first half. We don't have that visibility yet on the second half, but it is our current thinking. I can only give you some indication of what was said by some of the market research analyst firms, they expect -- and if they are right, and 25% bid growth for next year would mean about flat wafer capacity year-on-year 2016 versus 2015. So that means that if the initial shipments of the technology transition happen in the first half, then this second half will be lower than all capacity additions needed. How much lower, that's a bit too early to say, but clearly first half technology trends and transitions and the second half probably a lot less. On EUV throughput, yes, we've made good progress. We've agreed with our customer the targets for 2016 and we have good confidence that we're going to get there. You have to look at 2016, as I would say, the last phase in the EUV introduction of where our customers are developing their next nodes for which they're planning their production output in 2018 and '19. So it's the last phase of the development node and the qualification of those architectures that we will see in 2018, '19, hitting the market. That means that 2017 will be the year where we will start to see the start of the EUV shipments for production and they will be used in 2018. And that will accelerate throughout 2018. So 2017, we'll see the first start to make sure we can do -- customers can do the output in 2018, which will accelerate in 2018 further on So that's nothing different than we said I think last quarter, so no change from that respect.

The next question comes from Gareth Jenkins. Please state your company name followed by your question.

A couple, if I could or rather a one-on-one follow-up. So I just wondered if you could talk about 10-nanometer ramp. Now, you've got maybe a bit more visibility around it. Should we still be thinking about 10% pure wafer starts between the 20/16/14 nodes and the 7/10 nodes, but also a 40% to 50% lithography intensity increase. Just wondering if you could kind of elaborate on that, now you've got more line of sight? And then I have a follow-up on EUV.

Just for clarification, I just missed part of your question on your 10%. You referred to 10%, what was that exactly?

The low wafer starts, Peter, between your 20/16/14 and 7/10?

Basically, that's the assumption on the 10% that we still work with. We don't have any other data. I'll comment on that a bit later also. The 40% litho intensity, yes, that's what -- node-on-node that's still what we think is a realistic number. On the node-on-node wafer capacity reduction, the 10% is still what we are using. You have to bear in mind that starting from the 28-nanometer node we see those nodes extending a lot longer than what we saw in the past. And you've been around a long time, so you know that that between five and 10 years ago nodes had a two-year life. And then the previous node stopped, next node came and lasted about two years, and almost all logic customers moved in that same time period, which is less the case today or not the case today. We see the initial acceptance of a new node being driven by the leading-edge customers. And they install rather swiftly and rather fast, a quite a significant amount of capacity. And then you see the other customers in our segment, you could say, second tier customers following later. That is also what we see today, 28-nanometer still being shipped even in Q1 of 2016, which is more than four years after the initial introduction. So you can see a kind of a camelback in the first phase of that node, and then a much longer tail, which also makes it more difficult for us to say, how much will that wafer capacity for that node be. Yes, we assume 10%, but overtime it needs to be proven whether this is the correct number.

And just one follow-up. It looks like your Chinese orders are very strong, somewhere around €500 million, which I would assume is mostly Dalian. Can you just give us a sense of your market share into China through the course of this year or rather on those orders?

Well, our market share in China has always been pretty good and it will stay pretty good. I'm not going to give you an exact number, but we have no worries about our market share in China.

The next question comes from Jérôme Ramel. Please state your company name followed by your question. Jérôme Ramel: Jérôme Ramel from Exane BNP Paribas. Peter, just wanted to come back to the point you made that your clients seem to be keen on moving to the 10-nanometer node, because that's a significant improvement compared to the 16/14 versus the 20. Could you just give us a sense what is better? Is it in terms of field, it's in term of cost per transistor? What would make the 10-nanometer node more attractive than the 16/14?

From a lithography point of view, the 16/14 node is very similar from a little half-pitch dimension point of view, very similar to the 20-nanometer node. So you really should not compare 16 and 14 from a little pitch point of view to 10, but really 20 to 10, and that's a big shrink. And as you know, shrink has a big impact on the cost per bit. So it is driven by cost per bit. And for some customers, actually it's also value, just putting more functionality on the same square surface. That's what driving it. Jérôme Ramel: So you'd see -- to make clear, you see the cost per transistor going down at 10-nanometer node?

Yes. Jérôme Ramel: And maybe just one follow-up. You gave kind of guidance for the full year, I mean, competitive guidance for the memory, so flash being flat and DRAM going down. I'm not sure I understood for logic and foundry, what the trend are, quantitatively compared to 2015?

Well, I think, Wolfgang said, the first quarter is relatively benign, but we really see the ramp starting in Q2, and then also we're extending into IDM, not only for foundry, but also IDM on 10-nanometer, which will be the remainder of the year, which will drive the remainder of the year. Now, if you just look at the size of that ramp and the ultimate size for this year is always a bit difficult to predict in the third week of January, so as we were very early in this year. So if you look at that what customers are telling is it going to be a significant ramp, and we'll be driven by the leading-edge players, not so much by the followers, which will follow on, I said as an answer to an earlier question, which will probably have a much longer tail. But the initial ramp what we are seeing will be significant. And that causes us to state that the 2016 logic market for us, we will be significantly higher than 2015. Jérôme Ramel: For logic, that includes logic and foundry or that's just purely logic?

Well, there's logic versus foundry and IDM.

The next question comes from Mr. Muse. Please state your company name followed by your question. C.J. Muse: C.J. Muse with Evercore ISI. I guess first question, I know there's a lot of moving parts here, but curious what are the key milestones we should be watching for EUV to be designed in at the 7-nanometer node?

I think that we mentioned those. I think for our customers, it's most important that they have a certain level of productivity. But it's very important that when you start planning your production that the availability is critical. So the 80% availability target is what we agreed with our customers. This is what we are at least following. And I'd also like to refer to our presentation where you can see that. On productivity, quite interesting, one of our key customers in a question that was asked recently on where the 500 wafers per day would be sufficient for them to go into production assuming a reliable, or let's say, good availability of the tool. And the answer was, yes, they would use it. So you can draw a conclusion from that answer that it's not so much now about the productivity, it is about availability. So this is what we are really focusing on this year. C.J. Muse: And I guess as a follow-up, in terms of foundry logic spend this year, clearly a very good year for 10-nanometer. Curious if you could share your thoughts on sort of the mix this year between 10-nanometer, China foundries spending year-over-year, as well as whether you're seeing any incremental capacity at the 14/16-nanometer nodes? And then I guess to follow-on to that, your expectations for that 10-nanometer ramp to continue into 2017?

Yes. I've made some notes. The 10-nanometer ramp is predominantly outside China, if you refer to China. When we see the Chinese logic or foundry market, we are clearly focusing on the nodes before 10. So the 28-nanometer node is still relatively strong in China, but also very clear indications of their move to 16 and 14. And that is where we see Chinese foundries going. Like I said, the 10-nanometer ramp will be outside China. 2017, is a good question. I think it's too early for us to say anything about 2017. I think what we'll have to go through is the next two, three quarters of the initial ramp, which we have a pretty decent visibility of. And 2017 will be driven by, I think, by the end markets. And we're just a bit too early to comment on that C. J.

C. J., directionally, for 2016, the 10-nanometer will make out the majority of our shipments, but there will still be 28 and 16 and 14 shipments as well.

Yes, but those are then predominantly a deduction of China and somewhat in Taiwan.

The next question comes from Mr. Hosseini. Please state your company name followed by your question. Mr. Hosseini, you can ask your question.

It sounds like he is not there. Operator, you want to go to the next one.

The next question comes from Amit Harchandani from Citi.

This is Amit Harchandani from Citigroup. I've got two questions. Firstly, with regards to a topic that does come up often for discussion, which is equipment reuse. We've heard some of your peers talk about it or customers talk about it. Could you give us a sense of how you see equipment reuse impacting your prospects for this year as compared to last year and more so looking forward towards your medium-term financial roadmap across the different end segments? So that would be my first question and I have a follow-up.

The equipment reuse is always at weight, it appears to dominate somewhat our discussions lately, but it's always been there. Large IDMs have always done this. And it's a very sensible way to manage your capital efficiency. Now, with tool prices going up to where you currently are, advanced DPV prices are €50 million, there is more and more focus of our customers on capital efficiency. And the reuse program, which we have designed by adapting the architecture in such a way that we can upgrade from one node to the other and with little intensity going up with about 40% node-on-node, and there's a clear driver of our customers to say what part of my installed base can we upgrade to the next node, that's called a reuse. That doesn't cover the 40% little intensity at all, it just covers part of it. And we have had situations whereby reuses, let's say, upgrades where planned on the previous node that never happened, because the previous node extended longer than the original planning. So it's nothing new. I think it's going to be part of our business going forward. Very healthy part if you think about an extensive upgrade from a 1950 to a 1970, for instance, so €20 million upgrade with these margins, which is good business for us, helps us in increase our services and option sales business and helps our customers to measure installed base. So I think its going to be part of our business going forward. It might be new to some of our peers, but it isn't to us, because our tools have always been one of the most expensive in the customer fab. So it's here to stay and it's good.

And just to clarify the extent of reuse that you're seeing out there right now is in line with your 20-20 financial roadmap?

In our 20-20 roadmap, we have actually included that reuse. It is an inevitable event, and that the level of reuse that we're seeing is very much in line with our expectations.

And just as a follow-up, if I may, Wolfgang, could you kindly just once again explain or elaborate upon your comment of EUV revenue recognition for this year? I'm afraid I do not catch it correctly? Was it a €110 million in the middle of the year? So what are we looking for in terms of the EUV revenues this year, if you could shed some light on that?

You got that right. I mean, first of all, we told you that while we were in this transition period where we can recognize not when the tools leaves our factory, we will give you guidance as we ship the tools. And I said we shipped three tools last year, of which only one recognized and on top of that we started the shipment of another tool. Of the two tools that shipped and have not recognized last year, and the tool that has started to ship from these three tools together, you should expect approximately €110 million in revenue, and that will be somewhere in the Q2, Q3 midyear timeframe. And the balance of the revenue will likely recognize in 2017. Now that's the starting point, and then we also said that we will ship more tools this year. And those, of course, will -- we also told you there is for instance a 3300 amongst them which we'll recognize faster because there the recognition rules are different because we just need to demonstrate that we can print the wafer. So there is more to come. So you're going to bear with us as we make these shipments. And lastly, I'll say also that we will start to see some service revenue in the EUV field. We already had some -- last year our total revenue was about a €100 million or so, and now only a little bit over €60 million came from systems. So you have to bear with us, Amit. We'll give you information as we go through the year.

So just to confirm, a minimum of €110 million, but the actual number could vary depending on your updates as you go through the quarter now?

That is correct.

The next question comes from Mr. Sanganeria. Please state your company name followed by your question.

So you said memory at similar level to Q4, which has about €350 million, down 35%, so first half at that level and second half you said decline, that will indicate a significant down year-over-year. Did I understand that clearly?

Well, I referred to DRAM, and since we don't split out DRAM in the fourth quarter results, so DRAM shipments will be about equivalent to Q4, NAND will be over the year -- year-on-year will be about flat.

And I want to follow up on that China. You had pretty strong orders from China. Somebody mentioned €500 million, pretty close. And mostly, probably on NAND side, is that shipment to China mostly in Q2?

That shipment starts in Q1.

Starts in Q1.

And will continue.

The next question comes from Mr. Mehdi Hosseini. Please state your company name followed by your question.

Peter, going back to your 2015 performance, your foundry revenues are up 36%, but immersion systems shipment was down and ASPs were down too. Can you help me understand, how did this mix change, despite the fact that the foundries were up so much? And I have a follow-up.

For 2016, you said -- ?

2015.

Foundry shipments are up. The ASP -- we shipped significantly more systems altogether and there were also KrF and other systems in there. So like-for-like the ASP didn't go down, but because of the mix the ASP went down and it's KrF.

Does that reflect the foundry spend more on their trailing-edge versus leading-edge?

Absolutely, like we said earlier, we are currently shipping different node layers. We are shipping 28-nanometer, 14/16, the first 10-nanometer R&D tools. So it's quite a mixed bag of those tools. So to draw a conclusion on ASPs or ASP trends is a bit difficult, because it's quite a mixed bag. But what we can say is that with every node transition the ASP goes up, because of the richer configuration, including a lot more holistic litho options. Now, in 2015, which is true, we had a mixed bag of 28-nanometer node, 14/16-nanometer node, and some early 10-nanometer node shipments. So I think it's a bit difficult, and it did include as we put in the presentation, there is a lot more KrF.

And then my final question is absolutely China. Can you elaborate more, what the key end-markets or device stock that is driving such a strong growth in backlog, as it relates to China?

Well, it's both, foundry and now also memory. So it's shipped to -- so it's the ship to region is indeed stronger, and that's because memory is now also adding on top of the logic shipments, which are predominantly driven by 28-nanometer.

So there is a new NAND fab that is coming online, but also one of your Korean customers has a fab in Icheon. How should we think about the mix, new fab, existing fab on the 3D NAND? And also the foundry market, is it evenly split or is one more than others?

Well, I think, there are existing fabs. And the refurb fab, one of the existing fabs that you referred to is already full. So that means that the refurb fab is going to take tools. That's what it is. It's just as simple as that and not more difficult.

The next question comes from Pierre Ferragu. Please state your company name followed by your question.

It's Pierre Ferragu, Bernstein. So I have a question on your gross margin for the backend of the year. If I understand you correctly, foundry is going to be very strong. That's a 10-nanometer node, it's very high-end tools, a lot of options and upgrades and memory is going to be low. So should we expect like a very, very healthy gross margin development beyond Q2? So like Q3 and Q4 gross margin should be also heading in the right direction? So that's one question. And then I have just a quick follow-up on EUV. I got at some point confused about insertion. My understanding is that both your IDM customers and your foundry customers are going to insert EUV at the 5-nanometer node, so not the next one, but the one after. Is that also how you see the world from where you are? And then maybe on this 5-nanometer node insertion, how much visibility do you have today, on how heavy an insertion it is going to be? How are we going to use EUV tools only after a very, very low level for the first layers or are we heading into a more massive adoption of EUV at this 5-nanometer node?

Let me answer that the EUV question first. Whether it's called 5-nanometer or 7-nanometer, I don't want to go into that nomenclature, because there is a lot of confusion about what is what. And I'm also not going to tell you what we believe the lithography pitch is, because that's what's probably going to make it easier to understand that we're talking about this same thing. What is most important, and that's what we should focus on, whether you call it 5 or whether you call it 7, our customers, our leading-edge customers make it very clear to us that they will start at the output of their chip architectures that need EUV, whether it's 5 or whether that's 7, I don't know how they call it and I don't care. But when they need EUV for output is 2018 starting, 2018, 2019, which actually means that our shipments for production purposes need to start in 2017. It takes about a year to really qualify for a production ramp. So this is what we are focusing on. This is what we are discussing with our customers. And this is also driving the decision points and entry points for our customer for production insertion. It's just 2018, 2019 timeframe.

And in terms of the volume of insertion, so is that going -- do you have already visibility on how high in the architecture of the chip EUV is going to be used? Is it just going to be the most critical layer or more than that? Do you have that visibility already or is it still something that has to be defined?

Yes, we have that visibility, reasonable visibility from our customers. But it actually leads us to believe what we said in the past that the first year of volume ramp, and as you know, with the lengthening of the nodes, we discussed it a couple of quarter ago, everything is about six to 12 months later than we thought about a year ago. But we believe that the initial year of the production shipment will be a dozen or so tools, and then it will double every year that we move on. So we are still -- all our simulation models still show the same thing. So the first year of the production shipments is about a dozen tools, and then it will double the year after that, and double the year after that. That would [multiple speakers] the right model.

And that's relates to the gross margin question, Pierre. Yes, you are right from a direction in the second half, as logic will be a bigger part of it, directionally margin goes up. Of course, we don't know the exact volume yet, so we can't describe that volume effect. But directionally, you're right. One caveat, if there is a concentration of EUV revenue recognition in a quarter, you will see some distortion. But when we get to that bridge, we will explain to you how that works and show you the margin without EUV as well.

Next question comes from Andrew Gardiner. Please state your company name followed by your question.

It's Barclays. Just another one on EUV. In terms of the sort of looking at the milestones you are targeting for this year, I'm just trying to understand roughly when we might be able to sort of get a better idea, as to how you're making progress. You seem to be in a bit of a gap at the moment. Clearly, the customers have the 3300 tools installed and running, and those are a lot of the metrics you are talking about. The 3350s are there or on their way. So when can we expect to see some of the first news on the 3350 tools at the customer sites? Is SPIE too early or is it going to be a bit later than that?

That's a good question. And you pointed out the 3350 is really the tool that has the improvements on it that will give us the 1,500 wafers per day and the 80% availability capability. Now, those tools are just starting to ship. They need to be installed. The installation only takes a quarter or three months, which actually makes it too late for any, let's say, significant information on the SPIE conference, which is in February. So it will be around midyear. That's what you need to focus on.

And then just sort of quick follow-up. I mean, you've highlighted the six to seven tool shipments, EUV tools shipments this year. Can you give us any sense as to when those are coming or you haven't really -- I know there's no revenue recognition for those tools, but just in terms of the rough timing of shipment for us to gauge when those are leaving the facility?

Well, they are starting to leave the facility this quarter, the first one, so it will be throughout the year. There will be tools shipping every quarter. And as to revenue recognition, Wolfgang already said it, and gave some clear guidance for the revenue recognition of the tools that we shipped last year and the one that was in the process of shipping towards the end of the year. Now, of the six to seven tools that will ship in this year, there will be some revenue recognition, because there's a 3300 in there, which will very likely two of them could be, which will book revenue. And also on the 3350s, it depends on the commissions or the order conditions, where we can take some revenue already in 2016. So that's why Wolfgang said, it's the minimum, and there is very likely going to be EUV revenue on top of that in the course of the year.

Next question comes from Francois Meunier.

I just wanted to have some technical details about this 3350B that you are currently running in your factory. And first, congratulations for achieving the 1,250 wafers per day in Q4, that's a great achievement. What is or what was the availability of this machine actually during Q4? And maybe if you could give us some details about the laser, which was used, if it was a 120 watts or a different laser source for EUV? That's the first question. The second question is, I know, it's a bit cheeky, but it's good to announce the €1 billion share buyback, but why not more, given the progress made by EUV adornment and your confidence in the 2020 target of €10 billion revenues?

Well, I'll start with the second part, and then Peter will have the first part. This is just another layer. We have a very clearly stated financial policy that says we evaluate what our minimum cash needs are. Then we service the dividend first that we want to have at least stable, preferably growing. Last six years we have always been growing it. And then all the balance will go towards share buybacks. Now, we just roll out layers of these share buybacks. And as soon as we have used up the money, we will introduce the next layer of solution, be concerned by that all. We're executing the financial policy and as we generate the free cash flow, once we're through the remaining 1.5, you should expect us to announce the next tranche.

On your question on the technical details of the 3350, well, we started shipping 3350 in our factory, and we measure availability at a customer sites. While, we haven't installed it yet and are running it at the customer sites, so it would be a bit difficult to give you an answer there. It's too early. Also to the question of -- there's an earlier question, when are we going to see data on availability and wafers spread per day? Probably more towards the middle of the year. So the technical details, but there are some features in the 3350, like the in-situ cleaning and some other features, that give us confidence that we should have a higher availability, and also a higher throughput, not so much because we have a strong relation, the relation is the same, but more because of the transparency of the illuminator and the optics that gives us a better transmission, and that's why we get a better throughput.

The next question comes from Mr. Ahmad. Please state your company followed by your question.

This is Farhan Ahmad from Credit Suisse. Peter, my first question is regarding 3D NAND. I just wanted to make sure I understand correctly. Relative to your last call, it seems like you're upticking on the overall NAND CapEx. And I wanted to understand whether the linearity of the CapEx is also more weighted like the DRAM CapEx in the first half or do you expect like the NAND CapEx to be more evenly split throughout the year?

Well, 3D NAND guidance is up as compared to last quarter where we thought we were going to be down in this year. But that's because we have this new refurb fab of 3D NAND in China that actually came up, and it's relatively flat throughout the year. So it's not a big bias to one or the other half.

And then relative to your NXE:3350 systems, they are supposed to get you to about 80% availability and 1,500 wafers per day. Where do you see the performance in the lab today? And what should we expect when the tools are at the customer's site? Should we start getting like 1,500 wafers per day from the get-go or will that take some time to get demonstrated at the customer site?

Well, what customers will actually do, because they're very expensive machines, they will always ramp these tools slowly. They go through a process of qualification, and then they will start increasing the power, they will start increasing -- with increasing the power, they will get more wafers. So it's going to be through cycles of learning. So it is not to be expected that they will go gung-ho from day one. They will probably say, let's start where the 3300 left off and then they are going to gradually increase.

Next question comes from Patrick Ho. Please state your company name followed by your question.

Stifel Nicolaus. Peter, just in terms of the EUV insertion at the 7-nanometer logic node, do you believe the change in cadence with their tick-tock process, do you believe that buys you a little more time, in terms of, I guess, guaranteeing the insertion for high-volume production at 7-nanometers?

Well, I don't think that's related, I mean customers have their plans, and whether they follow cadence A or cadence B. What they tell us is that they need their production output in a certain period, in a certain year, which is starting 2018. And then you just calculate back using the cycle time that we need. And then we come to a moment where we need to start shipping the tool. And what ultimately drives the decision of the customer to say, I need that output in 2018 or 2019, there are many, many reasons, it is not only cadence, it's also time that they need to actually develop that next node. And for us, it's not that relevant. Most relevant is when do they need the tool, when do they tell us that they need the tool, and that's for the 2018, 2019 output, 2017 shipment starting for us.

And my follow-up question, in terms of the holistic lithography growth that you've seen over the last few years, how do you project the growth first in 2016 and maybe over the next couple of years, given that it's gotten strong adoption, do you see the growth rates tapering out somewhat or where do you expect to continue to see that growth in that segment of your business?

So in 2014, our holistic lithography business was just over €500 million. Last year, it was over €600 million, so it's growing nicely. We've previously talked about by 2017, 2018, we want to be at €1 billion for this business. And as a reminder, to a very large degree a software type of business with very healthy gross margins; only the YieldStar being the hardware there. And then, you can anticipate that it will continue to grow going into our 2020 plan as part of helping us on the equation of the gross margin to a 50% level. And we've also started to see and reported back to you, that customers are evaluating some of the software features for EUV as well. So it's not going to go away when EUV is coming in.

Yes. I think just to add to this, I think on the total service and options sales, which includes holistic litho that Wolfgang referred to, in our model we see that growing from the €2 billion, where we are today to €3 billion to €3.5 billion by that time, by 2020.

Ladies and gentlemen, we have time for one last call. So if you're unable to get through on this call and still have questions, please feel free to contact the Investor Relation department with your question, and we'll get back to you as soon as we can. Now, Aaron, can we have the last caller, please.

The final question comes from Timothy Arcuri. Please state your company name followed by your question.

Cowen and Company. I had two questions. I guess first of all, Peter, just on the overall '16 outlook, I know you said that DRAM is going to be down a lot, NAND is flat, and logic is up a lot due to 10-nanometer. So where do you and Wolfgang think that that leaves us for the year? Obviously, it's going to be up, but how much? Is like up 10 a comfortable number for the year, at least?

Well, we're all well-trained accountants. So if you take 2015, and I will give you a percentage up, we would guide you for 2016, which we said we weren't going to do. So unfortunately, we have to stick to what we called is qualitative guidance. And I think throughout the year, we get the -- there's a better feel for how the backend of the year is developing, and really talk about really the Q4 backend of 2016.

And then I guess just a follow-up on China. I know that there was some questions asked about this, but the big order number of roughly €500 million, it sounds like this is really 3D NAND. But if I divide the numbers, that suggests it's like 120K of wafer starts worth of capacity. And we know that the Korea fab in China is full, and the only fab this big is the one that's still searching for a technology partner. I think the fab you talked about in the prepared remarks was like half this size. So are we seeing orders for this other fab that's still looking for a technology partner or is my math not right?

Well, let me make one thing clear, China is not only 3D NAND, that's a new feature of the Chinese market on this particular refurb fab. But it's only a part of the story. There is also Chinese foundries in there. So there is also onesies and twosies shipment to a fab that are almost full. So it's a mix bag. It's definitely not only 3D NAND. End of Q&A

Well, thank you everybody. On behalf of ASML's Board and management, I'd like to thank you for joining us on the call today. So operator, if you could formally close the call, we'd appreciate it. Thank you.

Ladies and gentlemen, this concludes the ASML 2015 fourth quarter and annual results conference call. Thank you for participating. You may now disconnect your line.