艾司摩爾 (ASML) 2010 Q1 法說會逐字稿

Ladies and gentleman, thank you for standing by and welcome to the ASML 2010 first-quarter results conference call on April 14, 2010.

Throughout today's introduction all participants will be in listen-only mode and after ASML's introduction there will be an opportunity to ask questions.

(Operator Instructions)

I would now like to turn the call over to Mr.

Craig DeYoung.

Go ahead, please, sir.

Thank you, operator, and good afternoon and good morning, ladies and gentlemen.

This is Craig DeYoung, Vice President of Investor Relations and Corporate Communications here at ASML.

I would like to welcome you to our investor call and webcast.

As the operator mentioned, the subject of today's call is ASML's 2010 first-quarter results.

Co-hosting today's call from our headquarters here in Veldhoven, The Netherlands, are Mr.

Eric Meurice, our CEO, and Mr.

Peter Wennink, ASML's CFO.

At this time I would like to draw your attention to the Safe Harbor statement contained in today's press release and in our first quarter results presentation, both of which you can find on our website at www.ASML.com.

This Safe Harbor statement will apply to this call and all associated presentation materials.

As a reminder the length of this call is 60 minutes.

And now I would like to turn it over to Mr.

Eric Meurice for a brief introduction.

Thank you, Craig.

Good afternoon, good morning.

Thank you all for attending our quarter-one 2010 results conference call.

First, Peter and I would like to provide an overview and some commentary on the quarter, and as usual Peter will start with the review of the financial performance with added comments on short-term outlook.

We will complete the introduction with some further details on our current position as we close the quarter and head into 2010.

Following the introduction we will open up for questions.

So please, Peter?

Thank you, Eric, and welcome to everyone.

As Eric mentioned, I would like to take a moment to share some observations on the events of last quarter as well as some details on our first-quarter results.

Our first-quarter sales came in at EUR742 million versus EUR581 million in the fourth quarter of last year, which is well within the range of our guidance.

We shipped a total of 34 systems, 23 of which were new.

Our new system ASP of EUR25.8 million is a significant increase from the EUR19.7 million in Q4 and it indicates a shift back to a richer mix of emergent tools.

The mix of the prior quarter was impacted by the demand from foundries for capacity-related tools.

Our service and field option sales came in at EUR110 million, which can be considered normal given our current business conditions.

The Company's first-quarter gross margin was 40.3% versus 38% in Q4 and this reflects an improved coverage of our fixed manufacturing cost and cost reductions.

We continued our focus on R&D with a spend of EUR120 million in the first quarter.

SG&A came in at EUR41 million, slightly higher than Q4, but that is in support of the growing level of sales through the course of 2010.

On higher sales and improved gross margin in the first quarter operating income improved to 18.5% from 11.8% in Q4.

In the first quarter, despite the working capital ramp, we generated EUR50 million in net cash.

With that we again maintained a cash balance comfortably of EUR1 billion.

The first-quarter net bookings met our expectations at 50 systems valued just over EUR1 billion.

Net immersion bookings of 24 systems demonstrated a continued focus on technology investments driven again by DRAM and the foundry market segments.

But in addition we saw KrF bookings of 17 systems indicating a continued need for more non-critical systems in support of the leading-edge production ramp and is mainly in the non-memory applications.

First quarter Bookings pushed our backlog to over EUR2 billion containing 85 systems, which is by the way highest backlog value in our history.

We are highlighting also the fact that there are 58 immersion systems in backlog, 28 of which are our newest immersion product TWINSCAN NXT;1950.

As for our outlook we continue to increase factory production capacity in order to meet the growing demand as indicated by our Q4 and Q1 bookings and our Q2 bookings guidance.

The latter we expect again to be at the level that is approximately the same as we saw in Q1.

With this level of bookings and the ramp up of our production capability we have clearly created the potential to record a top-line revenue number for 2010 that will exceed the previous big revenue of the year 2007.

For the second quarter of 2010 we expect net sales to be about EUR1 billion.

Gross margin for the second quarter is expected to be around 42%, R&D expected to be around EUR125 million, and SG&A name around EUR42 million.

With that brief summary I would like to turn back over to Eric for more on our view of the coming quarters.

Eric?

Thank you, Peter.

As I said, Q1 sales and bookings came in as planned with high bookings expected to continue also into Q2 giving us added confidence in a prolonged recovery of the semiconductor industry.

Such an upturn can last two to three years if one replicates the profile of historical cycles.

Of course, our challenge is to remain cautious as to any signs which may signal a correction to this trend, but the two main drivers IC unit demand and technology conversion CapEx requirements currently remain in positive territory.

In fact, I would say IC unit demand is in positive territory and technology conversion CapEx requirement is in very, very positive territory.

Indeed the large first-half revenues continue to be driven by technology buys from the early adopters in the DRAM market segment currently converting capacity from their 65 nm or 55 nm products to 40 nm node and in the foundry segment building up at this moment their 40 nm capacity.

The second half of the year we will see this trend continue, in particular as these second-tier vendors are ramping in those same segments with also initial demand for 30 nm NAND flash capacities starting.

This NAND flash has been fairly weak so far.

Our 2010 revenues could, therefore, exceed our previous EUR3.8 billion peak annual revenues achieved in 2007.

The demand for IC units is still forecasted to be robust as up to 15% above 2009 and importantly the technology transitions that we discussed are requiring a significant increase of immersion lithography layers.

That is the number of lithography passes per wafer processed using these new immersion lithography systems.

Immersion layers will respectively double for the DRAM and Logic product and increase by about 40% or more for the NAND product.

Considering that there was a reduction of almost 20% of the memory capacity in the form of 200 mm fab which were retired in 2008 and 2009, we are comforted that the significant and prolonged lithography tool delivery level is absolutely necessary without creating overcapacity in the industry.

To enable these technically aggressive node transitions, we have now shipped nine NXT;1950, which is our most advanced TWINSCAN dual-stage scanner, having proven critical dimension emerging uniformity well below 1 nm and overlay of less than 2 nm, which is industry-leading performance of course.

At the end of Q1 we had 28 NXT systems in backlog, 22 in memory, and six for an IDM's Logic.

We expect to ship 11 of those 28 in the second quarter and we will still grow our NXT backlog in the second quarter.

On the technology front as immersion processes are extended by the use of complex double-patterning techniques we are deploying our holistic lithography project suite which enables the widening of critical process windows on one hand and allows for tighter process control in the fab on the other.

This capability further significantly differentiates our TWINSCAN immersion product from competition.

For the next-generation technology, we continues our EUV development and our On track to ship our first TWINSCAN NXE;3100s this year and Have now received our six order for this tool, confirming the interest of now older semiconductor segments from DRAM, NAND or Logic arena.

we Will ship six pilot production tools in next 18 months and our Currently planning was all the industry leaders, the 2012 production ramp for the new version of this machine called NXE;3300.

Orders will be taken already for these production EUV tools by the middle of this year.

In summary, we expect this cycle to be sustained by the normal technology transition of the early adopters and the subsequent technology conversions by the second tier DRAM makers, by the next flash memory upgrade cycle anticipated for the second half of 2010, as well as the foundry structural capacity built on the advanced nodes.

In lieu of the technical requirements for immersion in layer growth, which I have mentioned go as far as doubling for DRAM, we appear to be in no real risk of creating industry overcapacity between now and the end of 2011 although we will remain, of course, cautious as to the overall stage of the economy and its potential impact on overall semiconductor unit cost, which is a second driver of our growth behind the technology node transitions.

So with this explanation, Peter and I will now be pleased to take your questions.

Thanks, Eric.

Ladies and gentlemen, the operator will instruct you momentarily on the protocol for the Q&A session but beforehand I would like to ask that you kindly limit yourself to one question with one short follow-up if necessary.

This will allow us to get in as many callers as possible.

Operator, could we have the instructions and then the first question please?

(Operator Instructions) Didier Scemama.

Yes, gentlemen, it's RBS.

Thanks for taking my question.

I just have a simple question.

First of all, you mentioned that the 55% or 60% increase in CapEx this year for the industry will increase capacity by about 15% for IC units.

What do you think would happen if IC unit shipments were to grow about 20% or 25% this year?

At which point do you think we would need to or we would start to see the development of new factory from your customers among foundries and memories?

Then I just have a quick follow-up.

This is a good question because indeed at this moment, although nobody can read the GDP developments in the year, you do have this question very often in the industry.

Where there was only 15% unit we are too conservative and the numbers should in fact really be between 15% and 20%, so this is a very actual question.

So, of course, we are preparing capacity for this ourselves and I think I said last quarter our capacity will be at the level of about EUR1.5 billion of lithography tool per quarter within -- all by the end of the year.

So we will be ready in case this upside would come or is confirmed.

Obviously, the customers themselves are ready.

Most of the customers have enough space in their current buildings.

As you know, we do not have the time anymore where people would build walls and then fill them completely.

Most of our customers have already the walls and have filled them by phases, and you have in the industry a lot of fabs which obviously, definitely not full and are expecting the next phase.

But in addition you are starting to see, of course, fabs in Singapore, in Korea, in Taiwan which are getting to completion or are already completed and our shells.

So this is again a good preparation of the possibility of an upside.

Great.

And a quick follow-up, I just wanted to clarify one point.

On your second half outlook do you expect your book-to-bill to remain above one for the rest of the year?

Again, a very good question.

If I translate it differently I would say are we going to have a peak booking?

The current profile of the cycle when you start and, if the cycle is looking like a historical cycle and with the spread over two to three years, we do expect in fact that the book-to-bill will remain to plus or minus 10% from one, if you know what I mean.

With some quarter a bit higher, some quarter a bit lower around I would say the current level.

That plateau type activity could last, again as I said, two to three years and allow us to reach, as I said also last quarter, our potential target of sales of EUR5 billion as we said that we would reach on the next up cycle.

So that is within the possibilities of this.

Great.

Thanks so much.

Janardan Menon.

Hi, it's Liberum Capital; thanks for taking the question.

You said that you could see a flash memory upgrade cycle in the second half where flash goes to 32 nm.

Can you sort of quantify on a relative basis how big that can be versus the DRAM upgrade cycle that you are currently seeing to 40 nm?

Are you going -- without a big FSB-lead boom would you just get a significant flash technology upgrade cycle?

Can it come close to the levels of orders that the DRAM industry is placing?

And a short follow-up, if I may.

Your capacity on the 1950 seems to be moving, or at least your production is moving, from nine to 11.

What is your maximum capacity in that going to be in the second half of the year?

And when will you be in a position or are you are already in a position to satisfy customer requirements for any kind of rush orders on those tools at this point in time?

Okay, thank you.

Very good; two good questions.

So you could probably use a simple formula between DRAM and flash.

We think the flash is equal to one half of DRAM roughly in numbers of immersion tools in units.

So of course you cannot have this every quarter so some quarters will be higher, etc., but we would expect the flash 32 nm wave to start, as we said in Q3, to looks like half of the big wave of 40 nm DRAM.

So that is happening, as we say, at this moment.

Regarding the capacity of immersion tool, as I said, I expressed myself in euro, EUR1.5 billion of capacity.

That would be about 30-ish to 35-ish immersion tools per quarter plus KrF, etc., etc., of course.

But your question was on immersion.

Thank you very much.

Odon de Laporte.

Yes, good afternoon.

I have a question relating to gross margin.

When I am doing my math I conclude that with sales of EUR1 billion Q2 gross margin could have been a little bit higher, maybe 50 basis points higher.

So I was wondering whether it this will be a clean, clean gross margin or maybe you will be penalized by any items, maybe in relations with the ramp up of NXT.

Thank you.

If I could answer --

Before, I think nice accent.

Thank you.

I will answer quickly.

Clearly we have a gross margin percentage of around 42%, like we did around 40% for Q1.

In Q1 we were 30 basis points higher.

That is within the realm of, let's say, the differences that can always happen during the quarter; 30 or 50 basis points difference it could happen.

I am not saying that it will but if you come 50 basis points up that is pretty reasonable.

But as you have seen also in the actuals, there is always somewhat of a difference between our guidance and what we actually come out with.

But it is within the margin of you could call error.

Okay, thank you.

Timothy Acuri.

Hi, a couple of things.

First of all, Eric, you talked about revenue -- the revenue capacity being roughly EUR1.5 billion by year-end.

Does that suggest that there is a reset in lead times?

So, i.e., lead time is now nine to 12 months and maybe that comes down pretty significantly later on this year.

And if so, where does it come down to?

Again, very good question because we hear this from the customers on a regular basis.

The lead time by then would be about three months.

So we know the terms -- I don't say this to our customers too much, but we can turn orders within three months when we have reached the maximum capacity and have prepared the lenses.

So in other terms, at the moment we reach back a sort of plateau of work we can offer customers very short lead time, turnaround time, which then makes the business much more stable for them where they put orders on the long-term when they can do this and they can plan it.

And they have x percent of their business which can be done a very short lead time.

So just a quick follow-up on that.

Typically, whenever you see sort of a reset in lead times from such a long lead time to such a relatively short lead time doesn't that usually drive a correction in bookings because there is at least some portion of what is being booked today that is happening because of the fact that your lead times are so long and people want to get in the queue?

So I am sort of curious as to why that kind of a reset in lead times would not drive a correction.

It may be mild but a correction nonetheless in bookings.

Well, no, I think you always have the same -- you have always the same behavior.

Things that can be booked like a new fab as we discussed or a new line is going to be booked early because people understand that the real lead time of those machines are within nine months, six to nine months because you have to build up the lens, the glass, etc., etc.

So when they can do it, they will do it.

So a very large part of our bookings will always be on that basis.

Then you have a part of the bookings which are due to the fear factor.

As you said, we need a slot and you have a big part of our bookings which is based on complete surprise out of the blue.

So depending on the year the fear factor is higher than the blues.

There is the stuff coming from the blue end; sometimes things coming out of the blue is higher than the fear factor.

But this impact could be responsible for, say, 20% to 25% of the fluctuation quarter to quarter of the bookings, but they are not, I would say, more significant than that.

Okay, so just to make sure that I get that.

So you are saying that roughly a quarter of your bookings right now are because of the fact that lead times are so long?

You are putting words in my mouth.

At this moment I would say because we have now accumulated two or three even -- we are going to accumulate three quarters of high bookings, the question is not even real.

The bookings are based on the fact that the customers now have realized that they need more capacity.

It's driven not before for double booking; it's driven for definite output.

In fact, all of our machines are expected, waited for.

They are all running at a bit above 100% capacity.

So you can see in there there is no paper tigers.

This is at this moment highly, highly required.

In fact, the first question was is net going to be plus 15% or more.

At this moment everybody is thinking that it could be more than 15% therefore every order is for use.

So I would not say at this moment that there is any luxury for what you just mentioned the possibility of saying I have to double book because I am not sure.

That is not the case at this moment.

Also, if you look at the presentation, 76% of the backlog is shippable over the next six months.

So it's pretty tight and it actually is more proof that customers are not adhering to our standard lead times, which currently is not a very short lead time but it is the long lead time.

More for the leading-edge products closer to nine months than to six months.

So you have this management problem that we have that we have a total lead time in the supply chain which is, in fact, closer to nine months and the older lead time that customers are now granting us.

And that is a game.

That is a gain between a customer and a supplier that is sometimes played very smartly by our customers, sometimes very smartly by ourselves.

Again, you may not have really taken with the full seriousness the two messages we gave.

The first message is that most of the industry, semiconductors and beyond, really are under product pressure.

So there is clearly a volume pressure at this moment and you are driving a lot of demand for that.

And then the second reason, the second thing that I said which is very, very important in fact, is the numbers of layers on each of the chips are doubled in immersion for DRAM and will be a 40% to 50% increase on flash.

And in 2011 these ratios again are increased, potentially not doubled but not far from it.

So at this moment everybody -- there is no rush to be overly smart about double booking.

We are putting all the capacity necessary to in fact execute these ranked needs.

The double-booking opportunity in fact is impossible that at this moment because again we are running everything at full capacity and our customers are doing the same.

Understood.

Thanks so much.

Sandeep Deshpande.

Hi, JPMorgan.

Can I ask you regarding the orders in the current quarter itself -- is it mainly still in the second quarter the orders are still going to be very much DRAM and upgrades to immersion, and then in the latter half of the year NAND?

And, secondly, regarding the NAND orders for that half of the year, are we talking about that you are already in discussion about new capacity in NAND or are we just talking about a 32 nm shrink?

Peter, can you --?

Yes, the orders in -- just to give you a breakdown of the bookings in Q1, just to refresh your memory.

Of the bookings that we had in Q1 was about, I would say 20% was foundry/IDM, 55% was DRAM, about 10% was flash, and about 15% was what we call so-called hybrids.

So it could be used for either/or, which we think is largely going to be DRAM.

So that is the bookings split in the first quarter.

We don't expect in the second quarter that there will be a big difference in that split.

We will probably see the first few bookings coming in for the flash RAM, like you talked about the 30 nm ramp in the second half, but we don't expect a big change.

So that is the first part of your question.

The second part of your question was?

I was looking at the data when you were asking the question.

The second part was, Sandeep?

The second part I am asking is in terms of the flash bookings are they only shrink-related that you expect in the second half or are you also in discussion about new fab and new capacity launches in flash?

Definitely there are some shells, as you know, in the flash industry and some of our customers will take the first tool shipments to start filling those first lines.

So it's going to be a combination of the 30 nm shrink and the extension of capacity.

So you expect a secondhand flash capacity ramp which would mean that you sell both immersion and (multiple speakers) making an adjustment from the ASPs, etc.?

Yes, and it will not be in the front part of the second.

It will be like the fourth quarter.

Okay, thank you.

Atif Malik.

Morgan Stanley.

Thanks for taking my question.

Eric, in your prepared remarks you mentioned that we might not see oversupply for a while.

My question is on foundries and as foundries are spending, at least the tier 1 foundries have been spending quite aggressively since last year, and the capital intensity is above 40% level and the last time we saw this high tech capital intensity was 2004 which was followed by a correction.

So as this yield that foundries at 40 nm are low like 40%, 50% and you get 30% more incremental capacity when 40 nm yield improved, I am just curios how sustainable are the foundry bookings level from here onward.

First of all, the foundry business that we have is only tier one and I will not mention what name that means but only tier one.

Which means as the mechanics of this business [it said] the customers would always force a second source to the tier one, you are certain that you are going to have some duplicated capacity being put in the other foundry businesses.

These ones haven't started yet, so you know that this foundry -- your (inaudible) is going to happen anyway in parallel to the tier one.

On the top of that I would like to emphasize that you haven't compared us to 2004.

2004 was in 90 nm Logic RAM which had 90 nm.

We had many IDMs doing the same ramp.

It is quite different than we have in 2010 where 45 nm Logic RAMs is basically foundry.

So we had many IDMs at that same time that were doing the same thing and that is now all concentrated.

It's what you call consolidation so I think it's difficult and also a bit dangerous to make that particular comparison.

Okay.

And just to follow up on gross margins here, if you just assume that revenues stay at EUR1 billion level in the next three quarters and your margins are already 42%, from software holistic contributions how high can the margins go if you assume the revenues stay flat at EUR1 billion?

Or do we need to see a higher top line to get to the 45% target that you guys have talked about?

I would say both.

If you want to say the 45% would need -- we have always had at EUR5 billion.

We are not at EUR5 billion today, but at EUR5 billion in the future which we hope is not going to be the very far future.

That will be driven by the volume impact on the gross margin but also by that time the impact of the new, what we call the holistic litho products.

So we are seeing shipments of the first of those products this year.

They are going to add to the margin profile but to get to the 45% we have said that before that is between EUR1.2 billion and EUR1.3 billion per quarter.

That is a volume and a product mix issue.

We are well underway, especially on the holistic litho product, Eric alluded to that in his speech, that we are being increasingly successful in selling those products.

So that will add to the margin profile.

Thanks.

Simon Schaefer.

Thanks very much, it's Goldman Sachs.

I had a question on -- just going back to this debate on lead time.

When I look at your backlog today of the 58 in immersion systems, am I right in then saying that you can deliver all of those in 2010 and that is basically incorporated in your forecast?

Just to understand the mix dynamics and from a pricing point of view going forward.

Yes, I think what you have on backlog is 90% shippable in nine months.

There is always a few strategic machines which have been designed by the customer put a bit longer than normal as I tried to explain at the beginning.

Lead time is one thing but when you know for sure that you need something at a given time because your walls are finished [you will say, have that so].

You can say this about 10% I think is shippable later.

But on the other hand there are tons of -- sorry, there are certain numbers of orders which, quote-unquote, are [virtual] at this moment.

That is the customers have told us they absolutely need the units but the paperwork is not finished.

So that, in fact, has always represented a very large part of our business which is why we do this forecast of bookings as well as a forecast of billings with usually some accuracy.

Because we have so many, I will call it virtual orders, which are being discussed in terms of your spec, the timing, in fact sometimes to price, etc.

So where we actually know that the likelihood of that happening is virtually 100% but we don't have the paperwork.

I think I would like to correct -- there is a feeling out there very often that there is a one-to-one correlation between our lead time, our production lead time, and the lead time that our customers give us in terms of a sales order.

That is not the case.

We have an internal lead time; it takes us nine months to build a tool.

And from day one, from that first day of the nine-month period we can have a discussion with a customer for a particular shipment and it can take two, three months before they finally give us the deal which is signed, sealed, and delivered.

Some customers are very good.

They give us POs 12 months before, that is why we have POs for a shipment in the first quarter of 2011.

But there are also customers which are a bit tougher.

They always have a bit -- we need to put a bit more effort in in getting the POs on time.

So don't assume that one-to-one correlation.

Right.

I see.

I guess I am just trying to go back to this earlier phrase that you used, I think, about the fear factor order I think is what you called it.

So how much of that do you think genuinely is out there now or is any at all?

No, I don't see it.

I answered theoretically your question in case -- if you are in a situation of constraint, people will have fear factor order.

But in this business at this moment the need for machines and the placement of each of the machine -- I don't know how to express this better but the placement of each of the machines in each of the places of each of the fabs are negotiated one by one.

Remember these are EUR40 million each.

You are not having fear factor order for the pleasure, especially that you have to take deliveries of them.

As you know, our terms and conditions are pretty solid on that.

So I apologize for having misled people.

At this moment there is no issues of us.

The order backlog we have and the virtual backlog that we have are absolutely required machines which will be immediately put in production to the maximum speed.

Got it, okay.

And my second question would be just if I look at your orders, your order book over the last nine months or so and you said NAND has been very dormant but maybe close to EUR2 billion worth of orders being placed from the DRAM manufacturers as their shrink.

I appreciate that NAND of course has a slightly different configuration of machine requirements in relation to wafer starts they are trying to get to.

But I guess my question is going forward is the NAND upgrade cycle to the three X worth the magnitude of this EUR2 billion figure to what you have just seen for DRAM over the last nine months or so?

I am just checking one quick thing, the numbers of layers.

Give me a second.

No, DRAM is always going to be a much bigger business than flash for two reasons.

One, the numbers of layers of immersion required by the next technologies from 40 to 30 for instance or 30 later into smaller or flash; there is always a factor of about 50% more.

50% more immersion layers in DRAM.

So that is one thing.

The second thing, if I am not mistaken, the number of units of flash is that about half of the numbers of DRAM.

So DRAM is a bigger business and it has more technology requirement.

To give you some color on the backlog, as you can see from our presentation about 25% is foundry and IDM.

I am going to give you the breakdown in the memory sector, 45% is DRAM, 15% is flash, and 15% is what we call hybrid but that is largely, we think, is going to go to the DRAM side.

So it's a pretty low number for a flash that we have in the backlog.

Like Eric said, if you -- about roughly 50% of the DRAM demand will finally end up being flash demand.

There is still some way to go.

Yes.

Okay, thank you very much.

Gunnar Plagge.

Nomura.

If I look at the unit immersion backlog it's now this quarter almost 50/50 between XTs and NXTs.

Now I was asking myself and looking 12 months out who still would buy NXTs in 12 months because I mean if I look at --

XTs, right?

XTs, yes.

If I look at NAND we have the double-patterning development, in DRAM overlay requirements, and if I understand correctly in Logic the overlayer requirements are even worse than DRAM.

So if I -- who is buying XTs in 12 months time?

And if I look on at the implied ASP for NXT is about EUR40 million so is it totally unreasonable to model EUR40 million immersion ASPs in 12 months time?

This is a very good question.

In fact we cannot answer for real that question, not that we are hiding but we are not 100% sure.

It is true that if the customers were to want to have the appropriate target overlay, for instance, they would convert 100% to NXT.

However, we also know that there are some layers within the numbers of immersion layers which are less critical than others.

So it may be that some customers would say I want to optimize CapEx for one or two years and they would have a mix.

But it may be that the customer would say optimization for one or two years makes no sense and I prefer to buy the machine which I will more extendedly need.

So to be clear our history has been that the customers have always gone for extendability, which is why we have been in some ways so successful.

But this time we clearly offer the customer a choice between an NXT or an XT.

Both machines will have about the same margin for us so it's not a big issue of what their choice is.

But the customers may have then the choice between a cheaper version if they do not need to use full performance.

And having said that I think to answer your question, we do believe that our shipments will be predominantly NXT within 12 months.

But like Eric said, the percentage of XT -- there will be XT sales and how much that will be in the end has been difficult.

So your assumption of a EUR40 million flat ASP for immersion that is a bit too optimistic.

Okay, thanks.

Maybe as a follow-up, coming back to your famous model I think you said before that in order to entertain a 50% bit growth by the year ending 2011 you would need about a quarter of the million additional wafers.

I think that was based on the assumption that 2010 is flat capacity and I think what we see now in DRAM that indeed we see some capacity increases.

So I am just trying to understand does it mean what you said before, you need a quarter of the million between end 2009 to 2011 to entertain basically two years 50% bit growth?

I do not remember fully all those numbers.

I am not trying to --?

I think you said 250,000 is 223 fabs, so that might make it a little bit easier for you.

Wafer starts.

Wafer starts per month?

Yes, with 223 fabs.

With a fab of 100,000

250,000 wafer starts in --

Wafer starts per month?

Okay, sorry.

In DRAM?

In DRAM, yes.

Now all of what I said was probably wrong, not because I was lying.

It's just because I was probably talking about 50 nm nodes.

Now if you talk about 40 nm nodes you have to multiply by a factor which is the numbers of passes as I just said, the number of layers.

So what I can do is separately or, frankly -- and Craig could give those numbers to you there is no secret.

There is a need for much more immersion machine now on every new node and there is another ratio for the 30 nm DRAM node which is already being worked on.

So all these nodes have a different response, so whatever I told you then was probably correct because it was based on an average mix of different nodes.

If I have to re-update those numbers for 2010 or 2011 we can do it, there is no secret but they are probably higher than what I ever said.

Okay, thanks.

I might speak then to [Robert].

Thank you.

Adrien Bommelaer.

It's Piper Jaffray.

I had a question on EUV.

I was wondering if you could give us some color on the progress done by the rest of the [semi-cap] ecosystem to launch EUV.

I am always a bit concerned that I seem up a little too far ahead of the others on this.

And then I have a quick follow-up.

So the first question about are we too much ahead versus the market needs, the answer is no.

In fact, because EUV is considered a cheaper solution to a current problem, if you do double-patterning immersion at the level of say 22 nm node or 18 nm node depending on which products you are talking about, EUV at the current prices is a cheap version.

So if we succeeded customers like it.

The proof is that we have got now six orders and by July we probably will announce a certain number of orders for the new version called the 3300.

So we are time wise with the current market lead okay.

Are we ahead of the ecosystem?

Good question.

There are, I would say, four areas of concern.

One is the light source, which is basically a supplier of ours, and at this moment the issue with the light source is not so much if it works or not, it's what is the power we are going to get out of that source.

At this moment we have a minimum power, which is already demonstrated, which already gets us into a mode of nearly breakeven of total cost-wise of EUV versus immersion.

So every effort we do now and improvement, and granted we don't how fast we can do that at this moment, but every effort is in green territory.

So we think that is under control.

The second part or potential limiter would be the resist and at this moment resist at 10 mJ per square inch or 15 mJ per square inch, depending on usage, is being proven.

And you can see these resist at -- I make at Albany, New York, and at customers.

The issue here would be the line roughness, whether we can clean it up so that it allows perfection and good yield.

So, again, I don't think people will tell you that there is any showstopper.

The next one is the mask, the blank mask in particular and the writing.

So here again most of the customers who are interested by this are big mask vendors themselves and the fact that they buy our machines we kind of think that it is proof that they can find a way to build a mask and a blank.

There are some issues, of course, of getting enough blanks at this moment or having the right defect density on the blanks.

But again, as I said, if they had it as a showstopper anyway it's under their own control.

And when you look at the amount of billions that we invest in R&D versus the total amount of money that is required in that business, we don't think that is an issue.

And the fourth issue is an investigation on the mask -- inspection and that it is true, at this moment the industry has not completely converted on an industrial solution.

There are solutions; solutions like you test your defects not on the mask itself but on an exposure.

That is possible but that is not (inaudible).

And there are also ways of the inspection based on non-EUV source but on 193 source and you can see a significant amount of defect.

Probably not everything you are wanting to see, but etc.

So this again could be considered, I would say, today as the area of least excitement.

But those guys, as usual, react more when you are nearer to production because they are not a showstopper in the whole R&D and recipe management.

They could become a showstopper when you want to get into full production, but full production is 2012 to 2013 so there is significant time to solve that.

Also on top of that, Adrien, this is a fundamental issue in the sense that all these four things that Eric just mentioned you can only really only manage when you have this exposure tool.

So you first have to have the tool before you can really solve the other.

So it is logical that if you don't have an exposure that the exposure tool will be the first there and then the rest is to follow.

But you cannot solve those problems if you don't have the tool.

Thanks, that is fair.

Just as a quick follow-up, Intel has actually made some fairly lukewarm comments regarding EUV in the last few months and I was wondering in terms of acceptance by the industry who would look at this tool first for mass production.

Today we are certain now that DRAM doesn't have easy workaround and Logic -- according to Logic aggressive does not have easy solutions.

So for instance when you are a foundry business and you try to do the maximum shrink on a given node, that is you give the maximum flexibility to your customers to shrink, there is no real solution.

Flash could potentially get one generation more by (inaudible) system, but that is it.

Some other customers could decide to be less aggressive on some nodes and therefore go around certain numbers [of peers].

But I will repeat my full (inaudible) statement, what is important with EUV is that we reach a cost factor, it's an economic decision.

So every customer, including the one you mentioned, will be moving into EUV at the moment ASML and their peers develop the right economic solution.

So we are not dreaming here and people would tell you at this moment they plan to convert in 2010 -- sorry, 2012, 2013, 2014, 2015, 2016, etc.

At this moment this is irrelevant if they do not know what economic equations we are given.

And if we are able to give an economic equation which at the end of the day is cheaper than anything else you and I know that there will be a jump into it.

Now again we do not feel highly concerned ourselves because if we cannot reach exactly the level of performance that we want, we can still do one or more years with our immersion machine.

And we would not be -- the business impacted in some way.

So we are doing this because we think it is necessary for the industry.

We think that in any event past 2015 there are going to be some nodes where there is absolutely no choice, but between 2010, 2015 it's for us to define the economic equation that makes the transfer happen.

Thank you very much.

That is very clear.

Ben Pang.

Caris and Company.

Two quick questions.

First, on your capacity calculations that you used for the unit growth for semiconductors, can you give us an idea of the type of -- the number of units that you are looking at for lithography tools?

The numbers of lithography tools, I don't know if I want to give you the very precise number at this moment about what we expect during the year.

So I think it could be a bit -- how do you -- too much to --

Let's put it this way, our top-line guidance, which we said is very likely going to peak our 2007 revenue, that top-line revenue guidance -- and I am not going to give you the details on what we think the breakdown is per litho tool -- but that is going to support Eric's statements.

I apologize; we told you that guidance for the year.

It's very rare we give guidance for the year so early if you have seen our past.

The fact that we give guidance which is above EUR3.8 billion is already a measure of our bullishness.

And confidence.

And now of course you can take the EUR3.8 million plus something and divide by the ASP price and you will figure it out for yourself, how many immersion tools that is.

Okay.

One quick follow-up on your backlog.

At what point will you add in the EUV tools into your backlog?

We will do that once we understand that we have visibility of revenue recognition within the next 12 months.

So because EUV will be shipped in the second half of the year there will probably, it's not fully certain yet.

But there will probably be governed by what we call SAB 101, which is a revenue recognition principle which actually says if you have a completely new technology then you can only have revenue recognition when you have fully and latest sign off at the customer side.

And that could be 2011.

So when we will get sign on when we will book that revenue within the next 12 months.

So it's not order-wise; it's really revenue recognition.

Then we will put it in the backlog.

And that basically shows up as a turns business in that quarter?

No, I am sorry, it could be early.

For instance, if we see that six months that we see good progress and that we expect over the next six months that we will get final sign-off then we will put it in.

Thank you very much.

Jagadish Iyer.

Arete Research.

Thanks for taking my question.

Two questions, Eric and Peter.

First question is I wanted to find out how should we think about what is (inaudible) of booking to acceleration, which customer should come back and start to buy?

Is it the tier one guys coming back who make capacity purchases that you will see the bookings accelerate?

And I have a quick follow-up, please.

Well, in fact, the reason of our bullishness again is because at the moment when they -- two things apply.

At this moment, hedge first half, we have got the first tier DRAM, we have got not much flash, and we have got the first tier foundry.

And therefore in the second half we expect the natural second-tier DRAM, new flash, and second-tier foundry.

But in addition in 2011 or in a progressive fashion from now until the end of 2011 rate of conversion of nodes, which requires so much more passes of immersion, which the first tier will have been doing some things in the first half of this year, will start moving back in and they probably will start moving back in in Q4, Q1, Q2 2011.

So it's going to be -- these are a lot of ways that will add up in the next, at least four quarters if not six quarters.

What Eric is basically saying is that the litho diffusion rate is going up.

Now where will that end, we don't know.

But clearly because they need more of those immersion passes the percentage of litho as total of CapEx spend will be going up over the next two to three years.

And just as a follow up, you have said that could beat the EUR3.8 billion from 2007 EBIT.

Is it fair to assume that you have EBIT margins also you will eclipse in this particular cycle?

Thanks.

Yes, I think operating margins or the net margin will be higher than --

Than the previous cycle, yes.

Okay, thank you.

Thanks so much.

Ladies and gentlemen, I think we have time for one additional call.

If you weren't able to get through and still have a compelling question or any interest, give us a call here in the Netherlands or in the US and we would be glad to follow up with you.

So now, operator, if we could have the last call I would appreciate it.

Mehdi Hosseini.

Mehdi Hosseini, FBR Capital Markets, and thanks for taking my question.

Eric, I am looking at slide number 19 where you were indicating DRAM capacity or memory capacity is going to be (inaudible) for 2010.

Does that mean that even with the significant backlog from the memory customers you have is there still going to be shortage?

And if that is true then would we see the first-year memory [DM] guys coming back and actually start placing orders for capacity buys?

That is exactly what I said so it's good understanding of what we tried to say.

We believe that we are going to have a sustained period of booking and billings because of this phenomenon.

We start with a very low level of capacity because of the retirements of 200 mm.

You add to this a sort of volume correction which globally is what we call the 15% unit.

You add to this -- the numbers of layers are multiplied by two now and it would be multiplied by another factor in 2011, so everybody is going to come back.

Is there going to be a shortage?

I hope not.

In fact, we are working extremely hard to build up that capacity to the famous level of, as I said, 30 to 36 big immersion tools per quarter which is big enough, we think, to cover customers needs.

So is what you have in the backlog for the most part is for shrink among the memory customer and assuming that the capacity buys will be higher euro values then at some point this year we should see an acceleration in bookings, correct?

Yes, in fact now -- it's good that we finish the call with the first statement I made is we expect this plateau-ish of bookings, billings to potentially creep.

And I used the term plus or minus 10% per quarter as a proxy of the fact that you never know in which quarter things really happen.

So the fact that you hint or we hinted that there is a potential probability of growing the plateau is correct, but I would not see a profile where the bookings rose from EUR1 billion a quarter to EUR2 billion and then go to zero.

I would see more creeping up and sometime, maybe one quarter, a bit lower than anything at this moment because all the waves are fairly independent.

The wave of the first tier, the second tier, this node, that node, the foundry, the IDM putting together a business, etc., and the very large processor companies are starting to put orders.

Thank you so much.

With that, on behalf of ASML's Board and management I would like to thank you for your interest and for joining us today.

Operator, if we could formally conclude the call I would appreciate it.

Thank you.

Ladies and gentlemen, this concludes the ASML 2010 first-quarter results conference call.

Thank you for participating.

You may disconnect your line now.