艾司摩爾 (ASML) 2012 Q2 法說會逐字稿

Ladies and gentlemen, thank you for holding.

Welcome to the ASML 2012 second quarter results conference call on July 18th, 2012.

(Operator Instructions).

I would now like to turn the conference over to Mr. Craig DeYoung.

Please go ahead, sir.

Thank you, Peter.

Good afternoon and good morning, ladies and gentlemen.

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

I'd like to welcome you to our investor call and webcast.

Joining me from our headquarters here in Veldhoven, the Netherlands, is Mr. Eric Meurice, ASML's CEO, and Mr. Peter Wennink, ASML's CFO.

As the operator mentioned, the subject of today's call is ASML's second quarter 2012 results.

However, on July 9th, ASML announced a co-investment program in which customers will potentially contribute up to EUR1.4 billion over the next five years to accelerate development of 450 millimeter wafer platform and the next generation of EUV systems, both expected to enter volume production in the second half of this decade.

So, we would welcome any remaining questions you might have about this program, in addition to those questions you might have about our Q2 results.

At this time, I'd like to draw your attention to the Safe Harbor statement contained in today's press release and in our second 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.

The length of the call will be 60 minutes, as normal.

Now I'd like to turn the call over to Eric Meurice for a brief introduction.

Thank you, Craig.

Good afternoon, good morning.

Thank you for attending our conference call.

As usual, before we begin the Q&A session, Peter would like to provide an overview and commentary on the second quarter result and provide an outlook for the year.

I will, myself, complete the introduction with some further comments on the market view, longer term, and about our strategy.

So, Peter, please, if you will?

Thank you, Eric, and welcome to everyone.

Summarizing our second quarter, sales came in just above EUR1.2 billion, which is essentially the same as in the previous quarter.

This quarter sales remained largely skewed towards the Foundry and IDM sectors, which was about 73%, including non-critical KrF systems, which is supporting the capacity additions.

As in the first quarter, only 9% of second quarter sales went to the DRAM sector, with just under 20% going to NAND customers.

The average selling prince of all systems recognized in the second quarter was EUR22.4 million, reflecting a bit higher immersion shipments than the first quarter.

Service and field-option sales came in at a record of EUR243 million, which is, again, driven by significant adoption of system-performance-enhancing field options.

Updating our previously announced share buyback program as of July 10th, 2012, last week, ASML had repurchased 33.2 million shares, of which 7.5 million shares in 2012, for a total of EUR970 million, of which EUR270 million in 2012, giving an average buyback price of EUR29.

For regulatory reasons in connection with the customer co-investment program, which we announced last week, ASML has suspended its share buyback program from July 10th, 2012, until further notice, and we intend to resume share buybacks when permitted under applicable regulations.

The second quarter net bookings came in at 43 systems valued at EUR949 million, excluding EUV.

Booked average selling prices came in at EUR22 million, with the quarter's booking profile seeing a significant moderation in foundry orders at 36% of total orders.

Half of that -- roughly half of what they were in the previous quarter.

The balance of orders was almost evenly split between IDMs, NAND, and DRAM.

Our order backlog at the end of the second quarter was EUR1.5 billion, excluding EUV, by the way, totaling 55 systems with a strong average selling price of EUR27.3 million.

The backlog profile at the end of the quarter changed as combined memory moved upwards to 36% of total versus 23% at the end of the prior quarter, much of that change coming from DRAM as it moved from 2% to 14% of the backlog, reminding you, by the way, that this change is only 4 high-end units divided over 3 customers.

As to the outlook, we estimate second half sales coming in between EUR2.2 billion and EUR2.4 billion, with third quarter estimate at EUR1.2 billion, at a gross margin of 43%.

R&D and SG&A expenses will be about EUR145 million for R&D and EUR60 million for SG&A, which is slightly above the previous quarters as we decided to upgrade and invest in our IT infrastructure throughout the remainder of the year.

The second half looks to be sustained by an increased -- by an increase of system shipments supporting NAND shrink plants and by a low, but relatively stable, level of DRAM system shipments.

We see lower 28 and 32 nanometer logic shipments in the second half compared with the first half and we see IDMs coming in slightly higher, when compared to the first half of this year.

The exact level of sales achieved in the second half will largely depend on the strength of NAND pickup, to be mostly fueled by new ultrabook PCs and new smart phone RAMs.

With this short summary, I would like to turn it back over to you, Eric.

Thank you, Peter.

So, as confirmed by Peter, the balance of 2012 looks, in fact, steady.

This should lead to nine consecutive quarters of about EUR1.2 billion or more of sales.

This is pretty good in a non-fully-supportive economy worldwide.

Further, our developing views on 2013 suggest that it will be supported by NAND and DRAM at the beginning of the year, assuming a good pickup of the PC ultrabook business and a continuation of the high hand mobile unit growth, along with a steady shipment to IDMs, which is also driven by PCs and a startup of 20 nanometer foundry logic by midyear 2013.

The reason for the acceleration of the logic business in 2013 is a shortened node transition period between the current 28 nanometer node and a future 22 nanometer node, which will be about only one year, driven by the current race for best integration, best feature, best power consumption in the mobile arena where Intel architecture, the Intel architecture, and the ARM architecture will both play.

This transition is also very, very lithography intensive, as we expect between 1.7 and 2 times more immersion systems use per wafer start in 22 nanometer compared to 28 nanometer.

EUV for R&D use will also contribute to 2013 sales, as we will ship up to 11 systems of the 3300 as we planned now for a long time.

These will not contribute to production by customers.

They will be only used for recipe engineering, but this will create nearly EUR800 million of revenue for ASML, which will add to the base sales for 2013.

Technically, to keep our immersion leadership, we continue to enhance the overlay throughput performance of our TWINSCAN NXT-1950 system.

One of these systems, by the way, has exceed a productivity milestone, an enormous one, of more than 5,100 wafers in a single day at the customer manufacturing site, which is 600 wafers more per day than the previous record achieved only three months ago.

It illustrates the continuing productivity improvement of our platforms for high-volume chip manufacturing.

In addition, we will introduce a new version of the NXT, even capable of higher performance, higher throughput, and also better overlay, along with the install base upgrade by mid-2013.

We will continue, in parallel, to develop and introduce the extensive line of holistic lithography software/hardware products so as to enable recoil process control machine matching improvements, which are necessary in the complex nodes of the future.

As part of this product portfolio our computational lithography unit line has delivered significant enhancement to its leading Mask 3D models and applications, which are required at the 20/22 nanometer node and below.

The full accuracy of the Brion Mask 3D model can now be realized with virtually zero incremental computational cost versus substantially less-accurate thin mask models.

Regarding our EUV program, customers have now exposed more than 15,000 wafers on the six machines, six NXE-3100 process development systems which are currently installed and we are confirming steady progress qualifying the platform towards customer production ramp, in fact the new platform, towards customer production ramp for 2014.

This platform has shown exceptional performance, with a demonstrated 6 nanometer on-product overlay and with dedicated shock overlay of less than 2 nanometer, which is extraordinary performance.

We have made constant progress on the EUV source, which, as you know, has been for a while the critical path technology that we have to master.

With regards to the productivity itself, 50-watt power capability has now been repeatedly demonstrated at one supplier and 105-watt concept potential has been confirmed in lab experiments, supporting our road map to volume production systems, starting at 70 wafers per hour in 2014 and upgradeable to 125 wafers per hour two years later.

However, in situ, not only lab, but in situ experiments on the new machine called the NXE-3300 will be still necessary for full confirmation of this road map, although, as already planned, we still target to gather in effects by late summer to prove and confirm that road map.

In view of the progress made, we have received a customer commitment to purchase 4 additional NXE-3300 systems.

That makes it a total of 15, preparing, in fact, for the first semiconductor device production on EUV in 2014.

So, in summary, we are encouraged, as Peter said, by the short-term view, 2012, and by the mid-to-long-term prospect of this lithography market, which is very, very much driven by a huge industry technology transition appetite.

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

Thanks, Eric, and thanks, Peter.

Ladies and gentlemen, the operator will instruct you momentarily on the protocol for the Q&A session, but beforehand and, as always, 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 in as possible.

Now, operator, could we have your instructions and the first caller, please?

Of course, sir.

Thank you.

(Operator Instructions).

The first question comes from Sumant Wahi.

Please state your company name, followed by your question.

This Sumant here from Redburn Partners.

Good afternoon, guys, and thanks for taking my call.

My question is actually more to do with the announcement last week, I mean, and your R&D expense.

Given last week's announcement, you obviously intend to increase your R&D budget by 25%, I suppose starting from next year.

And I was just wondering how you will be recording this on your P&L and how will you be recording the cash from your customers which is being paid for this extra R&D.

I mean, should we be anticipating a contraction in your operating margin next year and, hence, EPS, as well as the real revenue benefit from this higher R&D will only be coming a few years -- in a few years' time?

So, that's my first question.

Then I have a follow-up.

No, I don't think you should simulate at this moment any reduction of profitability, short-term, through that deal.

The accounting is going to be a bit different per customer.

In case of a significant party, the accounting of this funding will be done below or above the gross margin line.

So, there will be a bit of difficult to trace all these numbers, but in -- as a generality, we intend to manage the P&L as we have always been doing, which is providing to the market a steady increase of profitability.

We expect that steady increase to come through this funding, a bit, but also through the fact that this acceleration of new products is going to deliver a higher top line.

So, we're going to have a higher top line.

We probably are going to have a higher gross margin because of these accounting situations.

We are going to have a higher R&D gross expense because, in fact, we will invest more in R&D for these accelerations, and a part of that will be compensated by the R&D funding.

But, in any case, we do expect a steady increase of our profitability.

Obviously, on higher revenue for next year, I suppose, which is --

Well, I did not guide 2013 yet, but we expect higher profitability.

Okay.

And my follow-up question, very quickly, is on the demand from the NAND sector, which is coming towards the second half.

Is this more capacity expansion, i.e., new factories, or mostly just technology upgrade?

And kind of within that, I mean, the bit growth you're expecting for DRAM and NAND this year or next year would be really helpful.

Thank you.

So, on NAND it will -- the pickup we are expecting is a pickup that the customers will, in fact, see in their own shipment of semiconductors with -- by the beginning of 2013.

So, we are going to have to ship lithography tool and then six months or so later, you will see the chips.

And so we're talking about chips out in 2013, lithography in, starting Q4, Q1, Q2, of 2013.

This pickup will be new factories, yes, absolutely, and requirement for more machines because of the technology transition.

So, you're going to have the two drivers, more factories, more wafers, and more machines per wafer to adapt to the new technology.

Your other question was?

It was the bit growth in DRAM and NAND for next year, I suppose?

Bit growth.

So, we're not forecasting those.

We just follow what Gartner says and at this moment I think Gartner kind of says of 40-ish, 45% growth of DRAM, I guess, and about 60% growth on NAND.

If these numbers are realized, we will grow, as we said.

We also believe, from what we heard from the importance of the new form factor PC and new solid-state drive that the Gartner numbers may be conservative.

So, in other terms, if the Gartner number is what they are, they will sustain growth for us, and it is possible that the Gartner numbers are a bit conservative.

Okay, thank you very much.

The next question comes from Mr. Janardan Menon.

Please state your company name, followed by your question.

Hi.

It's Janardan from Liberum Capital.

Just to follow up a little bit on the memory side, you said that in the first half of next year your sales will be more to the DRAM and NAND segment and then with the foundries picking up around the middle of the year.

So what kind of visibility do you have on that memory spending for the first half of next year?

Do you have sort of commitments which are not yet reflected in your purchase orders and, therefore, not shown in your backlog?

Or is it that you're expecting those orders to come through to you, order commitment to come through to you over the next couple of quarters?

And a follow up is on EUV.

You said you'll hit 70 wafers an hour by 2014 and 120 wafers a couple years later.

So, what kind of throughput do you expect to get to when you start shipping the units at the end of this year or early next year with the NXE-3300.

Okay.

So, regarding the commitment, backlog, et cetera, the answer is yes and yes.

We have startup of orders, bookings, which came, driven by this growth curve.

So, we have a bit -- we have received numbers of orders for that, but most of the orders haven't come yet.

They are more commitments/production expectations.

So, they will come into the backlog through new bookings in Q3 and Q4.

Regarding the speed of EUV, so the reason why we still don't have factual data on the speed itself of the EUV machine is because the lab experiments are driven in the lab, as well as on the 3100 architecture, which is not similar architecture than the 3300 machine, which is planned to be available to experiment by October-November timeframe.

So, only by then will we know exactly how all of these technologies, as you say, get installed on to the quote/unquote final machine.

The minimum performance from what we can see at this moment would be, I would say, a 30-ish wafer per hour, if we get unlucky on the first machine, with more work to stabilize the controls mechanism potentially taking three months, six months or so, to get to a point where the 30 wafer per hour will transform itself in 70.

So, if you get to 30 by the end of this year, would that be enough to convince your customers to move EUV into volume production in 2014?

We already got the orders for four.

We are negotiating a number, additional number.

The data today is good enough to suggest that you should starting into production in 2014.

I would even say to you with the data we have today, if we were to stop new development, we would get to this 30, 40 wafer per hour, so, yes, without any more development, based on the new machine, the 3300, which is more effective, energy effective.

And that already would not be a bad number for production.

So, the level of customer risk for 2014 at this moment has reduced significantly.

Now, don't get me to say that 40 wafer per hour is a good number and that's my new commitment.

That's not what I mean, but it says that the worst case situation, in case a terrible situation happened, by which we couldn't do even any progress, is already not a catastrophe.

Okay, thank you very much.

The next question comes from Mr. Stephane Houri.

Please state your company name, followed by your question.

Yes, good afternoon.

Stephane Houri, Natixis.

Two questions, if I may.

The first one is that you're talking already of the 22 nanometer or 20, 22 nanometer transition and you are saying there is some weakness in the 28-32.

Can you tell us where we are exactly in the current transition?

And you also said that 20-22 was twice as much litho intensive than the previous generation.

Can you remind us how much more intensive is the 32 for the 28?

And also, a second question on the bunch of orders for EUV, can you specify if it's for memory, if it's only for customer?

And I heard you say that you had -- you are negotiating currently.

Are you currently negotiating a new bunch of orders?

Thank you.

Okay, so a good set of questions.

So, the -- I didn't say the 28 was weak, I said that we are getting to the end of the build-up of 28 nanometer capacity.

So, if we ship all the plans up to the end of Q4, we would have reached about 200,000 wafers per month, 200,000, 220,000.

There is a set of customers who tell us that 28 will continue well into 2013 and there are plans, at this moment, to get from this 200, 220, to about 350,000 wafer per month capacity.

So, if I were to listen to customers at this moment, I would, in fact, tell you that 28 nanometer is only 70% done and you need another 30% or whatever capacity to be built on.

But even with upside possibility by Q4 and clearly significant shipment in Q1 and Q2 2013.

So call it a potential upside for 2013.

Okay.

So, [2000 28] in our press release is like we would have expected this to stabilize at 220, 230,000 wafers, but people are talking about more.

20 nanometer, 22 nanometer is -- usually would have started in 2014, but in view of the huge drive by the mobile environment in which you have the big fight between the ARM architecture and the Intel architecture, both architectures driving to excellence and excellence being far away, let me try to explain.

When you get to an application where you still have multiple chips, you have a chipset into a smart phone, you know that you're forced to try to get the chipset to be integrated on to one chip.

Until that continues, there will be a huge incentive to customers to try to make that leap and get to the next technology, which allows you to have the famous one-chip solution.

So, we expect the one-chip solution to happen beyond 20, 22 nanometer and we expect this to be a hard-fought battle between the Intel architecture and the ARM architecture.

So, that creates a moment in the market on the side of Intel, and on the side of the others, to build up these new technologies faster than ever before.

So, that will benefit lithography, because, in addition, those technologies are very complicated to build, as I said, 1.7 times to 2 times more lithography intensive by wafer on 20, 22, versus 28, 32, because 28, 32 is the same node.

So, if you wanted to compare 28, 32 to 45 nanometer, I cannot answer your question easily, but we can get you that data in due time if you call Franki or Craig.

Okay.

Regarding EUV, so, yes at the moment the units that we have accepted an order are for DRAM, as we always said DRAM at least two critical layers are so complicated that I guess if we do not execute EUV by 2014, there will be a delay in the shrink.

So, customers have realized that and, therefore, are very hugely incentivized to do that.

The second set -- batch of orders that we are currently negotiating are for, also, 2014, and they are DRAM.

The third set is for logic and in the logic arena we expect to have to be in production for 2015 or so in a node which will be called anywhere near 14 nanometer or 11 or 12 nanometer, whatever, but they will mean the same thing.

We are going to have in the logic arena a node which will look like a 20 nanometer design rule using architectures of transistors like FinFET.

That will be a node that will come just beyond 20, 22, and the, on the back of it, there will be, in 2015 or so, a node using the same type of architecture, FinFET, but with a shrink factor.

And that node will be called 14 or 12 or whatever for, I would say, marketing reasons, and that node is -- absolutely requires EUV.

With EUV, you cannot shrink.

So, in other terms, if I make myself clear, we are going to have 20, 22.

We are going to have 20, 22 with a type of FinFET architecture which allows an improved performance but not yet a shrink, and you are going to have that technology shrunk.

And that will happen from now on until 2015, '16, and that's, again, hugely lithography intensive.

Even whether you use immersion, double-double patterning, or you use EUV, you're talking about huge numbers of critical layers, and a huge number of lithography-enabling tools like the holistic lithography set of products that we have.

So, we see logic dwarfing, basically, DRAM anyway by 2015 or so, back (inaudible).

Just on your question what the difference was between the 28 and the 20 nanometer node in terms of lithography usage, on the 20 nanometer node, the average that we've looked at is about one or two double-patterning layers for the 28 nanometer node going to be about 10 at 20.

So, that is quite a significant increase in double patterning.

Of course, some of that we will compensate by giving customers access to new technology that gives them higher throughput and better yields.

So, some of it will be compensated through the new technology that we are shipping.

Okay.

Thank you very much.

Very clear.

Thank you.

The next question comes from Mr. Simon Schafer.

Please state your company name, followed by your question.

Thanks so much.

It's Goldman Sachs.

Actually, I had a similar question about the layer intensity, but I was wondering whether maybe you could help us understand how we should translate that sort of layer intensity into a percentage of sales computation.

I think the one thing that is obvious is that lithography has clearly been the one area in semi CapEx overall or wafer fab equipment, at least, where capital intensity is not falling.

But the historical average, I think, is sort of in the 15% range.

Any sense as to whether that 15% is not structurally 20% or 18% or any sort of measure for the type of increase that you may expect would be very helpful.

Thanks.

Yes.

No, we think we're reaching -- we will be reaching more towards 20%, 22%, mainly, also because we -- I think we are more intensive in lithography in the logic arena.

So, in this -- and logic may represent a bigger share than it used to, so that's also a factor that tell us that we can get into a higher percentage of the total CapEx, semiconductor CapEx.

Got it.

That's very clear, thank you.

And my second question, actually, is on the capital structure.

Obviously, great to see the announcement from last week and a big underpinning of your positioning, but the one thing, of course, it doesn't really change is the fact that when all is said and done your capital structure, arguably, is still relatively cash-rich if you're in a good environment cyclically.

So, any update as to what you may look to do, then, once the transaction is complete?

Thanks.

Yes, I think it's going to be a repeat of what we said last quarter.

Basically, we have the policy to return cash that we don't need and we do that through share buybacks and through dividends whereby the latter we intend to let it grow, but once we have decided to let it grow, it at least has to be stable at that level.

Clearly, what we have decided internally that we will use the fall period to reassess.

Every one or two years we are going to reassess the policy, the capital policy, of the Company and we're going to review whether the ratio between share buybacks and dividends, as we have seen in the past, whether that should remain or whether we should adapt that to different views.

So, that is more, let's say, further detailing of our overall policy, which is the money that we don't need, we will give back.

As I said in my introductory comments, we have temporarily suspended the share buyback program because of regulatory reasons.

Once those have been resolved -- and they result from the transaction that you mentioned, last week -- then we will just resume share buybacks as we did before.

Great.

Thanks, Peter.

The next question comes from Mr. Mahesh Sanganeria.

Please state your company name, followed by your question.

Thank you.

Mahesh Sanganeria, RBC Capital Markets.

Eric, I have a question on the four new EUV tools.

Are the commercial agreements and margins and throughput commitment and all those, are they similar to the first 11 tools or these four have a different arrangement?

Good news, they are a different arrangement.

Bad news, they are a different arrangement.

As some of you know, the 11 original tools are as-is, because the customers accepted to participate to this project as an R&D project.

As we know, this is beyond commercial relationships, this is a development of a new technology, so they accepted to take the 11 tools as-is, so we can ship at the best specification we can get and that's it, which is why we believe that we are going to recognize revenue next year, no matter what the exact performance of the tool is.

However, we mentioned clearly to the customers that we now are ready to take firm commitments on the specification.

We did not get push-back from the customers.

Indeed, they understand that we have to take that responsibility, which solidifies, in fact, the fact that we are serious for production in 2014.

So, the specification, indeed, says you've got to pay this price for a machine capable of 70 wafers per hour, upgradeable within two years to 125.

Okay and then on -- a question on the near term.

As you're look in the second half, in terms of your order profile, will it be similar to Q2, like memory and logic split almost 50/50 or are you seeing more memory in the second half in terms of orders?

Difficult to call the bookings.

As usual, as we said to you 10 times, it's a bit difficult to time the bookings.

It depends on the customers.

Sometimes they book with three months lead time.

Sometimes they book with six.

Sometimes they book with nine.

If it's logic, it's sometimes longer cycle times (inaudible).

So, your question is nearly impossible to answer.

I could say that I should -- we should get in Q3 more memory bookings than we should get logic, but I would not even be sure of that, so, sorry, I retire my comment.

It can be anything.

But the billings will be what I said, which is a decrease of logic and IDM and increase of memory.

Okay, thank you very much.

The next question comes from Mr. Sandeep Deshpande.

Please state your company name, followed by your question.

Yes.

JPMorgan.

Thanks for letting me on.

Eric, I have just a question.

I mean, overall in the comments you made in the release this morning, ASML seems to be much more sanguine or bullish than the peers in semi cap.

So, I mean, can you make a comment on why you think you are different versus your peer group?

I mean, is it because of your long lead times?

Is it your market share?

So, what do you think, internally, is the reason for that?

And I have one follow up.

Yes, it's difficult for me to judge how our peers are seeing -- what they are seeing and the specifics of their technologies, but I can guess that the lead time is a key point, that we are starting to see what they haven't yet seen.

Okay.

And then, just moving on in terms of your existing products, we've seen that lithography has been increasing as a percentage of sales for your customers, but also your customers are seeing a lot of concentration.

We've seen DRAM has concentrated quite significantly over the last couple of years.

Do you see that as a risk for your ASP, say, two years from now, despite your own very strong market position?

No, in fact, I see the opposite.

The consolidation has two values.

One, it creates an opportunity for customers to have R&D capability and the fear factor for us remains only whether customers can shrink.

If customers can shrink and there is enough design to take the shrink, lithography will be a huge growth business for the future.

So, the fundamental is we need customers capable of shrinking.

We need customers capable of R&D.

So -- and, therefore, we are very, very happy with consolidation.

This is huge compared to any commercial, short-term situation.

The second bit about consolidation is what happened last week is, in fact, due to the fact that the consolidation of large customers allows us to make a point that lithography becomes a nature or it's a weapon required by the whole industry for the whole industry's benefit.

And if you had 15, 20 customers, this message would have been extremely difficult to pass, where the 15, 20 customers would fight themselves to try to have a smart way of solving the lithography system.

But with fundamental, say 10, 8, large customers, they understand that lithography has become the weapons that they have to help, to subsidize, to get going for their own business model, which, at the end of the day, is designing chips for shrink.

So consolidation is a very good thing for that and it has helped us tremendously in the discussions of the deal, first with Intel, and at this moment with the others.

Thank you.

The next question comes from Mr. Didier Scemama.

Please state your company name, followed by your question.

Good afternoon.

It's Merrill Lynch.

Thanks for taking my question.

My first question, if I may, is about gross margins.

Can you, maybe, talk about the underlying gross margin assumption we should have for 2013 on the EUV machines and whether you also have, maybe slightly higher or improved gross margin for the non-EUV part of the business?

But a quick follow up.

Thanks.

yes, we've -- at the danger of repeating myself, last quarter I said that the gross margin target for 2013 is the high 20s.

EUV.

That is what we are focusing at on EUV.

Yes, yes, thank you, Eric.

Just to make that sure.

And we need about a two years' period to get to the corporate average.

That's what we've said before.

This is still our target.

Now, on the gross margin of the non-EUV tools, you could always -- if you really look a bit deeper into our current results and you look at our gross margins that we're currently showing, that gross margin, currently, includes running EUV costs that we have in the EUV factory that we currently have.

So, the gross margins stay where they are from a corporate point of view.

So, from a corporate average point of view, they stay flat, while our costs have gone up.

So, you can deduce from that, that the underlying gross margins of our non-EUV tools have actually gone, which is true, and that has been driven by the increase of the system enhancement, which we sell on top of the bare tool.

That has given a positive side or what is a positive boost to the gross margin profile of the non-EUV tools, and has helped us cover the current EUV running costs that are not part of cost of goods.

Yes, brilliant.

I mean, that's very clear.

And then a quick follow up to one of your statements, Eric, a bit earlier.

You said that at 28 nanometer for logic, the build up is more or less finished in your view, but one or, I guess, some of your customers want to increase that to 350,000 wafers per month.

So, I'm just trying to understand what, exactly, you're trying to say there.

Is one of your customers trying to substantially increase capacity and you don't believe them or how should we understand that?

No, I think what I tried to say is that in the current press release we are not taking care of the possibility of an upside.

And is that one particular customer or is it multiple foundries?

No, it's all of the foundries.

Brilliant.

Thank you very much.

The question asked -- and it's not our opinion, sorry, then I missed it.

I did not communicate our opinion.

We did not want in the press release to say the EUR2.2 billion to EUR2.4 billion guidance reflects an upside.

It doesn't.

Now, there's always a possibility of an upside.

Why would they go for an upside?

What's the rationale?

The rationale is that they have under-invested in 45 nanometer, 40, 45 nanometer.

65 was a big node.

28 could be a very big node to compensate for the fact that they didn't transfer too many things on 45 nanometer.

So, there is a business model out there that says, in fact, 28 should be bigger and we don't have an opinion on that.

We have to see whether they are right and whether, in fact, there is going to be so many mobile systems, et cetera.

If you talk to the fabless guys and they talk about the potential numbers of smart phones at 2 billion, going from 1.2 billion or so, 1.2 billion, 1.3 billion, to 2 billion, if that's the case, yes, you can imagine that they will need much more wafers.

Just so I understand, your guidance for Q4, what you're talking about at the moment, does not include this potential upside and what you're concerned about is what, the macro side and the fact that yields at 28 are going to get better?

Well, no, because we usually don't guide just on rumors.

So, that's all it is.

Don't read anything that we don't believe in anything.

We just tell you that there is a debate in the market as to how many wafers more are needed in 28 and we have not put this into our forecast.

So, just to be clear, EUR2.2 billion to EUR2.4 billion does not include any upside to this 350,000 wafer starts.

220,000.

Sorry, yes, that's what Eric's said.

Okay, super.

Thanks very much.

The next question comes from Mr. Gareth Jenkins.

Please state your company, followed by your question.

Yes, it's UBS.

Thanks for taking the question.

Just a quick question on 20 nanometer lithographer requirements.

I think you're now saying, effectively, that the requirements over 28 would be 70% to 100% higher.

I think previously you said somewhere around 60% higher.

I just wondered what the increment that you have learned in the last few months has been that has warranted that increase in intensity as you look forward?

And then just secondly, I guess, one for Peter, maybe, in terms of hiring the additional R&D staff that you need, the 25% to 30% additional R&D engineers, what, I guess problems that poses and whether you feel that you can get the available labor for 450 and the second phase of EUV.

So, I can answer the 20 nanometer.

You talked about ratio 1.7 to 2 times more early, so, no, I think we always said that.

I do not remember having said less than that.

In fact, we were waiting a bit to know if that ratio is based on the same yield and the same utilization rate than 28, because you don't want to be biased by a difficult start and the 2 times, the 1.7 to 2 times, is based on a good start.

So, if 20, 22 is a big issue where they have trouble to do yield and to do utilization, then we will have to ship even more than what we plan.

So, this number that we have is an acceptable economic target.

Regarding the R&D buildup, Peter?

Yes, like we said, we need between 1,000 and 1,500 engineers.

We don't intend to put all those people on the payroll.

As you know, we have a flex model.

We have dozens of companies that actually help us find those people.

Those are specialized engineering companies, which, basically temp labor companies.

So, they're already gearing up.

I mentioned -- I think I mentioned on the previous call that over the last 12 to 18 months about 50% of the people that we hired were non-Dutch, so we basically get them out of the European zone.

So, we're not confined to the Netherlands only.

So, it is us and our supply chain, by which I mean the flex labor supply chain, that have to deal with this particular challenge, but also some of that work and I think a significant part of that work will go into the supply chain with our knowledge and research institute partners, which, of course, have also people available now and especially in this macroeconomic circumstances and they're glad to take up the challenge to provide us with additional engineers.

So, yes, it is going to be -- also, it's going to take a bit of time until we ramp to the peak of the program, we're two, three years down the road, probably two years, which also gives us a bit of time to actually hire those engineers and hiring 750 or 800 engineers per year, we have done that before.

So, yes, it'll be a challenge for our HR organization, but in this times, that's a nice challenge.

Thank you.

The next question comes from Mr. Francois Meunier.

Please state your company name, followed by your question.

Yes, hello.

It's Francois at Morgan Stanley.

Thanks for taking my question.

I've got a very simple question, actually, because the announcement last week was also about 450 millimeters.

Is 450 millimeters a good thing for ASML?

It's not a bad thing.

It's -- I think it's a good thing for the industry, so it's going to be a good thing for us.

As you know, you may be asking this question because some of our peers are concerned about 450.

If you are a chamber-type business, 450 millimeter is a big problem, because the same chamber in the same time would process twice as many dies.

So, theoretically, if your price is about the same because the chamber adds the same price, you would cut your business by a factor of two.

So, 450 millimeter would be a very negative point for a chamber-type business.

We're not a chamber-type business.

We are a painting business.

The painter is a big lens and it paints centimeter by centimeter or nanometer by nanometer and, therefore, if you have on the same wafer more nanometer, well, the painter will spend more time.

So, the machine itself you will need as many machines per square centimeter on the 300 millimeter base or 450 millimeter.

It doesn't change at all our economics.

And we were a bit concerned that we would have to do this work without the industry complete aligned on a specification and a timing and we would have to over-invest in R&D in a wait-and-see attitude and then, as usual, wait and then have the machine ready, nobody takes it, and then two or three years later, somebody takes it.

That would have been a cost-inefficient way.

Now with this consortium that we've built, we are now pretty happy that the execution is an execution between consolidated customers, consolidated supplies, and we can align the dates, align the money and, therefore, do this in the most efficient way.

So, a bit of a follow up on this one.

When you said -- I mean, you probably have to redesign the chassis, because, obviously, the wafer will be -- it will be quite happy as it happened before when you switched from 200 to 300 millimeter.

Is the price going to change much, basically, for a given machine?

The price will go up by a factor about 10%-ish, if we execute correctly, between the 300 millimeter and the 450 millimeter machine of the equivalent spec.

But in 2018, when those machines will be in production a 300-millimeter machine will cost X times more than what we are doing today.

So, the NXT machine, 300-millimeter machine of 2018 will cost, I don't want to give you a number, but EUR70 million, anyway.

And the 450 equivalent would cost EUR77 million or something of this nature.

So, the ASP growth comes through the resolution and the performance of the machine and not so much of the die size -- I mean, the wafer size.

The wafer size will get us an additional inflation of 10%, roughly.

Okay.

So we agreed to push from the biggest maker of chips today.

Do you think the others will follow or they are still discussing and feeling a bit (inaudible) about actually moving to 450?

So, sorry, are you asking the question, are they joining the consortium or are they joining 450, or both?

450.

450.

On 450 it is not abnormal that a leader leads and you need one to build up momentum in the industry and I think Intel has been clear that they wanted to do so and they believe that it's good for the industry for them to sponsor that technology.

And that will be useful for the others to get in when, in fact, this infrastructure will exist, again helped by Intel.

At this moment, we suppose that the best date for really production -- and I don't mean recipe, R&D, and all that stuff, would be 2018 or so.

Okay.

Very helpful.

Thank you, Mr. Meurice.

Merci.

The next question comes from Mr. Mehdi Hosseini.

Please state your company name, followed by your question.

Thank you.

Mehdi Hosseini, Susquehanna International.

Two questions, first on the EUV, how should we think about the booking and revenue recognition.

Where would you start to actually include it and to get the revenue recognition?

What are the key milestones and would it take more than one quarter?

And I do have a follow up.

Yes, on the revenue recognition, Eric said it's, I think, at the beginning of this call that somebody asked a similar question.

We have, for the 11, 12 units we have an agreement with our customers that they accept the tools as-is, because it's more an R&D project and we basically share the risk of the introduction of that tool.

So, that will be when we ship the tool, we book the revenue.

But also we will book the order.

I mean, you know how many orders we have, so if you want to see when it flows into the backlog and when it goes back out, it will be a turns business when we ship.

So, we give you separate number of units that we have on order.

Beyond that, we said the tools that we are booking today, they will-- they are currently being booked with certain performance criteria and that we need to meet at shipment, which will be for 2014, which will, basically, mean that we will have revenue recognition as we have it today, and then, also, we will have bookings as we have it today.

But for the first 12, so that's for the next 12 months, you will see turns when we book the revenue for the first 11, 12 units.

Great.

And just one very quick follow up on the foundry-related booking and how the capital intensity could go up by 2 times.

In terms of the dollar value of booking, should we expect similar trend like last year, where foundries took a pause, it did have an adverse impact on your overall booking, but then it came back very strong in Q4 of last year with shipment for 2012?

Should we assume the same kind of trend, especially given the higher dollar of booking because of the increased capital intensity?

We don't know.

Again, bookings timing is an impossible question.

We just don't know.

Everything is possible.

Got it.

Thank you.

The next question comes from Mr. Jagadish Iyer.

Please state your company name, followed by your question.

Yes.

Piper Jaffray.

Two questions.

First, Eric, you had called out the color in terms of DRAM and NAND for the first half of '13 and also foundry for the second half.

I was just wondering, how should we think about overall immersion tool shipments in 2013, directionally, given that you're concurrently going to be shipping several EUV systems?

And then I have a follow up.

You want me to guide on 2013, but I will resist.

So, I'm not going to help you.

No, no, no.

It's too early.

It's too early.

What we said at the beginning of the call is we have numbers of positive drivers.

The biggest one is 20, 22 nanometer ramp, which could be a big one.

As I said, it's a multiplier by 1.7 to 2 times the run rate of 28 nanometer.

So, if it starts early enough in 2013 you will have a run rate which is 1.7, double times, what will happen in 2012, then, of course, it'll only be six months, but then in the first six months, if you have NAND and DRAM, as we said we are going to have and if the up side on 28 nanometer comes in, 2013 is a fairly good year and, therefore, there is growth of the immersion tools.

Then you add to this the famous EUR800 million of EUV and then 2013 is a pretty good year.

But it's too early, six months.

22 nanometer may be delayed because of a technology issue.

Qualcomm may say the price is too high, so I'm not going to convert my 28 to 22 so quickly.

You may have some impact.

So, it's too early to build up your firm scenario.

Okay, fair enough.

So, just a follow up on the EUV.

I was just wondering, how are you going to be de-risking yourself in terms of the laser source vendors?

It looks like one of them is ahead of the pack.

I was just wondering that, given there could be so many challenges ahead, so how are you de-risking yourself and how are you pushing the other vendors to come up to a certain level in terms of the source power so that you can kind of pick and choose between those vendors?

Any color on that will be great.

Thank you.

Yes, absolutely.

So, indeed, at this moment one of the vendors is ahead of the pack, as you know, and, as you also know, we are investing significantly into this relationship for them to succeed.

On the other hand, it is our responsibility to de-risk the situation.

So, there are numbers of ways of de-risking.

One is we are still working with Ushio and with Gigaphoton on their own business model.

Clearly, one of them has a business model of insertion of EUV in due time.

So, this is not a race of only one horse.

The timing is not immediate, however it does help in the R&D-- sorry, it does help in the production-type environment, because production, as we said, is only a start in 2014, '15, but then in '16, '17, '18 is when the unit comes in.

So, if you had a second player come in for production, it's a de-risking situation.

On the other hand, also, we are building, as you know, a knowledge of the sources and so, for us, it will be easier, also, for us to know what to do if we need to be even more involved into source development.

Thank you.

I appreciate it.

Ladies and gentlemen, I'm afraid we've run out of time for this call.

If you were unable to get through on the call and still have questions, feel free to contact the Investor Relations Department here at ASML.

We're happy to answer your questions.

Now, operator, if we can -- if you can close the call for us, I'd really appreciate it.

Thank you.

Of course, Mr. DeYoung.

Thank you.

Ladies and gentlemen, this concludes the ASML 2012 second quarter results conference call.

Thank you for participating.

You may disconnect now.