艾司摩爾 (ASML) 2011 Q3 法說會逐字稿

Ladies and gentlemen, thank you for standing by.

Welcome to that ASML 2011 third-quarter results conference call on October 12, 2011.

Throughout today's introduction, all participants will be in a listen-only mode.

After ASML's introduction, there will be an opportunity to ask questions.

(Operator Instructions).

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

Craig DeYoung.

Go ahead please sir.

Thank you operator.

Good afternoon and good morning, ladies and gentlemen.

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

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

As the operator mentioned, the subject of today's call is ASML's third-quarter 2011 financial results.

Joining us today from Silicon Valley in the Bay area of California is -- and cohosting the call is Mr.

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

Given that we're in two different locations, should there be any kind of technical difficulties that would separate one of the parties, that separated party will join rejoin the call as soon as we possibly can get them back on.

So at this time, I'd like to draw your attention to the Safe Harbor statement contained in our press release today and in our third-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 the call will be 60 minutes.

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

Thank you Craig.

Good afternoon, good morning everybody.

Thank you for attending our conference call.

Now that I am in the Silicon Valley, I realize how much courage it is for all of us for you to wake up at six o'clock to listen to us.

But before we begin the Q&A session, Peter and I will provide, as usual, an overview and some commentary on the third quarter and the view forward.

As usual, Peter will start with a review of the financials in Q3 with added comment on the short-term outlook.

I will complete the introduction with some specific commentary on current status and planning regarding immersion and EUV.

So Peter please.

Thank you 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 of our third-quarter results.

Our third-quarter sales came in over EUR1.4 billion, and were evenly distributed across all sectors.

Again, as in the second quarter, they included roughly equivalent unit shipments of noncritical KrF and i-Line systems, and critical level ArF immersion systems.

I would like to highlight that we have now shipped more than 100 TWINSCAN NXT 1950 immersion systems, bringing the total number of ASML immersion systems sold to over 320.

The average selling price of all systems in the third quarter was EUR23.2 million.

That's an increase of EUR2 million per unit versus the second quarter with an average selling price of new systems of EUR27 million, which is a EUR4.4 million increase over the second quarter.

Service and field option sales for the quarter came in at EUR185 million, which is not substantially different than the second quarter.

We also recognized sales for the first time of two EUV systems for a total amount of EUR80 million.

Our third-quarter net income was EUR355 million, which is 24% of sales, creating an earnings per share of EUR0.84 for the quarter.

Excluding the zero margin UAV business, net income would have been 26% of sales.

Updating our previously announced share buyback program, as of September 25, 2011, ASML has repurchased roughly 21 million shares or EUR560 million, giving an average buyback price of around EUR27.

As stated before, ASML intends to cancel the repurchased shares, and at the end of the third quarter, we had about EUR2.8 billion in cash and cash equivalents.

Third-quarter net bookings, excluding EUV, came in at 23 systems valued at EUR514 million, which is essentially one system above our guided level.

Booked ASPs declined from EUR24.7 million in the second quarter to EUR22.4 million in the third quarter, which is due to a product mix shift toward some lower ASP non-critical tools, especially KrF.

Bookings strength was in the Foundry sector 41% of total bookings, followed by NAND Flash of 30%, with 18% from DRAM and the rest came from IDM.

Our order backlog existing Q3 was EUR2 billion, excluding EUV, which totaled 74 systems with an average selling price of nearly EUR27 million per unit.

For the fourth quarter of 2011, we expect net sales to be about EUR1.1 billion, including revenue recognition for one EUV system, an NXE 3100, at EUR40 million.

This system will be recognized again with zero profit margin, resulting in a gross margin of total expected sales for the quarter of about 41%.

But excluding this EUV system, the gross margin would be about 42%.

R&D expenses are expected to come in -- to be EUR150 million, giving support to our dual product leader -- to our dual product leadership strategy with focus on the EUV system development and also on immersion platform announcements.

SG&A will be at EUR56 million in the quarter.

We expect fourth-quarter bookings to be at the level above the third quarter, reflecting customers' clear and continuing need for new technology capability in support of their aggressive shrink roadmaps for new product development at lower manufacturing costs.

However, it's too early to understand how overall semiconductor demand will contribute to our business in 2012 and to what extent our business will benefit from what appears to be a solid technology transition year.

At this point, our customers are starting to place orders for these next-generation technology upgrades, but are currently not yet capable of quoting their total capacity needs for 2012.

They will most certainly wait for macroeconomic and industry specific market developments before confirming the full extent of their investment plans for next year.

With that, I would like to turn it back over to Eric.

Thank you Peter.

So Peter has addressed our short-term and midterm perspective.

I would like myself to focus my introduction on the next four to five years, which we will see one of the industry's most extensive technology transitions, the transition from - for critical layers from immersion lithography and its multiple versions, spacer, double patterning, double double patterning, etc., to EU.

This next-generation technology has been chosen unanimously by the industry.

It is based on EUV light, the wavelengths of which is not limited, or not limited to further geometrical shrinks up to 2020 or even longer.

As you know, immersion lithography, which is currently the leading edge technology, will not be able to carry the industry's geometric scaling on critical layers beyond, say, 15 to 20 nanometer in the DRAM segment and beyond 10 to 15 nanometers in the NAND and Logic segment even by stretching immersion lithography with self-aligned or double double patterning or any of these techniques.

These immersion technologies will have a hard stop for use in critical layers.

But of course immersion will however be used for many non-critical layers well into the 2020s, along with KrF technology.

The transition from immersion to EUV on the critical layers will, however -- and this is important, this is our key message -- will be gradual.

First, customers with a huge install base in immersion lithography systems in the fabs and other dedicated process equipment will use amortized equipment for as long as possible and will bring the EUV system in only after maximum technically feasible utilization of the install base first.

The second question is that each layers at each new node will have more or less complexity for which EUV transition will be more or less economically and technically obvious.

It's not a question of node; it's a question of layers per node.

Each of the layers have I would call its personality.

So for instance, key enabling [Vios], or contact layers for logic, DRAM, and even NAND are currently the most obvious insertion layers for EUV, as EUV helps manufacturing control yield while providing the possibility of better shrink factors.

A die on EUV could be as small as 10% smaller than a die made by an alternative technology -- and also making use of less design restriction.

The third point, third aspect, which explains gradual is that the suppliers providing EUV systems like an infrastructure like us will provide continuously improved equipment adapted first to the most complex challenges until a time when the production capacity and the product development allows a shipment of a larger number of systems at improved specifications and costs for all critical layers.

So in other terms, our machines will improve as we go, and as it improves, its economics are better and better and (inaudible) better and better, they will move into a different set of layers.

So these are three factors, the use of current install base, ever more technically challenging image requirement and progressive availability of the EUV system with better economics will make EUV transition gradual.

Most customers are proceeding today on this with the development of some layers to be either processed with immersion and/or with EUV, thus having the security of a known technology as a backup, but the full use of EUV and its enabling capability if and when it works.

XML will support this gradual ramp with an aggressive dual product strategy, of course.

We continue innovation on our immersion platform concurrent with the huge development of the EUV platform.

Regarding critical layers which will use immersion, we are developing the next generation machine, the next generation NXT, which will set new throughput and overlay standards.

We are also in parallel strengthening our suite of holistic lithograph software and hardware products which enable the process window enlargement and improve the process drift management and the capability of matching a machine, which becomes more and more a requirement.

Our customers' process complexity and our associated product solution will contribute to XML revenue growth of course, but also will ensure leadership, as for instance machine matching and compatibility is becoming an ever ever growing requirement.

If you have an install base of ASML machine, it becomes natural and necessary to adopt more ASML machines for this important compatibility.

Regarding EUV, we have shipped five at EUV preproduction [an] NXE 3100 at this moment.

There's six shipping in fact as we speak.

Integration has become on the first NXE 3300, which is a new generation of production tools slated for delivery on the second half of 2012.

Our current EUV throughput roadmap has been updated.

So we indeed incurred a cumulative 1.5 years delay on this development in large part due to the investment in the EUV source development area which was not commensurate with the complexity of the task.

After several months of heavy involvement with the three potential source suppliers, we now clearly understand the key issues to be addressed and are intensively cooperating with the suppliers on sub programs, on numbers of sub programs, to drive execution of this roadmap.

We now plan to achieve commercially viable throughput of our target 60 wafer per hour in the field in the summer of 2012.

We will achieve our target 125 wafer per hour on the 3300 a year later.

We are currently building capacity for one system per month from mid 2012 out through end of 2013, ramping to two per month in 2014 and then three per month in 2015 in anticipation of the gradual demand curve which I described before.

So in summary, in the perspective of the development of, on one hand, multi-patterning immersion and on the other hand the gradual insertion of EUV, we see the next four to five years as a path to good secular growth.

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

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

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

Now, with that, operator, could you please give your instructions and then the first question please?

(Operator Instructions).

Sumant Wahi.

This is Sumant from Redburn Partners.

I think the question is actually for Eric.

The question is basically that in the Q2 conference call, you kind of pointed out that the minimum spend by industry for technology upgrade will be about EUR750 million per quarter, which was excluding EUV.

Now, if I add about 5 (inaudible) to that, I get to about EUR3.35 billion, EUR3.4 billion of sales for 2012, which is kind of the minimum spend you were pointing to.

I was wondering.

Looking at your famous fab model, where do you see the semiconductor industry today?

Is it looking purely as a tech upgrade happening in 2012 or is there some capacity addition as well?

I have a quick follow-up.

As we would like to be disciplined today and not guide on 2012 because we don't have enough facts.

We usually report to you facts, meaning when our customers are ready to put orders for our machines which have long leadtime and we usually give those facts, this time we don't have facts, so we cannot really guide on 2012.

What we said, however, what Peter said, is that there is a good technology transition wave in 2012.

As we also mentioned before, a lot of our partners in the DRAM business and in the logic business are improving or reducing their usual cycles of transition.

We see DRAM now every year, and we see logic even better than the usual two years going to 18 months, so we know this is positive.

But again, we cannot, at this moment, comment at all on the total picture for 2012, which will have to include capacity build.

Okay.

I guess my follow-up is looking at the current capacity, in the beginning of this year, you were talking about building up the capacity adds ASML to suffice EUR2 billion per quarter of revenue.

But looking at the order pattern right now, it looks like 2012 you would have much lower sales.

So would you be considering any sort of cost cuts at this point in time?

What are your fixed factory costs at this current moment?

Absolutely.

We commit to the world to scale our OpEx to the level of sales.

But indeed, as you say, we also have such a large market share that we need to provide for the customers a very short-term demand.

So we are playing between this maximum flexibility up in terms of output and the possibility of always providing a P&L, which creates value.

So we do have this variability on OpEx.

We can exercise it between three to six months, and we can go towards very low OpEx if necessary.

Most of the activity on OpEx would be the reduction of contracted work, which will be because we leverage a lot of activities, R&D as well as development -- I mean support with companies, but we also have a larger flex workforce.

So again, today, we are still running our business to have in 2012 a good year because we need to be capable of delivering, but we also have a scenario in which we need to reduce our OpEx to adapt to the possibility of lower (inaudible).

Could you give any sort of figures on COGS for example, fixed factory costs within the COGS or not at this point?

Peter, would you --?

Yes, I think we don't do that.

We've done that when our sales dropped to almost 0, and then people were still counting with gross margins that were in the 30s%, well, if you have no sales and you have only fixed costs, that's why we started to talk about fixed costs which of course makes sense.

So no, that is not the case.

If you want to know what particular levels of gross margin you can expect, I would take the last let's say seven quarters as a proxy of the different levels of sales and the gross margin.

That would be good for your model.

Thank you.

Gareth Jenkins.

It's UBS.

Just a few on EUV, if I may, fairly quick ones.

I just wonder Eric what really gives you the additional confidence that, in terms of EUV, that customers will take EUV for the 1X node in DRAM and NAND.

Just two quick follow-ups.

I just wonder if you could give us a sense of how many layers you see as being critical with EUV.

Than the last one probably for Peter, just on R&D, what percentage of R&D is EUV currently?

Is that kind of EUR20 million to EUR30 million step up of kind of abnormal R&D currently in these projects that will then step back down again as it matures?

Thanks.

So we believe that, for the NAND business, around 18 nanometers will have a contact layer, which is extremely, extremely, extremely difficult to manage without EUV.

As usual [in life] things should be done differently, but we have huge push on an 18 nano one layer contact in that.

In the DRAM arena, we have two layers in the 20-ish, 20, 22 nanometer environment, which has the same opportunity.

So to be clear on this one, it's for us to take.

So if the machine runs well enough, we will take those layers and they will justify a fairly good ramp, by the way.

If we delay this, indeed then it will be a significantly difficult process using double patterning, trouble pattering or whatever, which also will contribute with about the same level of I would say [lethal] spend.

So we're not considering the EUV introduction as being an (inaudible) for us, but it is for our customers because EUV is more cost effective, so we are working extremely hard to get this to happen on time.

The R&D effort in EUV at this moment is I would say about simplify more than one-third of our spend.

We are building up a bit more than that, again, for another year or so.

After that, we could stabilize.

But again, as we've noticed in this business, the R&D curve can go down if we only go to maintenance, but good for us we always see another opportunity.

So in other terms, if I'm right with the secular growth options, which we're starting to see, we will need in fact to upgrade our EUV machines to new generations for the good of creating growth.

And so on one hand, I see the possibility of going down in R&D if and when the business matures, but if the business doesn't mature as we could start seeing now between 2015 and '20, then we will probably maintain the current level of R&D.

Thanks.

[Gunar Plaha].

It's [Lowat City].

Eric, you provided a bit longer-term outlook.

I was wondering.

I understand clearly from a customer perspective EUV is a preferable solution given that it gives you profitability advantages through [shrinking] competitive efficiency gains from 450 millimeters.

But at what stage would you expect to make expenses for (inaudible) EUR50 million as well.

You see some companies now in the supply-chain starting to spend next year.

And my follow-up for Peter would be is it possible to qualify slightly the sequential Q4 order increase that you're expecting?

Is it purely seasonal affect?

Could you give a little bit more color on that?

Thanks.

So let me start with 450 and then Peter will handle the bookings question on 450.

So clearly we believe that the top four to top six largest manufacturers or semi conductors will develop, will push for a 450 transition in due time.

We think, at this moment, due time is probably a real prototyping around 2016, 2017, and potential production around 2018.

Before that, there could be some demonstration units, and you've seen and heard some reports on consortium, particularly in the New York State, discussing the possibly of demonstration and feasibility.

But I would say this is minimum investment.

This is more demonstration than it is industrialization.

So I would think at this moment the whole industry would agree with what I said, which is, in the best case, serious prototyping beyond 2016, serious production beyond 2018.

On these dates, I guess, or most suppliers, increment suppliers, are working and discussing with their largest customers to confirm that there is an insertion point, and obviously to share the economics.

This is not yet -- we haven't yet reached agreement on that, the whole industry hasn't yet reached agreement on that.

But we're very I would say open book to -- on these questions and expect that there will be conversions in due time.

Okay, thank you.

I've got on the question on the I say the qualitative statement on the order intake on Q4, I have a question on whether there's any seasonality there.

Now, we did mention on previous occasions that what we normally see is indeed that, in Q4, customer budgets are actually set and the [volume] purchase agreements are signed with us, and that means that the purchasing people then have a proxy to start negotiating and placing the orders.

But I want to make it quite clear that what I said also in the opening statement that we are of course focusing on the next generation technology upgrades for next year which customers have taken their decisions on.

It's just what we see is a start of placing the orders for that next generation technology upgrade.

Having said that, that it is logical that we would see that let's say next-generation upgrade order coming in in Q4 but it was hard to imagine that it would stop in Q4 because that would mean customers would place everything that they need for 2012 in only three months' time, which they won't.

So, I don't think there is any seasonal trend, but it's just a matter that that time of the year is a logical time for our customers to actually start thinking and acting on what they need the next year.

Simon Schafer.

It's Goldman Sachs.

My first question was also on EUV actually.

I think I understood, Eric, that as you said before, second half of 2012 would be revenue recognition for the initial 3300 set.

But when would that revenue actually show up in orders?

Would that be a Q2 issue when you would actually be able to print that in your order book?

Peter?

Could you (multiple speakers) --

(multiple speakers) bookings for EUV.

The timing of the EUV bookings for the next generation for -- yes.

It's clear that at the time of those bookings we would indeed expect when the EUV program delivers the results that we're currently planning for, like we said, we will see upgrade to the midteens as we have said.

That will actually drive clearly the need for our customers to take into account the fact that order leadtimes for EUV tools are beyond 12 months.

They are actually quite a bit longer, so that means that bookings for the 3300 beyond what we currently see need to come in clearly I would say in the first half of next year, taking into account that the first 10 tools have already been ordered and we have about the capacity to do 18 units until the end of 2013.

That's about one per month starting from July 2012 onwards.

So, that means that there are only a few systems that we can book until the end of 2013, and we would expect that those orders would have to come in the first half of 2012.

Got it.

But just to clarify, Peter, I know the initial 10 orders that have already been placed, but you haven't necessarily printed those in your order intake.

Is that right for the (multiple speakers)?

Correct.

That's correct.

Like we said, we actually have viewed the EUV systems as the R&D systems for the 3100 series, and clearly these are the production series.

So this would be a time and we are considering internally to say when are we going to show them in the backlog.

Everybody knows it's about EUR70 million, EUR75 million average, and we have 10 units, so it's EUR700 million to EUR750 million of EUV orders that we currently have.

That is not reflected in the backlog yet.

Where we anticipate that we will start shipping in the third quarter of next year with a -- you could say with a speed of one, two per month going forward for the next 18 months.

So this is clearly a time when we need to start considering taking them into the regular backlog.

Like I said earlier, they are orders that we take for EUV.

They have somewhat longer time because the order leadtime that we actually need for the total chain, for the supply-chain and our own build time goes beyond 12 months.

So we always said we want to show the order backlog actually reflecting what we think we can sell in the next 12 months, so we need to decide internally what is the right moment to start putting the EUV orders in.

But it won't be far away.

Got it, thank you very much.

My second question would be just on the balance sheet.

It impacts the share buyback program, this up cycle, but still sitting there with EUR2.1 billion net cash for the [first] -- as usual at this point in the cycle, you've actually seen of course an increased amount of M&A overall.

I wonder whether you can maybe just update us as to how you're thinking about potential opportunities, perhaps venturing out of your core expertise area of lithography?

Historically, that something you've contemplated in the past, and perhaps other times sort of abandoned again, but maybe you could update us on your latest thoughts.

Thank you.

I think, as I say, the EUR2.8 billion cash balance in the context of M&A, you can rest assured we have no specific plans there, so that means taking into account that on the EUR2.8 billion, we have about EUR800 million of prepayments on EUV for which clearly we still need to invest.

We are actually starting to invest in work in process.

We will have receivable balances that we need to finance.

So the EUR2.8 billion, there's about EUR800 million of current prepayments in there.

But clearly we will, given that the 2012 profitability profile, we will generate cash, so we have said clearly that above our cash balances we will keep returning cash back to the shareholders which will be in the form of dividends and of share buybacks.

I think that is the most obvious use of the excess cash balance that we currently see.

Thanks Peter, thank you.

Janardan Menon.

It's Janardan from Liberum Capital.

Actually, it's another question on EUV if I might.

You said that you're getting a new light source which will take you up to the midteens of wafer throughput in Q4.

I was just wondering.

Going forward from there up to about 50, 60 wafers per hour by the second half of next year and then to 125 wafers per hour in the second half of 2013.

What are the milestones?

What are the expectations you have from light source supply or any other and how you expect to ramp that up?

If you give a little bit of color to give us a feel on how that's going to happen?

Just to clarify, if a leading DRAM manufacture, say, is going to go to, say, a sub -- a low 20 nanometer production on DRAM by the second half of next year -- I was a bit confused.

Do you think he will use an EUV for at least two layers or is EUV for commercial production clearly going to be pushed into 2013?

So the reason why we feel now that we're comfortable about the roadmap is -- and the understanding of what we need to do is we have identified that there are numbers of things which are difficult to do and there are numbers of things which are not that difficult to do.

They are more deterministic.

They just take time.

So today you see us showing good body language because there are two or three things of difficulty that we have to prove by December.

They are in I would say the critical areas of generation of the radiation.

But then there's a lot of activities which is just multiplying I would say the results of this radiation by improving the effectiveness of the delivery of this radiation to the wafer.

This is we say just work.

So we see a step-by-step approach by which at the end of December, as we said, we will get between, say, 15, 20 wafer per hour proven.

Then from there, we will have those multipliers coming in on a regular basis in 2012 so that we achieve what I just said, 60 wafer per hour by midyear and then going further with even more efficiency move towards the 120, 25 wafer per hour.

So in other terms, the separation now of complex stuff from less complex seems -- is giving us this confidence level that we are now much more under control of the development of this effort, this [source].

Regarding market timing, the machines that are used today, the 3100s, will be used only for recipe.

That is starting -- in fact started this year and will continue until the end of 2012, and the 3300s are the preproduction machine that could be used on some layers.

So you could see them being delivered in mid 2012 on, and starting to ramp up production of some layers in the first half of 2013.

On Slide 22, we have a few of the examples mentioned which gives us the higher confidence level that you noticed.

Now we're not going to go into the technical detail because it is clear that some of it also has some competitive elements there, but it gives you an indication that we do clearly know where we need to look for that increased productivity and what the reason is for that high confidence.

And a small follow-up if I may.

I know you're not saying anything at all about 2012 today, but which segments would you be more confident about for next year, and where would you be less confident?

Would your confidence be higher on the NAND and the Foundry side and less on the DRAM?

Would that be a fair characterization of how you would look at 2012?

(multiple speakers) NAND is a good business for everybody.

As you know, there is more and more excitement about the product, the end products and the creation of more opportunity of NAND in PCs as well as in tablets.

So as you know, NAND was driven by tablets and mobile phone.

Tomorrow it will be driven by mobile phone again, by tablet, but also by PCs with SSDs and more effective user (inaudible).

So indeed NAND is going to be a growth business I would say in 2012.

DRAM was so bad today, in 2011, that we think there is an opportunity to create a good litho business, not so much because of the wafer starts, possible that the wafer starts 2012 will not be so good, but the technology is a big user of lithography.

So we could expect growth of DRAM from a fairly low base 2011.

The Foundry is in fact the key segment which will make 2012 a good year or an average year.

This is what we have a bit more difficulty to call.

Okay.

Thank you very much.

Mehdi Hosseini.

This is Mehdi Hosseini from Susquehanna International.

Eric, going back to late '08, '09, it seems to me that your customers cut through the maintenance level and everything just froze.

This time around, it's quite different than back then.

So we're not in the same environment as 2008.

Is that correct or do you have any thoughts?

And I have a follow-up.

I cannot even (inaudible) that macroeconomic question.

Yes indeed, we have in 2008, '09, a freeze of six months because the world was stunned by the banking crisis.

Remember, this was a time when people thought the whole economy would implode.

This created a sense of we don't need to take long-term decision now, and we can delay decision, so it was not a rational view of life.

It was an absolute life stop until things become normal.

So today indeed we have a crisis which is not yet obviously consumption driven.

It's a crisis of confidence due to the too high leverage of the world in general, deficits, management, foreign debt creditability, etc.

But we do not have yet a feeling at all at our customer that there is a freeze of decision.

Indeed, this is why Peter mentioned that we expect more bookings in Q4, because in fact the customers are saying, "Well, at least we can take the decisions which are the obvious decisions, which are the technology decisions, and we need these numbers." Then from there we gain time before we can take the decisions on the capacity.

So we are in a completely different environment at this moment.

Sure.

Then as a follow-up, or rather more like a qualification -- clarification, when we look into the first half of next year as bookings for 3300 come in and assuming like a worst-case scenario your customers would still be focused on R&D and the non-EUV orders would be more maintenance CapEx, then ASML could see an order environment that could be quite different than the rest of the equipment industry because of the 3300.

I'm not asking for guidance on Q1 or Q2 of next year.

It's just qualitatively.

Indeed we are proud of our model, which survives from the increase of lithography intensity and EUV is going to be a key factor through this lithography intensity, so no matter what the macroeconomic picture is, we have an underlying rich opportunity here.

EUV, again, will make -- will underline that.

As a measure of this performance, you can see 2011; 2011 has been already shaken since April.

As we all know, since April, the market has started to be concerned about the total need for semiconductors.

But 2011 for us will, one, be a record year, and two, even our guidance early into the year has not really wavered.

We are able to make that number.

So, this litho intensity (inaudible) no matter what is sustaining the business model indeed.

I would like to add to that, Mehdi, that you might remember what I said earlier, that for let's say leading-edge double patterning NXTs which are now coming at 200 wafers per hour leadtime, so are six to maximum eight months but I would say six, while EUV 3300 has significantly higher leadtime, it could be a year longer.

So that you could say if you look at the order time, when do they need to place the orders, and when would they need to get the shipment of the tool, then the order placement could be about the same time while the shipment of that R&D tool if it relates to EUV could be about 12 months later.

So yes, that could indeed happen.

Great, very helpful.

Thank you.

Satya Kumar.

Credit Suisse.

Eric, I was wondering if the order levels stay at sort of the EUR600-ish million level in Q4.

Do you have a sense as to what type of [bit] supply growth we can see from the memory industry for DRAM and NAND in 2012?

What sort of a split do you mean?

(multiple speakers)

(inaudible)

Based on the 600, let's say the EUR600 million because that's what he -- EUR600 million plus EUR150 million, that's what you mean is the EUR750 million maintenance level, and say what would that support?

That's what you're asking.

Is that correct?

Correct.

You mean in terms of mix of DRAMs, NAND and logic?

(multiple speakers)

What kind of [bit] supply group can DRAM and NAND (multiple speakers)

sorry, Peter, sorry I could not -- I cannot answer that question now.

You're saying so if we go into a low [only] technology transition activity, which is the famous EUR750 million or whatever by quarter, what is the [bit] level by segment?

I unfortunately don't remember this, and if nobody on the phone does on our side, we should have that discussion with you as a follow-up.

Satya, we do have that information but I don't want to start guessing or give you the wrong number.

Just made a note to ask to call the Investor Relations people, Craig and Frankie.

They will have that data.

Okay.

A couple of follow-ups.

One, on the Foundry side, what type of utilization rates are you seeing on your immersion machines in the Foundry segment?

We've seen, in the Foundry segment, a reduction in the third quarter, and we've seen the pick-up in the last month, but all this still at fairly good level.

Remember that a fairly good level in Foundry means it's still low because you have such a difficulty to process those new nodes that the machines are running but of course the numbers of wafers and dies that gets out of it is not huge, again, due to these complexities.

So you're talking about a sort of down I would say starting from June, July, August, and then up September and until now from numbers which are good but which shows that technologically-wise, these processes are difficult.

That's very useful.

Lastly, on the OpEx front, the fact you've not really changed your OpEx in Q4 on a couple of quarters of back-to-back low orders, is that a sign that perhaps you're seeing a pipeline of orders that might improve a bit more meaningfully in the first half of next year?

If the orders stay at the current second-half levels and you feel that might be the case into Q1, would you make a decision to perhaps throttle back on the OpEx at some point in Q4?

Let me answer that question.

I think OpEx is not -- there's never a proxy for us on where we think the business is going.

It's just the other way around.

When we see the business is going in a certain direction, we will adjust the OpEx.

Like Eric said earlier, we have a flexibility to take close to 20% of our total cost base, which includes fixed cost down, in three to six months.

So indeed, we have still ample time to review what's going to happen next year in order to be able to bring the cost levels down.

So OpEx is not a proxy.

I would say orders and business expectations are.

Thank you.

Jagadish Iyer.

Piper Jaffray.

Two questions.

First, Peter, where do you think backlog could potentially trough please?

Well, the issue with backlog is that's from where we currently are is the difference between what we will book and what we will ship.

Since it is difficult for us to give you an exact statement of what we will book, it's difficult to give you an exact statement of where we think the backlog will go.

So it's a kind of function of the fact that on one of the two elements of that equation we don't know.

So, that's I think a very difficult question to answer.

Second one, on the Foundry side, Eric, you had mentioned that it is kind of a wild card for 2012.

So what needs to happen for a resurgence in orders?

Is it going to be demand or is it the real that these foundries have to overcome before they start to place orders?

Thank you.

I think it is the demand on new nodes.

The foundries are now seeing a new business model, I will say, which is this transition, the node to node, which would happen not every two years but every 18 months.

There are some planning that says that the ramp itself within this period, when the node happens, [it] happens earlier, 18 months, but in addition to volume seems to be going faster into this business.

This model is in fact what we all await to be proven.

If indeed the customers see an application push to do this transition faster and go and ramp immediately in production, then we are going to see a sustained logic business and foundry business.

If it is sustained in 2012, it's in fact indeed a very good year because, as I said at the beginning, DRAM and NAND will probably be higher in 2012.

Therefore, logic, if it would be sustained, would make 2012 a good year.

But again, this is a theoretical question.

We have to wait to see whether the demand factor for these new nodes justifies that.

Thank you.

Ben Pang.

Caris & Company.

Thank you for taking my question.

First, on 2011, your forecasts have held up pretty well versus the rest of the industry.

Do you think there's any overcapacity of lithography tools in the DRAM space right now?

In the DRAMs space, not at all.

In (inaudible) I should say -- I should be cautious.

Not in the critical layer arena because, as we said, the ramping of the new nodes in DRAM are complicated.

They require these new machines.

So we are shipping these new machines.

We know there's no other machines in the install base that are capable of doing this.

So therefore, on this critical layers, we have no concern at all that there is excess capacity.

Of course, you could always have excess capacity on some non-critical layers on the install base, but I would say this is a bit irrelevant to sales.

These are old machinery and things, although I even think the old machineries are going to -- are transferred to NAND.

My follow-up in regards to the EUV program and the roadmap for throughput that you commented on, are those going to be step [for audio] completely step functions, in other words 60 wafers and then the next release goes directly to the 125?

No, not at all, not at all.

In fact, this is one of the confusion that we'll all have, you will all have in the press and in different communications.

There are so many factors that yield throughput that you can, for a period of time, have on one experiment achieving 60 wafer per hour, but in another you're not.

You're [at half this].

In another experiment you are a bit better.

So you're going to have a, I will say a continuous gradual set of positive data coming in towards the 60, and in fact the real number (inaudible) I want to give you the truth, I think the calculated number is 69 wafers per hour, and then we'll go from there.

So you will see that every quarter.

I think we will be able to report a gradual improvement towards those targets and then beyond.

That's very helpful.

Thank you very much.

Sandeep Deshpande.

Just one quick question on your customers, Peter.

Just talking about your customers, how do you see them reacting if things maybe get even worse in the next couple of months?

Have they talked to you on what they think in terms of 2008 and now?

Are they saying that things are -- well, you highlighted the financial difference between the situation in 2008, but I just want to understand how your customers are talking -- in terms of what they're talking to you at this point?

I think I can repeat what Eric said earlier.

The situation between now and 2008 and '09 is different.

It's quite different.

In 2008 and '09, we were all surprised, taken by surprise, and there was an absolute freeze for at least six months where doing nothing was the best Option.

That's absolutely not the case today.

Customers have very clear cost reduction roadmaps and their technology.

Node transitions are also very clear, actually quite strong in every segment.

So they talk to us about the latter.

They talk about what they need to do to get to the next nodes, to get the costs down.

They seem all to be quite firm on that particular transition that -- of course, against the background that we are currently not in a situation whereby we have to doubt the viability of the whole financial industry like we did in the first few months of 2009 despite the fact that clearly we have issues.

That bears on I think the overall uncertainty.

But no, I don't think there is -- that customers act the same.

As a matter of fact, they show a high level of confidence in what they absolutely need to do.

And then a follow-up, Eric, in terms of -- there have been some investors worried about the Foundry spending, well Foundry spending in 2011 is the highest it has been for ten years.

Is it because there is a change in capital intensity in these new generation processes 28 nanometers, or is there a real underlying -- you've said previously that there's no underlying issue.

So are you suggesting clear changes in capital intensity in the business?

Yes, absolutely, absolutely.

So first of all, there was a sort of catch-up.

Clearly the Foundry business has underinvested for a period of time, so 2011 includes a bit of catch-up.

But secondly, as I said, there's the Logic business seems to have accelerated from a two-year transition to an 18-month transition, so you get that to be done.

The third aspect is that you are now getting to nodes, which have a significant amount of critical layer increase.

So for instance, when you are in the DRAM business, the next node would force you to do, say, two to three layers which are very, very complicated, which are we call the [huge] critical.

These are the layers which drive in fact the buy of new machines.

While in the Logic arena, when you go to 22 nanometer from 22 days, 28, so you go to 22, 20, 14, then you have the numbers of critical layers, you're talking about 10, 15.

So you're five times bigger [than the] memories.

So these transitions are hugely (inaudible) incentive.

And that helps us consider that the Foundry could in fact become foundries and Logic in general would become a continuous engine in the next few years.

But again, there will be digestion as we go on some periods, six months, one year, indeed, but the secular growth on Foundry is probably significant.

Thank you very much.

Ladies and gentlemen, we are coming very near our end time here for the call, but I'd like to suggest we squeeze in one more question.

Operator -- if you didn't get a chance to ask a question on the call, please feel free to give Investor Relations department, who's available for a lot of the rest of the day here call, and we'll attempt to get back to you as soon as is physically possible.

So operator, what that, may we have the last question please?

Didier Scemama.

It's RBS.

Many thanks for taking my question actually.

Just coming back to EUV bookings (technical difficulty) hello?

Peter, can you be maybe a bit more specific?

Is it possible to tell us whether you're going to book that in Q1 or in Q2?

I've got a very quick follow-up.

Well, I have a very quick answer.

No.

We cannot be very specific.

But it will be in that time frame.

And your follow-up was?

My follow-up was to the extent that you're going to start to see critical layers moving to EUV in addition to double patterning and immersion, and that's more and more of the end markets adopting those technologies, would it be reasonable to assume that your longer-term market share could actually go even slightly above 80%?

We think that we came from a natural 70% market share, when I tried to explain what would be proper economic decisions at the customers between critical and noncritical.

We think that now 80%, or 75%, 80% could become the next natural in what you just said.

The terms that a company like ours, if we are ahead of our competition, would benefit of critical layers, but the critical layers, as I explained to Sandeep, represent a bigger share so indeed, so we go from 70% to 80%-ish, but I would not venture that this would go above 80%.

I think we mathematically -- so a natural market share between 75% and 80% is probably correct at this point.

Great, thanks.

So now, on behalf of ASML's Board of Management, I'd like to thank you all for joining in the call today.

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

Thank you very much.

Of course.

Ladies and gentlemen, this will conclude the ASML 2011 third-quarter results conference call.

Thank you for participating.

You may now disconnect.

Thank you.