ASML Holdings (NASDAQ:ASML)
Company Conference Presentation
Feb 27, 2012 14:00 PAM ET
Good morning, I am (inaudible) in London and I am very pleased to have Peter Wennink, CFO of ASML with us today. Obviously we won't recap what ASML is doing because that’s known to everyone and maybe we will start by the topic of the week which is Elpida filing for bankruptcy so maybe two questions, one which is short term, do you have any exposure on our books to Elpida and more you know long term or maybe for H2 because you have been with Eric talking about maybe (inaudible) CapEx spending for Q1. What could happen to the (inaudible) market depending on this scenario for Elpida.
Too relevant your questions I would say first more the biggest impact on the company. We have limited receivables outstanding and as a matter of fact we have a prepayment of balance Elpida that is higher than the capacity where we currently have, so that’s okay, we have an immersion tool there on loan which is actually ASML property and also with the relevant document there. So we don’t expect any major exposure from the current situation.
From a market point of view probably more interesting, it's our estimate based on some probably data we think that our own bit supplies to the DRAM model Elpida itself is about 10% of what we believe but the Elpida group which of course includes Rexchip could be 15% and 20% in total.
So, that’s quite a significant part of the DRAM market. I just got information from some of our people that were actually creditors meeting that the whole process that is called rehabilitation process takes about five to six month before you have a final go, no go on that plans of there will be a superiority in which of course there is a lot of uncertainty of what’s going to happen.
Clearly the invention of (inaudible) to continue with the operation as to of the call the whole rehabilitation plan that issue is going to be will that be acceptable, will there be a goal, will there be a no go.
And that will have a significant impact I think on the DRAM market, especially since like I said Elpida itself unless the group of this growth associated, I think 15% to 20% the total bit capacity that could have a significant impact if the plan would not go then you would clearly be in a situation where we currently estimate about 10% of all the capacity in the DRAM business, that would have a significant effect on the balance between supply and demand.
And that could be a possibility to a certain extent because last time something Pete like this happened it in (inaudible) and the facts will not useful in Europe.
And in that particular case why weren’t those facts used for DRAM because at that point (inaudible) at that point was not competitive from the cost point of view, so the tool set that was used in those space was not competitive, if you change the shareholder nothing changed on the comparative point of view. So those tools were also let’s say into the logic market we can remember one effect I think it was in Virginia. We sold to TI and again that’s millimeter (ph) what they use for LF (ph) process.
So actually those tools those tools will find its way if it's not used by dealers so.
That’s could be even better.
If that’s the case then of course you would need additional tools at the leading…
Okay now if we move on to the order of (inaudible) for each year, it looks like Q3 was could have been the trust last year with a strong recovery in Q4 and order to BF in Q1. How do you feel, we usually felt superiority in the quarter. How do you feel it's like a one month pay down already.
I think it's nothing changed as compared to the time when we went out with the fourth quarter results. Our customers gave us enough insights and guidance for us to guide us out for about 2.4 billion of sales in the first six months of 2012 about 1.2 billion in the first quarter. So it's evenly spread. That is a reflection of what their visibility is right now. Orders will follow, we believe to support that level and for the second half of the year it's still uncertain because we are just six weeks I have heard of that when we were at the time of the fourth quarter results and nothing has basically changed, in terms of where we could see things going, first half, logic will be store, so family customers will be strong and as which we will be pay a loss in the second half. Kind of first half of 2012 and that will be to our expectation, if we will run our simulation model based on 77% bit growth.
We think that if the industry needs to support 77% bit growth then NAND will be strongly in the second half than in the first half. Then the big question will be DRAM, now DRAM bit growth is currently the expectation is mid-30s in terms of bit growth in overall percentage which would mean that what we currently have on the books plus what we think we will ship in the first six months of this year could satisfy now the demand, so the big question where DRAM is, is what is the expectation of 2013 for next year.
Now in the discussion we have had with our customers most of them simply indicated it actually could be the mid-40s level, driven by mobile application. If that’s the case then by the end of the year there is an upside opportunity for DRAM. Now clearly also with the new situation we just discussed makes DRAM the big swing factor for the second half of the year.
So it's not obvious that this generally would appear like a super strong H1 and then maybe a slightly weaker on a (inaudible). Now you say something which is a bit unusual for 57, you say all those will follow revenue in a way that nobody you get the order and then you ship, this year to be differ by the look of it where you have all the upcoming maybe a bit last minute and you have to ship so quickly.
Yes and the issue is basically where we are today in terms of our market position as the main provider of lithography. The customers are really interested in when we will be able to ship and at most of our customers we have a very good position and basically we are more or less tied to each other. So when we know what customers want we plan the shipment and then we agree with the customers that orders will follow and which they normally do.
For a lot of customers its they have made that choice a leading edge, the lithography as it's more important for them to agree with us when we have the shipment ready and when we have that agreement with our acute customer and you could say we are their only supplier, they are the only customers then once you have that agreement then the orders are they will be following as a kind of we could say administrative process which was of course not the case in 2010 and 2010 we had a situation where we under ship the industry three to four months that was almost a panic at the customer size and at the end of the 2010 everybody remembers that if you follow ASML more than $2 billion of order intake just the reaction of I don’t care you know put the orders in at least we still have place in the line and which was clearly the result of the (inaudible) in the ship in the industry, but in the situation where we are today where we have a very clear relationship with our customers almost on a one to one basis, we are the only suppliers they are the only customer that we have stabled order of what is needed and give us a visibility of six months and the orders will follow.
We have customers that actually give us an order four weeks, three weeks before shipment, not that we have any doubt that we will ship because we know the technology road map of the customer, we know the pass away is going, we collaborate with the tech suppliers, everything is scheduled for that tool to commence, customers check every week, and we are still on schedule and we have missed the process of getting the piece crossed and (inaudible) that takes time and is almost….
So now if you will see when you are doing your one and one missing in rooms sort of there, what are the presentations here in semis focus on the big rate between Intel and the ARM this is even more the case this year, it's going to be the more the case next year. Obviously some (inaudible) then much more than they need last year.
Intel is spending much more than they ever spend in terms of CapEx to size and also in terms of absolute dollars. How sustainable it is, basically it looks to me it looks a bit like what we have seen in the past we have done, what we have seen in the past we met there is this new thing which you know smart phone tablets and stuff like that and how sustainable is this something which is last year on this year maybe in 2013 and then someone we will have to loss and maybe it's going to be inside or the other camp.
Well I think that’s a very difficult question to answer, who can win and who can loss. Actually for us it's not that interesting to be honest. But six months ago I would have shared your worry you know there is a lot of investments in the family business, isn’t this too much. If you look at our historical simulation model where we have put in customer roadmaps, dye sizes assumptions on yields that you know assumptions of the new technology and we just recalculate that back to the number of data tools, they actually saw I think this company six, eight months ago between our let’s say top down simulation and the bottom that the customer roadmap which was a big gap and because actually this could create over capacity but since we saw the tools shipping and the tools being very high in their utilization.
Nothing pointed at the (inaudible) so we started to look deeper into this. We went to our customers and customers, so we spent a lot of time with (inaudible) companies and we discovered three things actually four, one is that the 28 and 32 nanometer run is faster in terms of the number of diseases that we saw with 45. Rejection means you need more 28 nanometer capable tools.
Number two it's time to enter yields because of the difficulties of 28 and 32 nanometer is longer and when the yield is lower you need more capacity to make sure that the end demand can be satisfied. Three is you just mentioned the mobile revolution, as a one of leading as logic to go in there. So what we are seeing is that there is lot more functionality going into 28, 32 nanometer devices which leads to bigger chip, a bigger dye and the dye size is actually a big contributor to the current extra lithography demand just to give you an example of the Apple A4 processor is less than half a size of A5 which is 45 nanometer design, while the A6 goes to 28 which is even bigger. Now if you have bigger dye you need more weight, those are three big drivers and more units.
So it means you need more tools, on top of that we have three major lithography, three major customers for leading lithography in the global foundries (inaudible) and all focusing on this 32 nanometer, 32 to 28 nanometer section of this market.
So those four drivers and the other to have as spare capacity to take orders, so four drivers which is the 28 nanometer design are higher in the number than they were as 45 and 65 for DRAM.
The time to yield is also longer, the dye size is significantly larger and we have a situation where three major foundries at 28 and 32 are battling it out and they need to match the capacity and if you put all those parameters in our simulation model actually it shows that they are buying what is needed in all the capacity.
It is also collaborated by the fact that we follow the utilization of our tools it's max,
So basically there is room for (inaudible), because of (inaudible) of all those new product.
Okay. Now the big question, what was going on, because I am not working at…
Well that will be too much. I think the issue is you don’t have at the time of the quarterly results we have to basically set we have chopped up the issue in several work packages few of the most important ones were going into detail which was the kind of later and the later technology we are using it's called Malta (ph).
We have also an issue with public (inaudible) is a way too more efficiently create this plasma, this and then we have issue with the collector and the mitigation. On Zimmer and on SBI, SBI is a lithography conference which we had two, three weeks ago, they clearly come out with results that were mobile and the repulse actually works, they work together, I mean they were infested, yes, to work together which will have a significant increase in the role power which actually if you try that power you could argue that it's been 30 to 40 ways the speaker run the risk, you really not clearly when you fix it.
But it's a very good step forward, now before we move whether we can get to 60 wafers we need to integrate all the other issues as we are working also which can happen over the next five to six months.
So somewhere in the summer we will have a clear picture of whether the integration works. Now based on that last three scenarios, it all works according to plan and we have a more power in 60 wafers. If that’s the case the (inaudible) will ship next year starting in second half of this year and about want you to per month.
Now 60 wafers per hour by the end of the year, would that mean end of the next year, (inaudible) remain 2014 we would use two units per month which could go to three units per month in 2015.
That will be a very nice gradual ramp up, the second scenario is that we have currently have the role power to 30 to 40 wafers per hour and we actually have a duty of cycle of what we call an ability of the tools of over 70%-75% and also the recondition for scenario number one. So you have 30 to 40 wafers and customers will say when I have a availability of 75% I can start using it in my private line. And for the most critical layers, in DRAM and in logic, we believe customers want to even if it's 40 wafers per hour as low as the availability it gives them a tool that they can use in their planning.
Now that would mean that you will see a slower RAM from 30 to 40 to 60 to 100 which would mean into 2014 you will see one unit per month, that’s the second scenario then the third scenario we will have 30 to 40 wafers, role power that like we have today but the availability or the duty cycle of the source simply wasn’t hit 70% to 75% which means the two can be used for development purposes that is not stable enough to put into a pilot production line which then means customers will probably develop, make sure you get that stable so that we can start, you know using it which could mean that six or 12 months of the for the ramp up in the pilot, in the production that’s where we are.
Those are three scenarios of course been able to show the combination of (inaudible) mobile. Two weeks ago it is very skeptical, so we get confirmation of the track (ph).
Okay and how do you are consistent building and because obviously EUV is completed, it's a big power station on the back of it's because it consumes lot of power. The math are very complicated, if you look like mirrors and it's based on some fancy (inaudible) everything. Is there any (inaudible).
I would basically say there is a biggest bump of that is now the throughput, you need to have the power of source, the biggest in a bottleneck, clear any issues with both resist but that is something where we currently already developing or customers are already developing processes and recipes that you actually deal with that and not one customer has indicated the rest of that either masked or the resistance is going to be a bunker for the EUV introduction, it's really economies of the EUV lithology tools, the throughput that is the biggest issue today and as we go into drive you know EUV ramp it's not a whether we will have EVU but when do we have EUV in full production.
We stop talking about EUV but another subject which could be interesting for this year and of course you remember in did the synthetic buyback, you did not announce anything for this year but it might be the case again. Can you remind us how it worked, because it's a bit complicated compared.
It is a bit complicated and I wish the administration, finance would listen and would do away and withholding tax, because that dividend we are holding tax is basically driving this and in fact such tax will say that a share buyback is an a different distribution and on dividends we need to withhold 15%. Now since we buy those shares on the other market the Texas sector will come to us and say every euro that you spent on a buyback we will charge you with $0.16 of business we are holding tax which would make share buyback a bit too expense.
Now you are exempt for this different weight on the tax, based on different paying history. Now so that means major Dutch companies on any problem because they had had a long history but they only started paying dividends few years ago so we are limited to what we can do. That limitation is based on what we did last year and what we can do this year gives us a maximum of buybacks this year of 430 million euro. On top of that we need to cover our ESOP plans of another 70 million it's always about 0.5 billion we can do in straight buyback.
On top of that we have 200 million of dividend, so we can distribute 700 million, last year we did 900 million of buyback and dividend. So 700 million would be the max.
You will do something on top of that, you can do so called infected share buyback which effectively means that prepayments of share premium. So you would have a repayment of share premium then a corresponding reduction of a number of outstanding shares.
So you would have 10 shares through the euros, we would repay share premium with 30 euros in cash we have nine shares of 30 euros left.
That’s what we did in 2007, but since this is a repayment of capital there is a great protection in the issues, you have a waiting period of eight to 10 weeks where you have to announce that you are going to do this which means you need a general meeting of shareholders that yesterday’s and then you have a waiting period and you have an execution period of two weeks and through all the logistics. The season takes you a quarter, three months to get it executed. Now clearly you don’t want to do this what 10 million, 20 million, 30 million euro so you want to do this in a big bulk that’s what we did in 2007 over a 1 billion.
So you won't have a big number which you then can announce in time at the moment in time that is what we are going to do and then you execute the whole profit. That’s how it works, so it's yes it's relatively stay it all and bit uncommon and it's because like I said we have still economies that which is abolished in most of the countries in European Union, so again I would wish our administration would listen to this.
I am not sure if they are listening to this.
I am sure he isn’t.
Anyway is there any question on the floor
You can maybe just elaborate on your DRAM comment, you said two different things that we between what’s on the book and what you are planning on shipping your supply, you have the bip increase supply for this year and the first half and then you talked about next year, are you sort of saying that the second half for the DRAM side of this drops with the NAND side of it, you talked about the stronger second half or you are still it's still too early to figure out.
Yes of course it's ignoring what just happened with (inaudible). If you ignore what happened with Elpida then if you do that market this year because the big growth is currently averaged about 35% due to customers and to even if you listen to the analyst. So 35% we put the cover with what was shipped last year and what will be shipped in the first half of this year that’s what I am saying.
Now that under the assumption then that will be it, and the assumption that 2013 would mean nothing which is when you looked at 2013 and listen to customers, our customer tell us we were (inaudible) bit growth will be next year based on the end applications which is largely mobile devices, DRAM goes up in all those tablets, DRAM company goes up in smart phones, if you look at their expectation they think it's going to mid-40s, if you mid-40s bit growth in 2013 you need to have that capacity ready in early 2010-13 or late 2012.
So this upside as we have in the DRAM market is really driven by the expectation of 2010-13 because 2010-12 with the current bit growth for (inaudible) basically supported by what we ship in the end of last year and we are going to build ship in the first, six, seven months of this year, that’s what I wanted to say.
Now on top of that of course we have the Elpida situation which we know how that’s going to end up.
Peter, your market share right now is about 80% in terms of value and with EUV you could potentially become a monopoly player in the business and at the same time your customers are consolidating, Intel said yesterday they think of the next, in the foreseeable future it's Dell and Samsung and we are the only companies to be able to invest in NextGen phase and everybody else will have to rely on foundry.
So how do you think about that dynamic, you becoming more powerful that supplier but also your customers are consolidating as well?
Yes I think it's a perfect we saw security balance, we didn’t show that and I think you can joke about situation but what effectively Intel is saying that next generation production technologies require so much capital and investment in R&D and so much risk is only very large companies can afford to do this.
It is also true in the equipment business. If you think about the development of EUV and next is in 450 millimeters that could easily go over 2 billion euros few of our needs because you need to have 450 millimeter on anything, on EUV, on immersion on dry everything, that can only be done by large number. So this economic drive is economical position that we are currently seeing in our (inaudible) is a very unfortunate example of this what is currently happening is that company’s need to be larger, you have the size and volume to pay for these high risk R&D ventures.
And that also means that you almost in a kind of a symbiotic relationship with each other you need each other but to be very honest on top of our customers that is already the case. If you have chosen a known technology as your leading edge, technology and you have basically tuned your process you are (inaudible) process. So that too and that machine and your (inaudible) product that we are delivering extremely difficult to switch. It's already there, and it's our task is to make sure that we can meet the cost road maps of our customers, that’s the big issue. If you can meet the cost road maps, then they will keep buying, if we don’t, the whole thing will go down and it will stop by.
For EUV as you work with the light source vendors, do you have a sense of whether at this point they have a good handle on what are the challenges they face and how they are going to solve all those challenges or do you feel that even at this stage this high risk your schedule could slip further.
I think the issue is not so much with the fundamental question, in which like we said we showed also the quite significant increase in the raw power. Issue is about engineering challenges, to give you an example this source basically operating in a vacuum it is a big vessel, now we found out that the vessel probably needs to be 5 centimeters longer, so you can just of grab it then point those you need to go to a supplier, it makes the vessel 5 centimeters long, takes you two to three months those things happen. So there is not so much of a fundamental issue it is more an engineering challenge of finding out when you put everything together when things work you find things that you have a very short development period and you have somewhat longer development period, that is a big question mark as we cannot answer today and that’s why we said we need to put everything that we currently know that we are working all and put this together into one source that will operate as an integrated source in four to six months and we know.
And meanwhile we will find things, we will find things that can as a sure delay that can have somewhat delay that’s why I thing that we a scenarios, it's the three scenarios that are currently out there. Not so much about whether EUV works but will it work, with a duty cycle of 75% and up that is going to be the big question and that’s engineering, it's a of engineering.
Next question there, can you just remind us when – what nodes for the various customers segments are going to require EUV in order to continue on the roadmap? Either in technological sense or in a just a cost competitive sense?
Yes I think it is, you can say anything sub-20 which for logic needs 20 nanometer node just below the 20 nanometer node for DRAM about 18 nanometers, but some would argue it's already at 22 because they are facing significant issues with double (inaudible) and now it's least issued they can probably continue to 16, 14 so it's couple of nodes out, what they really need is talk about 2015-16 and that moment in time global patent exclusions are there but they will create higher cost and more complexities which is of course something customers would like to avoid when they have to that give them right for productivity.
Okay. Thank you Peter.