2023-08-08 15:22:06 ET
ASML Holding N.V. (ASML)
KeyBanc Technology Leadership Forum Conference Call
August 08, 2023 12:00 PM ET
Company Participants
Skip Miller - Vice President of Investor Relations
Conference Call Participants
Stephen Barger - KeyBanc Capital Markets Inc.
Presentation
Stephen Barger
[Call Start Abruptly] with KeyBanc Capital Markets, I cover Industrial Machinery and Semi Cap Equipment. And I'm here with Skip Miller, who is the Vice President of Investor Relations for ASML. And Skip, thanks for being here again this year.
Skip Miller
Thank you, Steve.
Stephen Barger
Just to start, do you want to give a little background of ASML and maybe get people up to-date on your outlook?
Skip Miller
Sure. So yes, so maybe talk about what we've given for this year in terms of an outlook for 2023. Last, of course, this July quarter, we basically guided that we are – see the revenue moving towards a growth of 30% year-on-year. That was a bit stronger than we had at the start of the year, what we said the prior quarter. Primarily that was driven by additional deep UV revenue. So if you then look at the different segment starting with or I'll say the different businesses that you look at the EUV first, we expect this year to grow roughly 25% year-on-year. That's on shipments of 52 units, which were a bit lower than what we had in prior quarter expectations, which was primarily driven by demand timing due to fab readiness. So that shifted those a bit to the right.
On installed base, this year, we expect that to be similar to last year, so roughly flat year-on-year. A bit different starting into the – end of the year, we thought we'd see some growth there around upgrades, but that's really tied to the timing and recovery. I think the move to the right, if you will, the recovery timing has pushed some of those upgrades a bit to the right. So when you start to see that recovery pick up, that's where you'll likely start seeing more of those upgrades come in. But today we see it a flat growth year-on-year with respect to upgrades.
And then lastly, on deep UV, we had two things really that drove the higher deep UV revenue. One, we had a change in our revenue recognition as it relates to fast shipments. And on fast shipments, not to get too far off to explaining fast shipments, but we ship these machines with a reduced testing protocol at our factory. And that would – prior to this agreement this past quarter, that would result in a revenue that would have to be finalized after we completed all the tests at the customer site.
So that was the shift, usually a quarter in terms of delayed revenue. We are still doing that with EUV, so we still have a delayed revenue of roughly $2.3 billion going out of 2023 into 2024. But for deep UV systems, we now can revenue recognize even doing the fast shipments at the time of shipment from the factory. That was a part of the growth, but we also had higher or more I'll say deep UV units and more – before we said 25% of our deep UV units would be immersion, we now expect more than 25% to be immersion. And that combination grew. The deep UV from what we thought originally started with the 30% year-on-year growth, we're now at 50% year-on-year growth. So you roll that up, it basically translates to a growth that's moving towards 30% with a gross margin. That's a slight improvement over what we had a year-ago.
Stephen Barger
Terrific. Yes. And in terms of the cycle itself, on the last call, you talked about how there's still a lot of semiconductor inventory in the channel that's – wafer starts are declining to help balance that. Can you talk about memory specifically, but then any other parts of the chip supply chain that either positive or negative?
Skip Miller
Yes. So maybe, I guess if you just look first at the higher level in the macro, there's still a lot of uncertainty. The uncertainty that coming from the whole rates, inflation, recession, concerns and even the geopolitical piece. So that's still there. And if you look at the end market segments, there are certain segments that are starting to, you say, bottoming out or reaching normal type levels. But even in those segments, when you start looking back upstream towards the actual fabs, you still see high inventory levels. So our customers, both logic and memory customers are basically trying to moderate wafer output to bring those inventory levels down. And they do that by lowering utilization on their litho systems as well as other, but we monitor, obviously, see the litho utilization levels, and you can see that lower coming down – have been coming down at the start of the year.
And on memory, it's still coming down. On advanced logic, we saw over the past few months what looked like a flattening, albeit at a much lower level. And so we'll just have to monitor that. I think in a more mature space, you see a more stable and a higher level of inventory, sorry, higher level of utilization. And so that's kind of the dynamic right now. We'll continue to monitor the utilization levels, see that they start stabilizing and then eventually return and start going the other direction, at the same time expecting the utilization levels to come down.
So I think in general, although there's different market segments reaching that, there's still high inventory, customers are still moderating their level – moderating the wafer output by lowering utilization levels. And we'll continue to monitor that. And as they expect to see a recovery in the second half, I think the real question is what will the slope or the shape of that recovery look like as we see that to start to unfold. Start of that has probably shifted a bit more towards the end of the year as opposed to the middle. But that's what we'll have to be monitoring, and we'll see how that unfolds as it relates to demand into next year.
Stephen Barger
You've been in the industry a long time. Maybe you can compare and contrast this cycle to others that you've seen to maybe give a clue to how you see it playing out?
Skip Miller
Yes. So I guess, yes – it's a reminder, I've been in this industry a little while.
Stephen Barger
You look great though.
Skip Miller
So yes, I guess you look back, it clearly this, you call it a classic semiconductor cycle in the sense that you have an overbuild and now you're working through that inventory and time. And I think if you go back and look at 2018, 2019, that was more just a memory cycle. Maybe you go back and look at the financial crisis around the 2008, you had a different event there, so you almost have to go back as far to like 2000, 2001, where you had both the Dot Com, but also Y2K, which created this massive buildup, and then it took some time to bring that inventory back down, or you could say it differently or the cycle to recover.
What's different about now? I think if you look at comparative back then, you only had really one-end market application which were PCs. Today, you have many different applications. So that I think will help in terms of duration to recover. You also have, I think customers today after some consolidation over the past 20 years, customers are a bit more disciplined on how they – with the moderation right now, wafer output controlling, inventory is a bit more, you could say disciplined, I guess, or effectively. That's another piece.
And I think the last piece is probably there's just tech sovereignty we call it, which are the different countries desire to onshore, bring the manufacturing capability back to their respective regions. That's creating, you could say, a strategic build that's ongoing as well. That's fueling obviously equipment demand that will help in this recovery as well. So we'll just have to monitor, see how he has. But I think there's definitely maybe a similarity back in that timeframe, 20 years ago, but there's a different set of parameters that exist today that should obviously shorten that, but we'll have to see how things unfold.
Stephen Barger
Yes. I mean, obviously there's still a lot of uncertainty. Peter, on the last call, the CEO didn't want to talk about 2024 much, but he did talk about wanting to maintain capacity for 2025. Can you just expand on that and just in the context of what your major customers are saying about CapEx maybe coming down a little bit this year or in 2024?
Skip Miller
Yes. So maybe just to remind everyone, and in 2022 at our Investor Day, we gave an update on scenarios for 2025 and 2030. And as part of those scenarios in terms of where we see, end market-driven demand going, depending on whether it's a high or low market. We needed to significantly increase both EUV and deep UV capacity. So we started making these investments last year and into this year and continuing through next year for 2025 and beyond to get to 90 EUV machines and 600 deep UV machines of capacity by the 2025, 2026 timeframe. In parallel, obviously we're also investing in high-NA capacity, but it's a bit later, its 20 units by a 2027, 2028 timeframe. So we need to go make these investments now because a lot of these, not only at ASML specifically, but also in our supply chain and our supply chain partners, you have to also build buildings in many cases, which take quite some time. I think multi-year to get the permits, to get the buildings in place to eventually put the equipment and then people to ramp that capacity.
So that'll take some time, and those are long-term investments that we're going to make and get put ourselves in position to be ready to have the capacity to meet the demand for these scenarios that we talked about for not only 2025, but the second half of the decade. And so yes, I think, look at today, I know there's still some uncertainty as it relates to 2024, and that's what Peter commented a bit on our call about. There's pros and cons or there's opportunities for growth next year if all the demand comes through. However, there's also uncertainty, so a balance there in terms of customer caution and how things will unfold. But we need to make those investments, continue those investments as we see the long-term demand with all the secular growth drivers starting to unfold, we need to have the capacity in place and not restrict the industry to continue to grow.
Stephen Barger
Makes sense. Shifting gears a bit, major theme of this conference is AI. Can you talk about how ASML is using AI internally to operate the business more efficiently? And then maybe what you think it means for the industry near-term, long-term?
Skip Miller
Yes. So AI, I think, yes, if you look at what – how we specifically use AI, there's many different applications. I think one of the big ones obviously is the computational lithography. So it's actually bringing designs into the final images on the wafer and that requires a lot of computational power and a lot of AI to go with that. Another area or areas I should say, is about we generate a lot of data, not only through metrology systems, but also on our machines themselves. And actually utilizing that data in a predictive manner is another area where we utilize a lot of the AI capability. And I think you'll only see that continue to expand in time.
In terms of what AI means for semiconductor demand, and for us, litho demand. I think if you listen to our customers some what they've been recently commenting on is they are seeing some incremental demand due to the whole generative AI move and all the activities going on there. It's still in the early innings, so they commented, I think [indiscernible] said something around 6% of their revenue, however, they're seeing a 50% CAGR over the next five years, so healthy growth.
And I think if you look out in what we communicated in our 2025 and our 2030 scenarios, clearly AI was one of the mega trends that we talked about. So you're starting to see that unfold near-term. It's in the early innings, I would say today. But longer term we see it moving not only in the server space, but on the edge and ultimately end-to-end devices. So I think there's a great opportunity for AI going forward, and it's something we've been talking about for quite some time, and it's nice to see it continue to really develop and start to gain traction.
Stephen Barger
Sure. And I should have said this earlier, but if anybody has a question in the room, just let me know and we'll work that in. I'll talk about China, or I'll ask about China. It seems like that's been a bright spot. Can you update us what you're seeing there in this year or how you think about coming quarters, especially in the context of how you see the regulatory environment changing?
Skip Miller
Yes. So maybe first in terms of baselining, where we are in terms of China, what it means for revenue for ASML. Historically we've been around 15% in terms of system revenue going to the China region. We commented on in April and again, reiterated in July that if we look at our backlog, it's over 20% now. Our backlog is from the China region. And so we expect this year that our revenue will be more aligned, more reflecting of – aligned to what we see in our backlog. So we expect our revenue this year to be – over 20% of our revenue to the China region. What's driving that? I think if you look at, I think there's the demand, and we call it the fill rate meaning there's a certain demand and we have a certain supply. We've been unable to meet the demand.
And over, in fact, over the past couple of years, that fill rate has been less than 50%. So there's been a substantial amount of demand that we've been unable to supply. That now due to some of the shifts in memory, in terms of demand timing even some of the advanced logic shifting a bit to the right that's provided the opportunity for us to fill more of that demand to the China region, so that's kind of the driver, if you will, for the higher revenue this year.
I think if you – to your question on export controls, I think what came out in the communication from the Dutch government, the official communication at the end of June was largely aligned to what we communicated earlier in April and early in the year on our website. And that was that the export controls are really targeted to advanced applications. They will go into effect starting September of this year. And we will require an export license and have to apply for export license for our most advanced immersion machines, which for us means the TWINSCAN NXT:2000 and above, if you will. And so that, again, goes into effect September, apply for the license.
And I think as a reminder, we've already restricted EUV for a number of years now. And predominantly the focus of the business in China is around mature and mid-critical applications. So that's still is supported by 1980s immersion machines, and below, as well as a significant number of dry machines to address that mature and mid critical technology.
Stephen Barger
Okay. We have a question on the floor.
Question-and-Answer Session
Q - Unidentified Analyst
[Question Inaudible]
Skip Miller
Yes. So on the tech sovereignty piece, that’s what we call it. So that's we're saying basically, and what you see now going on in the U.S., you see it in Europe, in Japan, we basically quantified it in our Investor Day material, and we said that it would create roughly a 10% additional, we call it inefficiency, if you will, in terms of our – the capacity that's out there between now and 2030. So that's how we quantified it. Is that sufficient in terms clearly by definition, there's going to be some inefficiency, that's a strategic desire for each of these governments to put the fab on their soil. By doing that, you will create some inefficiency. I think these big customers that are doing this will do a good job of managing that to be as efficiently as possible, but there will be some there. So I think our quantification is 10% between now and 2030. And is that sufficient definition of that, or is it more or less? I think we'll have to see how things unfold in time.
Stephen Barger
I want to go back to supply chain. I think it's gotten better as the cycle has adjusted to some degree. Where are you seeing sticking points still? And if we do see a slope of recovery that's faster than you had expected, do you think they'll be able to keep up or do we risk getting back right into what we had, which was real constraints?
Skip Miller
Yes. So first on the supply chain, I think if you look at what occurred in 2022, there were a lot of, you could say, surprises in the supply chain in areas that you didn't design in terms of the bottleneck. Our bottleneck, by the way, design is around the optics. That's the most complex, the highest cost item in our supply chain. So that's where you design the bottleneck. And our end objective was where to get the other suppliers in the shadow of our optic supplier being Carl Zeiss. I think, like I said, in 2022, we are having all kinds of different supply chain challenges from things like valves and hoses and all kinds of things. You didn't want a $50 part or euro part limiting the shipment of a multi-million machine, but that was the reality of what we're facing.
So I think that if you look at, we came into 2023, we said by the maybe the middle of the year, we'd be in a position to be in this more predictable, where you had these other suppliers in the shadow of the optic supplier. And I think we're achieving that. The predictability is there. So I think that part of it is now you can see a bit under control. And now it's more about the ramp. And so to your last part of your question, where are you headed? And will we get into a situation where we're supply constrained? It kind of depends on how things unfold. Now what we're doing clearly is we're – this year we talked about having the capacity of more than 375 deep UV machines and more than – or 60 EUV machines.
And I think if you look now out to 2025, 2026 timeframe, or 90 EUV and 600 deep UV. So we still have some, you could say, additional capacity expansion to go, and that's actively going on. If you go over to Veldhoven, and you'll see a lot of cranes and a lot of buildings. But if you also go to our suppliers, you'll see that as well. And some of our individual sites like in Wilton, Connecticut, there are a lot of activity going on there to put this capacity in place. So we are ramping, it doesn't come in a linear fashion, it comes in steps. Will we have sufficient supply next year, 2025? Obviously that's our plan and we're not letting the – taking our foot off the throttle there. But it's going to also depend on where the demand ends up in the end.
Stephen Barger
Sure.
Unidentified Analyst
[Question Inaudible]
Skip Miller
Well, I think, we don't have – so in our supply chain right now aside the bottlenecks, we talk about being the optic supplier that we will, we manage accordingly in this ramp that we need to undertake with all the suppliers by the way. In terms of other items, you talk about packaging, I think you heard from our big customers saying they're in the middle of a ramp and they're limited by the backend, which is interesting. I think you see that as part of these – some of these new packaging strategies that are going on.
In terms of rare earth and others, we at least today don't hear something from our customers saying that that's limiting their ability to ramp doesn't mean there's not something out there unfolding. But we don't have anything today that at least is coming to us that's saying, hey, we're having this massive issue other than the fab readiness piece, which I mentioned, earlier which is more just a skills thing that they need to get the skills ramped up.
I think that's partially has to do with the fact that we haven't really been at this level of fab bills for quite some time. We did way back when – back when I started the industry, there was a lot of this activity going on, but that's kind of moved a lot of that to Asia, so bringing that back is going to take some time to bring those skills up to speed, but I think that's all something our customers view as manageable.
Stephen Barger
I'll ask a similar question on the skilled labor front, whether it's in Europe where you're doing a lot of manufacturing or what you're seeing from your customer conversations, is they build fabs in the U.S. or just build the infrastructure required to support your growth?
Skip Miller
Well, I think aside from the fab, fab readiness issues that we just talked about. I think in general there's a – obviously a hiring talent competition that's going on – yes, maybe that'll soften over time. But it's still quite significant today. But I don't think at least we're not – even though it's an area of focus and an area of challenge, I don't think at least we're hearing from customers saying that that's limiting their ability to either execute on their technology roadmap or execute on their capacity ramps or other areas of development. But I'm sure it's something that obviously that just like ASML that they're all focused on and try to get the best talent and bring them on in an efficient way. We brought a significant number of people on last year, and you want to make sure they land in an efficient way and we can contribute as quickly as possible.
Stephen Barger
Right. We just have a couple minutes left. Can you talk about how you see litho spend as a percentage of WFE or whatever metric you want to talk about expanding over the next few years? I guess, do you have a broader opinion about WFE intensity in general?
Skip Miller
Yes. So we just maybe a level set on WFE, so I know it's a nice metric and a lot of –used way to roll up all the numbers. We don't specifically talk about that either when we're guiding or whether we're talking about future growth in our 2025 or 2030 Investor Day scenarios. Why? Well, because we don't know all the non-litho side in great detail. So we can provide, and what we do provide are what we have in terms of our scenarios.
And if you look at our 2025 scenarios, you're talking between $30 billion and $40 billion in revenue. So how that will map into the WFE, I guess it comes down to how the other numbers will land. I think we see it as a nice growth. Will the others grow at the same pace and therefore as a percentage of WFE will we increase or decrease? I think our focus is more on how what we can control and what we see in terms of our growth. And I think that continues out to 2030 where we're somewhere between $44 billion and $60 billion. And so that's our growth trajectory and then you just have to map it in how the others see the rest of the industry. And that's something obviously we're not an expert in non-litho. So others will do that and determine what we grow as a percentage or what will it look like? But we think there's healthy growth opportunity out there.
Stephen Barger
But you think litho intensity as a percentage will increase as node transitions occur?
Skip Miller
We think the litho spend per node, we actually communicated in our Investor Day for Memory and Logic. We actually communicated to build a fab, what is the litho spend per wafer and that's definitely growing. In case of Memory put out there around 20% node on node. For Logic, we had somewhere around 30% node on node. So there's definitely significant opportunity as you get a richer mix of critical layers, more EUV adoption. That's the key drivers behind that growth.
Stephen Barger
Perfect.
Unidentified Analyst
Quick one.
Stephen Barger
Yes.
Unidentified Analyst
Philosophy and dividends and share repos, a lot of free cash flow, what do you guys think about that?
Skip Miller
Yes. So our capital allocation is, we obviously generate significant cash the free cash flow. We have some pressure this year because of down payments primarily that we mentioned on the past two quarters. But in terms of the cash that we generate, first we'll invest in our business. Second, we will grow our dividend. And lastly, in a remainder, excess cash will go to investors in the form of share buybacks. So that's our capital allocation policy unchanged. But the only item we missed – mentioned this year is some pressure on free cash flow and running at a bit higher levels of cash just to be prudent in this environment.
Stephen Barger
Thanks very much, Skip.
Skip Miller
Thanks, Steve. Thank you all.
For further details see:
ASML Holding N.V. (ASML) Presents at KeyBanc Technology Leadership Forum Conference (Transcript)