Great. Well, hi, everyone. I'm Alex Duval. I head up the Europe Tech Hardware team based in London for Goldman. Thank you so much for joining, and delighted to be joined by Skip Miller, head of Investor Relations for ASML. Before we kick off, I'd just like to state the conversation is not intended for the media and is off the record. And with that said, perhaps we can kick off with a high-level view on the latest trends you're seeing across logic, memory, EUV. What growth do you expect for each of them this year, next year, and then we can get into some more specifics.
All right. Thanks, Alex. Yeah, so for this year, first off, you know, we did guide on the specific market segments, logic and memory, because of currently being supply limited, so we really talked more about our businesses and what we can supply. I think clearly, a logic, you know, dominated year. But when you look at the businesses, and you first start with EUV, we talked about, you know, growth coming into the year. We said looking at this full year, we expected around 40%. We now expect that to be around 25% year-on-year growth. The change there is that we had some systems this year that we won't ship due to fab readiness as being the primary driver behind that.
And so that, you know, lowered some of those that will, you know, ship obviously next year when the fabs become ready. So that was one, so that's EUV. On Deep UV, we actually adjusted our outlook there for the year of a growth. We talked in April about a growth of 30% year-on-year, and we're now talking about a growth 50% year-on-year. There's really 2 drivers on that front. One, just, you know, more systems and a higher, you know, percentage of immersion. We're now seeing over 25% of those machines, over, I should say, over 375 units, first off, but of that, over 25% of those will be immersion. So that adds some of the revenue, but the other piece of it being around rev rec.
What we had changed, communicated in July, is that we will now, for machines where we do fast shipment, and maybe we can talk about fast shipment later on in, one of the follow-on questions here.
Great.
But we're doing fast shipments for DUV and EUV machines. And on DUV machines, we've reached agreement with our customers that, on the reduced set, the reduced set of tests that we actually run at our factory, that that will be the new standard. So now we can revenue recognize shipment from our factory. And that basically meant we saw roughly another EUR 700 million of revenue this year that would not... You know, we originally were planning it to be delayed out of 2023 into 2024. That'll now be revenue recognized in 2023. So that was DUV. So you had a combination of, again, the change in fast shipment revenue recognition, but also more systems.
And then the last one was the installed base, which initially in the year, we thought we'd see, you know, some growth there, you know, around 5%. This, you know, looking at now July, because of kind of the timing of the recovery, we look at it as maybe it'll be similar to the revenue that we had, last year. So we said, you know, flattish relative to 2022 on the installed base. That's primarily has to do with the upgrade side of it, meaning, the customers will wait and do upgrades when they start seeing some recovery there. So when you roll all that up, you know, the higher, obviously, DUV revenue net, we now say instead of growing 25% year-on-year for the total, sales, we now see that moving towards 30%.
That's kind of the summary where we are with respect to the, you know, different businesses and the outlook we have for this year.
Super helpful, Skip. I think if you look at the recent earnings season, definitely highlighted robust demand for leading-edge chips related to AI, generative AI, specifically, and NVIDIA talking about, a guidance that was ahead of expectations, and obviously in recent days and weeks, other bullish statements from other companies as well. I wonder if you could share with us what you see for 2023 and 2024 as far as ASML is concerned, related to AI. How are you seeing the picture?
Yeah. So AI clearly gets, you know, a lot of buzz in the media, but also you hear from our customers and our customers' customers about the opportunity that's provided and what they're actually seeing in terms of increasing demand. I think, you know, from a respect of AI, clearly long term, we've talked about that in our, you know, some of our key secular growth drivers, AI being one of them. It's been on our Investor Day material now for, you know, a number of years. And I think from a question you're specifically asking more near term, I think we're kind of in the early innings, so although the growth rates are quite high, you know, communicated by some of our customers, you're, you know, working still at lower numbers.
So again, it will become more material as you move further in time. So we see that as a great opportunity. And I think it's also maybe, you know, something we'll have to see how the adoption unfolds in terms of the, not only just in the cloud, but also at the edge and eventually in the end devices.
Super helpful. And do you see it sort of boosting orders, for example, next year, number 1? And then I guess number 2, do you think about 2025, to what degree could that provide a bit of an upside boost there?
Yeah, so there's clear opportunity there, but I think the question becomes: When does it actually, you know, translate into, you know, a meaningful number that we see back in demand? Is it gonna be nearer or longer term? I think we're quite comfortable in the long term, and we'll have to see, being in the early stages, how significant it really means in the shorter term. But clearly, again, that's gonna be, over time, a great opportunity.... for growth going forward with this, just another leg, if you will, on the whole, opportunity for growth here.
Makes total sense. Given you gave a sort of lower revenue target for this year on EUV, and I think you referenced that in your introduction, maybe could you update us on what customers are saying in terms of technology roadmaps and desire to have EUV tools for next year? You know, clearly, ASML's talked, talked about a push out in demand timing for EUV. So is that a one-off push from 2023 into 2024, or could there be some of the orders that actually see a push into 2025?
Yeah. Yeah, so there are multiple points there. I think first off, in terms of technology demand, customers continue to push innovation and therefore, technology needs. And you see that even, you know, some questions around memory demand and memory orders, and I think you're still seeing technology buys there. And that same is true in logic. So that'll continue, innovation will continue independent of where you are in the cycle. So that's one. I think the second piece, which is around the, you know, more the demand capacity-driven type demand. When you look at that, it's really a combination of inventory levels and how that is gonna come down in time.
And today, with inventory being higher, you know, looking at, you could say, upstream in the channel, they are moderating wafer output today and do that, you lower utilization. So we see lower tool utilization levels, and in time, as that inventory comes down, utilization levels would start coming up and eventually trigger new demand. So that's the dynamic that we're looking at, inventory levels and utilization levels. The specific item for 2023 and a shift of units had to do more with the fab readiness, which I mentioned earlier. And hopefully, as our customers work through bringing the skills level up, in some of the trades in terms of how they build out these fabs, that that will in turn, you know, address itself or resolve itself over the course of 2024.
So, I think I covered most of the three you asked. I don't know if there was another item on there or not.
Well, I guess you referenced the fast shipment recognition-
Mm-hmm.
And we know that sort of DUV is getting recognized, but there's this EUR 2.3 billion of EUV revenue. So, what's the latest there on, you know, whether that just goes into 2024 and stays there or-
Yeah.
Push on, how should we think about that?
Yeah. So maybe a backup, just a test for those that aren't fully familiar with this, whole fast shipment revenue recognition dynamic. Basically, when we came into the year, we said, "Okay, we're gonna still do fast shipments on some DUV machines," think immersion machines and EUV machines. And the objective there is to reduce the cycle time so that we can ship more machines out of our factory, but also more importantly, with the customers, they get the machines quicker and they can qualify and get them ramped in production sooner. So that was the origin of fast ships. Now, doing fast ships meant that because we reduced the number of tests that are run in the factory, that we had to wait to complete the whole 100% of the tests at the customer site.
So you had this delayed revenue, you know, of roughly a quarter or so. In July, and, and so—sorry, the start of this year, we said that due to fast shipments, both DUV and EUV, that we expected to delay roughly EUR 3 billion out of 2023 into 2024. And then in July, we said we were able to align with our customers such that they were willing to run the reduced set and sign off after that, which in turn meant that we could revenue recognize on shipment on a fast shipped machine, DUV machine, which meant around EUR 0.7 billion. So instead of EUR 3 billion going out of 2023 into 2024, that's now EUR 2.3 billion. So that was the dynamic that occurred in the first half of the year.
So going forward, any DUV machines, even if they're fast shipped, will be revenue recognized when they leave our factory. But EUV, if we do a fast shipment, we still have to wait until those machines are complete, their full testing at the customer site, if you're gonna do a fast shipment. Which means today, we still expect EUR 2.3 billion revenue ... EUR 2.3 billion euros of revenue will be delayed out of 2023 into 2024. Now, next year, the question becomes: What will fast shipments look like going forward? And I think there's two dynamics that, you have to take into account. One is that when you move to a new model machine, in this case, we're going from a 3600 D machine to a 3800 E machine next year.
When you go to that new machine, you end up wanting to run the full test. You don't want to reduce, run reduced tests. We want to see the full sequence for those machines. So, we'll likely see the 3800E machines that we ship next year will not receive fast shipment. So that's one thing to keep in mind. And then the second piece is we continue to ramp our capacity for EUV. You know, this year we were talking about capacity around 60 machines, and by 2025, 2026 timeframe, we plan to be around 90 machines.
So in that ramp profile, if we can get the capacity above the demand in such a way where you no longer there's not as much value in the fast shipment side, and our customers aren't as pulling as hard to get the machine on the fast shipment because we can ship what we need there, then I think you would argue you wouldn't do fast shipments in those two scenarios. So if that's if both of those were to come through, that would obviously reduce the number of fast shipments that we would do in 2024, and then in turn, could adjust the number of delayed revenue out of 2023 into 2024, but that's it or out of 2024 into 2025.
As that's the dynamic at play today, and so we'll have to see how between those two, I think these as we ship more of those in 2024, you won't see fast shipments on that front. And then the next question just becomes a supply-demand. Where are we on that front, and what makes sense to do fast shipments or not?
Understood. And I guess, Skip, if we stay on this topic about sort of EUV and the unit dynamics, I guess, you talked about how there's a shift from 2023 to 2024. Specifically, you have some fab readiness issues, and that's the majority of it. There's probably a bit of macro in there as well, and perhaps a minority of that. How should we think about a sort of shift from 2024 to 2025? Could this theoretically boost 2025? How should we be thinking about that?
Well, yeah, so I think the way to think about the second half of 2024 in terms of, you know, demand, you have to keep in mind that machines that we're shipping in the second half, in particular EUV, it takes some time before you install, qualify, and have those machines producing wafers. So those machines in the second half of 2024 are really about wafer demand in 2025. So I think that's a piece you just keep in mind. In terms of how the second half will shape up, 2024, you know, orders will continue in this quarter and the next. And yeah, we'll have to see how that in turn means in the, you know, the demand dynamics there.
In terms of if your question is specific around the fab readiness piece, as I said, hopefully we'll have that resolved over the course of 2024, so that doesn't play into this. I think, you know, the rest will have to just see how the recovery in terms of the shape, the slope, will occur in 2024. That depends on how many units are needed.
Makes total sense. On 2024, you sort of talked about a clear opportunity for growth.
Mm-hmm.
I think that was the phrase that was used. Obviously, that's against very tough comps. Also, EUR 700 million is kind of being pulled in. So, what would need to happen for that opportunity for growth to kind of materialize? Is this kind of the orders you're referencing, and the fact customers need to get confidence on the macro picture for 2025?
Yeah. Yeah, so I think what we said in July, you know, again, was trying to provide some balance between the fact that if you look at the, you know, current, forecast and the demand discussions, and what we're talking about with our customers, and you take that in addition to, the strong backlog we have of EUR 38 billion, that provides a clear opportunity for growth next year. Now, having said that, we also tried to balance somewhat, though you have to keep in mind that there are still a lot of uncertainty in the macro there. And if you, you know, look at that from a perspective of, yes, there's a recession that's continuing to loom, out there, that people are wondering when, that could happen and what impact that could have on demand.
You have the geopolitical dynamic that's out there. And so those two items, I think, are, you know, what's the major driver around this macro uncertainty that's creating some, you could say, caution with our customers in terms of how to, you know, see through the recovery in terms of timing and the slope and shape of that. I think our customers are still of the opinion that the recovery will happen in the second half of the year, but it's probably moving closer to 2024, as opposed to the middle of the year. And then the big debate is on the shape and the slope. So I think that's the piece that we have to, you know, to see how that will unfold. So a balance.
There's clear opportunity there with all of the forecasts and discussions we're having, and the growth opportunities are out there, against this uncertain piece in the macro, and we'll have to just see how these next two quarters play out and how our customers see the inventory levels, and in turn, what this means for the utilization levels.
Makes total sense. You're talking about this clear opportunity for growth next year, but then, of course, trying to sort of inject some balance into the conversation-
Yep
... which is totally understandable in the current environment. Then, how should we think about 2025? You know, there's obviously structural drivers you laid out in a lot of detail at your CMD. How confident are you there? What are the kind of different moving parts to think about?
Yeah. So, I think, so the 2025 scenarios that we provided last year during our Investor Day, it provided an opportunity between EUR 30 billion and EUR 40 billion in terms of revenue for 2025. And again, that depends on a high-low market. So, I, you know, our view today is we're still quite confident in our 2025 scenarios that we provided. I think if you look at the structural growth drivers that are out there, we talked a little bit about AI earlier, but you see all the others kicking in. We talked about the electrification. You see what's going on in the EV space. That's clearly an opportunity there. The Industrial IoT continues to unfold, and I think a lot of the secular growth drivers that we talked about are very much, you know, materializing.
And so, 2025, again, we have no change to what we said in terms of our opportunity there, and even beyond that, looking at, you know, 2030. And I think the other piece is still, you know, unfolding as this whole tech sovereignty play. I think you're seeing that, you know, materializing as well, which we spoke about in our Investor Day as well. That's also a, a piece of it, where you're seeing these customers now building fabs in the different areas in regions that are trying to reshore or onshore some semiconductor manufacturing. So that's also at play in addition to the whole, you know, technology transitions and then some question marks on the capacity timing.
Makes total sense. I guess you talked about building capacity next year. That's to make tools, but in order for you to have demand for those tools and ship those, there need to be the orders... Could you give a bit of an update on what we should be thinking about in terms of orders in the coming quarters? How should we be thinking about that?
Yeah, so we don't, as Alex says, we don't guide for orders. So, you know, it's more of, how we see second half of 2024. And clearly, we expect there'll be, you know, orders will continue in the second half of this year. That will be towards the 2024, second half of 2024, machine demand shipments. Again, as a reminder, that the machines in that second half of the year of 2024 are more about, you know, 2025 capacity. I think most view, or 2025 wafer capacity. So I think most view, by 2025, there's clearly we'll be back end or we need to have a lot of, you know, wafer output, in that time frame.
So I give some confidence that second half of 2024, we'll need more machines as well, in which case you should see orders in the coming quarters. But customers are tight on cash, so obviously, you know, they're gonna push us on the orders a bit. But that's the dynamic at play that's right now.
Super helpful. I guess you touched on sort of trailing edge demand and sort of all these sort of mature applications. Maybe we can talk about that, maybe leave China DUV to one side-
Yeah.
Because I'd love to ask you about that separately. But if we think about sort of trailing edge demand, it's obviously been strong in recent quarters. You know, maybe give us a sense of kind of what's driving that. How do you think about supply and demand, and how sustainable that could be?
So, yeah. So I think if you look at the auto industrial piece of it, that's, you've seen that being quite strong and resilient through these, you know, past few years, trying to catch up on the supply-demand dynamic with respect to the end markets. And we see that in terms of, you know, demand for our machines. Now, the challenge we faced over the past couple of years is our ability to ship these machines. We just, you know, the supply/demand we had at one point, I think our demand was somewhere, you know, 30%-50% above what we could supply. So as we said, we're ramping our supply capability there, but that demand remained.
And so I think what you see now is because some of the shift in the, demand timing for our memory customers and some of the, as well as some of the logic customers to the right in terms of demand timing, that's provided the opportunity for us to meet or ship more of these more mature technologies, to customers that are both, you know, outside of China as well as inside of China, that are supporting these auto, industrial, and more mature applications. And we see the... You know, if you look at all these things we've talked about in terms of secular growth drivers around, AI and, and, electrification and all these applications or these, secular growth drivers, there's clearly not only an advanced demand, advanced application demand, but there's also mature.
And so, you need all the infrastructure around that, and so we see this sustainable going forward. It's part of the reason why we talked about a need to grow our DUV capacity, from, you know, around 375 units today to around 600 by the 2025–2026-time frame. That's a, you know, significant growth. And we even quantified how many wafers a year that you need to add, or we think we need to add, the industry needs to add, and therefore, the units that we need to produce. So, we see it sustainable going forward. It's gonna be a key part of it.
We're able to meet more of the demand this year with some of the shift in demand timing of others, but still not fully meeting all the demand that's out there today for the more mature applications.
Got it. And I guess on the sort of China side of this, that's another part of the equation.
Yeah.
You know, you've clearly had a lot of benefit coming from China. I guess investors sort of typically have two questions. Firstly, how sustainable is that?
Yeah.
In the sense that is this all sort of a pull forward, that then it's gonna end? Number 1, and then number 2, you know, could there be a further regulatory change, that could impact this, particularly when you see news flow about a sort of major Chinese smartphone vendor being able to use technology on 7 nm? So obviously a bit of a sort of moving situation, but curious on your latest view of that.
Yeah. So, China, you know, first off, from a business perspective, I think if we look at this year, you know, we have a backlog that's somewhere 20% or more of our backlog is targeted for Chinese customers. And I think what we've said, both in April and July, is that our expectation is for this year, is that we will end up with a revenue that is somewhere aligned to our backlog. So, think of a revenue of, you know, around 20% or more. And that compares to around 15% that we've had the past few years in terms of the system revenue going to the China region.
In terms of the driver, why it's higher this year, it primarily has to do with what we call the fill rate, or a lack of ability to fill all the demand in prior years. So in the past couple of years, we've been only able to fill, you know, less than 50% of the demand of what they really wanted. And as I mentioned earlier, with the shift to the right, with some of the demand timing of some of the other customers, that's provided the opportunity for us to ship more. So that's the primary reason for a higher percentage of our revenue this year going to China. In terms of sustainability, our view is that it, you know, China will continue to be sustainable as they work to build out the capability.
They want to put the capability in place. Obviously, it'll be a focus on mature technology, which is what they can in fact produce. And it really is in support of the different industrial activities going on in China, the growth there, and some of the items that we pointed out, electrification being one of them, energy transition being another one. But I think we even talked about, you know, some of the real impact that the example, like an EV production has in terms of the wafer demand there. Every EV vehicle needs a third of a wafer, and therefore, it drives a lot of fabs. In China, they're working on building those fabs.
So, we see that piece of it being a sustainable story and not something just a one-time pull forward, and it's over, and we'll continue to do that going forward. And again, I think we can talk a bit more about the export controls, but it's gonna be, you know, the focus in China is gonna be on, you know, the more mature technology, the 20 nm and above.
Makes sense. And then obviously, in terms of the sort of regulation itself, maybe you could just give us a recap on sort of where we are at the moment.
Yeah.
And then, you know, you had obviously questions about whether it could get more stringent. I think your CEO said actually that wasn't his expectation. So what's the latest thinking?
Yeah. Yeah, so the export controls, well, and the regulations, you'll say the official regulations that were communicated by the Dutch government last month, sorry, at the end of June, were largely aligned to what we communicated in prior or largely aligned to our expectations as communicated in the, you know, some of the prior quarters. And that was that it would, you know, go into effect September 1, and it would be targeted for the most advanced immersion machines. So you require a license, so you have to apply for a license for NXT:2000 and above, going to China, starting September 1.
We expect that we will, again, beyond 2024, that those export licenses, there will, it'll be more about shipping 1980s versus 2000 s, and that it will not have a real material impact on our 2023 financial outlook and our longer term, our 2025 and beyond scenarios. Again, keep in mind that our 2025 and beyond scenarios are end-market driven, not independent of where they're built regionally. They can build anywhere on the globe, but it's about end-market demand there. Largely as expected, it will require export licenses going forward for NXT:2000s and above. Again, as a reminder, we're restricted already on shipping EUV machines today. In terms of speculation on what might happen going forward, I won't speculate.
Yeah, I think there's been different comments made on this front, but again, I think there's reason to believe that 1980s needed in support of you know 28 nm above for global supply, but I won't get into speculations on future government decisions.
Totally understand. I think we've got about five minutes left, so we should have time for a couple of questions from the audience. I think we have someone walking around at the back with a microphone. So if someone would like to put their hand up if they have a question.
Thanks for taking my question. On the topic of recent headlines around Chinese smartphone makers and some of the progression that they've made there, just a question on whether that has maybe resulted in some question marks around roadmap progression without the use of EUV and instead DUV multipatterning. If you could maybe provide a reminder on the comparison between DUV multipatterning and EUV as it relates to cost, performance, et cetera, when we talk about some of the more advanced nodes, like 7 nm and onwards. Thanks.
Yeah. So, yeah, so first off, you know, in terms of the production of, you could say, more advanced nodes using multi-patterning, I think our customers in the past, before EUV, because of some of the delays in EUV, had manufactured 7 nm and beyond using a multi-patterning technique. And so, the challenge with multi-patterning, and, you say, the value therefore with EUV, is that multi-patterning increases the process complexity. Meaning it takes multiple layers, multiple exposures to produce one layer, as opposed to one exposure with EUV. So that's the value, obviously, going to EUV. Now, producing it with a multi-patterning, is it possible? Yes. I think our customers have proven you can produce 7 nm, for example, using multi-patterning.
But by doing and utilizing multi-patterning, the process complexity translates into cost, cycle time, and yield challenges. Meaning, more steps, obviously higher cost. On the cycle time, you have more steps, obviously more cycle time. And then by process complexity, by trying to align all these different pattern sequences to make the one layer, you end up with, you know, edge-to-edge issues and so on and so forth, that create some yield challenges. So I think those are the three big challenges. Is it possible? Yes, you can make it. Does it have challenges with respect to cycle time, yield, and cost? Yes, that's also the case, which is why you see all of our customers move to EUV on future nodes for that exact reason.
The gentleman with the question in the front.
Yeah. I'd like to ask on IBM. Okay. Just gonna ask one on IBM. So I know you've clearly articulated why there was a revision in the expectations for this year. I think you guys are still guiding to some ramp through the back half of this year. Can you help us understand what are the signs that, like, what is it you look to see from customers for IBM to pick back up? What sort of early indicators have you seen, just so we can contextualize that?
Yeah. Yeah, so just to clarify, so IBM is Installed Base Management. So some people may wonder, talk about a specific customer. Installed Base Management, and the question is, we said originally growing 5%, now we're talking, you know, we're flat in terms of year-on-year growth, looking last year compared to this year. And the main driver why we're now saying flattish versus growth is really around the upgrades. And so the question was really more of when or what should we look for or expect to start seeing the upgrade business pick up? And the... If you go a year ago, we have said, well, the challenge with doing upgrades are that the utilization levels are high because customers are trying to produce all they can. So we need utilization levels to come down.
Now, utilization levels have come down, and we're saying: "Well, why can't you do upgrades now?" And it's because of the uncertainty. So now the fab manager is okay with it, but maybe the CFO is not okay with it, and that's the dynamic that's at play, with respect to this year. In terms of what to look for in terms of, recovery, I think if you look at the utilization levels as they come, down and then inflect and then start going back up, that somewhere in that point in time where you start seeing going back up but not getting to the, you know, high levels, again.
But in that recovery period, when the slope's starting to move up, it gives confidence to the CFO that, yes, I'm in a recovery, and yet the fab manager goes, "I'm still not, you know, at max utilization levels," would be a logical area where customers would say, "I'm willing to start doing upgrades." And some of them are quite short in their lead times, others could be a bit longer, but that's kind of why you see upgrades being a bit lumpy in terms of history and when they occur. It's gonna be tied more to the recovery. And as I mentioned, for logic, we're starting to see some flattening. If we start seeing that pick up, that could be an opportunity going forward. We still have to wait and see when that will occur on the memory side.
Okay.
Yeah. Yeah... So we haven't seen enough of it yet to warrant a growth, but as that picks up in time, obviously, you'd expect the upgrades to, you know, track along with that.
Great. I think we're just out of time.
All right.
Thank you so much.
Thank you.
Really appreciate the discussion. Thanks all for joining.
Thank you, and thank you all.