Bemiconductor Industries Aktie

Good morning, good afternoon, ladies and gentlemen, and welcome to Besi's quarterly conference call and audio webcast to discuss the company's 2026 second quarter and first half year results. You can register for the conference call or log into the audio webcast via Besi's website, www.besi.com.

Joining us today are Mr. Richard Blickman, Chief Executive Officer; and Mrs. Andrea Kopp, Senior Vice President of Finance. [Operator Instructions] As a reminder, ladies and gentlemen, this conference is being recorded and cannot be reproduced in whole or in part without written permission from the company. I would now like to turn the call over to Mr. Richard Blickman.

Thank you. Thank you all for joining the call today. I'd like to remind everyone that on today's call, management will be making forward-looking statements. All statements other than statements of historical facts may be forward-looking statements. Forward-looking statements reflect Besi's current views and assumptions regarding future events, many of which are, by nature, inherently uncertain and beyond Besi's control.

Actual results may differ materially from those in the forward-looking statements due to various risks and uncertainties, including, but not limited to factors that are discussed in the company's most recent periodic and current reports filed with the AFM. Such forward-looking statements, including guidance provided during today's call, speak only as of this date. Besi does not intend to update them in light of new information or future developments nor does Besi undertake any obligation to update the forward-looking statements.

For today's call, we'd like to review the key highlights of our second quarter, the 6 months ended June 30, 2026, and update you on the market, our strategy and the outlook. First, some overall thoughts on the second quarter and first half year '26. Besi reported strong second quarter and first half year '26 results as favorable order momentum continued for both our traditional and wafer level assembly systems.

Revenue of EUR 249.9 million increased by 68.7% and 35.2% versus the second quarter last year and Q1 2026, respectively. Growth versus the second quarter last year was primarily due to increased AI spending for photonics, data center and hybrid applications as well as increased demand for mobile applications. Additional hybrid orders were received in the second quarter this year from 2 repeat customers and 1 new hyperscaler customer. Similarly, orders of EUR 292.9 million rose by 128.8% versus the second quarter last year and 8.6% versus the first quarter of this year.

Of note, Besi's orders for the last 12 months increased to a record of EUR 987.6 million. Q2 net income of EUR 89 million increased by 177.3% versus the second quarter last year and 72.5% versus Q1 '26, as gross margins improved and operating expense growth was limited despite increased spending for development and customer support activities. Similarly, net margins increased to 35.6% versus 21.6% in the second quarter last year. For the first half year, Besi's revenue amounted to EUR 434.7 million and net income of EUR 140.6 million increased by 48.8% and 121.1%, respectively, versus the first half of '25, primarily due to significantly expanded AI infrastructure spending, a modest recovery in traditional mobile and industrial markets and disciplined overhead management.

Similarly, net margins increased from 21.7% in the first half of '25 to 32.3% in the first half of '26, aided by improved gross margins and significant operating leverage in our business model as baseline operating expenses reduced from 30% of revenue in Q2 last year to 18.9% in Q2 '26. H1 orders rose to EUR 562.6 million, an increase of EUR 302.7 million or 116.5% versus the first half of last year, due to broad-based growth across all Besi end-user markets and products. Order growth was strong for photonics and data center applications and hybrid bonding capacity expansion for both current and next-generation AI devices.

We also received new orders for AI power management applications in the second quarter this year from multiple customers. Overall, we estimate that system orders for AI applications rose to approximately 60% in the first half of this year versus approximately 50% in the first half of '25 last year. In addition, we saw renewed growth for high-end smartphone applications in the first half of this year versus cyclical lows reached in 2025 due primarily to incremental capacity purchases and new product introductions planned for 2026.

Our net cash position at the end of the second quarter '26 increased by 58.8% versus March 31 of this year to reach EUR 164 million. Growth was primarily due to the conversion into equity of Besi's EUR 175 million of convertible notes due 2029 and strong cash flow from operations, which offset the payment of EUR 125.4 million for the annual dividend paid out in the second quarter of this year.

Next, I'd like to discuss the current market environment and an update on our strategy. Overall, we see favorable industry conditions this year due to the ongoing AI infrastructure build, capacity shortages, particularly in memory and in CoWoS and CoWoS-like and renewed unit growth for both AI and traditional mainstream end markets. Similarly, we see ongoing improvement in assembly equipment market conditions as the impact of AI spending and the recovery in traditional mobile industrial applications becomes more apparent.

The latest Tech Insights forecast calls for 53% market growth between 2025 and 2028. We expect to significantly exceed such projected growth rates given our leadership position in 2.5D and 3D advanced packaging and wafer evel assembly. We continue to execute well on our strategic plan presented last year for both revenue and cost initiatives. Orders have already exceeded peak levels from the last cycle and disciplined expense management has enhanced our profit potential. We are adjusting our operating model, supply chain and service support activities accordingly. Progress also continued on our wafer level assembly agenda this year. Hybrid bonding customers adoption increased from 15 at year-end '25 to 21 at the end of Q2 '26. Use cases increased for logic, memory, co-packaged optics and consumer applications and orders increased material versus the first half of '25 as significant new capacity was added.

In addition, there were multiple new product announcements made this year related to data center and consumer CPU applications utilizing hybrid bonding. Progress also was made on our TC Next agenda with increased revenue and customer adoption versus the first half of last year. Now a few words about our guidance. We see order momentum continuing in the third quarter of this year due to ongoing demand strength for current and future AI applications as well as improvement in Besi's traditional mainstream end-user markets.

Customers indicate that we are in a multiyear AI CapEx cycle, further supported by increased demand for agentic AI applications, which are driving increased demand for data center CPUs and many of our advanced packaging systems. Besi's strategy is currently focused on expanding our opportunities in wafer level assembly, increasing our penetration of CoWoS, CoPoS, and Photonics markets and ramping our supply chain and service support capabilities in alignment with market conditions.

Based on our backlog and feedback from customers, we anticipate that Besi's third quarter '26 revenue will increase by 10% to 15% versus the second quarter of this year. In addition, gross margins are anticipated to decrease to a range between 63% and 65% due to a less favorable product mix than we had in the second quarter of '26. Operating expenses are anticipated to be flat to up 5%, due primarily to increased development spending. That ends my prepared remarks. I would like to open the call for questions. Operator?

[Operator Instructions] Our first question comes from Didier Scemama from Bank of America.

Just wanted to ask you a quick question on the Q4 bookings outlook, Richard. If you could tell us a little bit more about the contribution to your confidence about the strength of that order intake momentum. Is that also driven by memory? And if you could give us an update on your thoughts on the adoption for HBM4E, that would be great.

By the way, you asked Q4, but let me focus on Q3 first.

Sorry, I meant Q3, sorry.

No problem. We -- as we indicated, we see continued momentum for the business for the second quarter. But your question is very important from the, let's say, aspect of how that will develop for the different applications. And we certainly expect in the third quarter, continued orders for hybrid bonding. And as we have said on the previous quarter calls, whether that was February or also end of April, we are in a very interesting, exciting moment in time for the adoption of hybrid bonding for HBM applications, whether that is HBM4E or even HBM3.

As we speak, evaluations are going on at one of the major -- one of the 3 memory companies in our industry. Also, we've indicated that another one is starting the evaluation currently. So it will be exciting to see the outcome in the third quarter, but also into the fourth quarter, how much and which application then will be using the hybrid bonding technology. There's a lot of, let's say, views of how this will develop in the industry currently. Some are voicing that it may delay. Others are voicing that it is imminent. So in that exciting landscape, Besi with its hybrid bonding technology, together with Applied Materials in automated line concepts is at the forefront of this technology being used in the memory.

As we indicated in the prepared notes, we see continued adoption in the logic arena. We've added 3 more customers this quarter, one being an hyperscaler. That is for the use of the future classes. But there are many other applications, which are developed and in test applications currently using this hybrid technology. So broadening the applications is a fact to 21 customers so far. So anyway, a bit longer answer to your question. It will be very exciting to see how the third quarter will pan out.

Okay. Very clear, Richard. And for my follow-up, I wanted to ask you a little bit about how you're thinking about mobile. So one, you said that we have seen or you have seen some improvement in the mobile business from a very depressed level. As we move into the second half and in particular, Q4 and Q1, normally, the order intake is dominated by the refresh of high-end smartphones. So any thoughts around that? And also related to high-end smartphones, what are your thoughts on the adoption of hybrid bonding in high-end smartphones over the course of '27 and '28?

First of all, we made the comment because we have received significantly more orders so far this year compared to last year, as we called last year, sort of, a cyclical trough. And that is always driven by new features or updated features like, for instance, new cameras, but also other components in those high-end smartphones are in a way, cyclical because of new designs in next generations. And as we indicated, we have enjoyed a significant improvement in orders in that front.

At the same time, we mentioned 3 months ago that we saw the first adoption of hybrid bonding, hybrid bonded device, the M5, into high-end laptops, and they should find their way at some point into the high-end smartphones. Not yet we have seen those orders. But if you look at the -- simply the plan for capacity expansion in Taiwan as we have started now to fill the AP7 with the first systems, which is publicly informed to the world. We should see in the next quarters how that capacity is also expanded for other applications than for logic devices so far in the high-end compute. So data centers, that was the first. So that's well on track.

Okay. Perfect. And maybe my last question relates to the HBM again, adoption. So the JEDEC body has, sort of, loosened up the thickness requirement for HBM4E. And so I just wondered whether you think that has got any impact on the adoption curve of hybrid bonding in HBM4, 4E or 5 because there seems to be a lot of confusion around that in the market.

Well, we saw the first increase now over 2 years ago, which, in a way, stopped the market saying, well, we don't need hybrid bonding anymore. But we have seen the contrary in development. And as just explained, the year '26, all 3 are preparing to adopt at some point hybrid bonding for the very simple reason that the performance of the device is significantly better, we have told. And at the same time, it produces far less heat. Many numbers are publicly shared by some of the memory producers, which simply indicate whether you allow an increase in height and stacking that does not improve the performance or the heat characteristics.

So the technology moves on. One of the reasons we hear is simply because there's a huge shortage in the memory market as we speak and allowing a higher, and let's say, package height is simply offering the industry more devices, albeit at performance levels which are different from that using a hybrid bonding solution. So the amount of development is only increasing and the end markets simply require the best solutions for performance and other characteristics like heat. So again, my message is that the heat or the height does not change that characteristic.

The next question comes from Charles Shi from Needham & Company.

Richard, I have -- the first thing I want to start is on HBM qualification now, all top 3 HBM customers have your tools. And I think you mentioned a little bit, but can you kind of talk through hopefully one by one that where the qualification status is and what are maybe some of the major hurdles you are seeing in terms of getting hybrid bonding qualified at each one of the customers? And maybe I have a couple more questions afterwards.

The major hurdles are simply cost and yield. If you look at any technology in this industry, but also for other industries, any major change requires qualification, and that always starts with yield issues. You have material issues, you have different processes. So if you look at the entire production process of such a device, bonding is one step in that. Preparation is very critical. So all these processes need to be simply qualified.

At the same time, because the process is more advanced in terms of accuracy, also because of the nature of that process, that all requires -- and if you take simply as a reference how long it took for Taiwan, the major customer, to have the logic application up and running. It took us over 3 years. And finally, it has met those yield and cost challenges and is only expanding from here. The same in the U.S. after a long time of qualification at the major U.S. logic customer, we are finally in the stage that this is becoming a mainstream application. And that qualification process is identical with the 2 Korean customers, also with the U.S. customer.

So we're at the very beginning of this technology moving from the development stage, proving all of its performance characteristics in every way, defining the material specs, et cetera. And we are step-by-step coming closer to a mainstream market application. So between development and operations. Development is always completely different in terms of organization and companies as compared to operations. Operations are only interested in throughput in yield and ultimately, the cost of the device. So as we have indicated in previous quarter updates and also at the Capital Markets Day or Investor Day, we are making significant progress at all 3.

One is currently driving the bus the fastest, and we all know that, who is also announcing that publicly. So we will shortly find out do we pass the test for full production at this moment? Or does it take a bit more time? The other 2 also have stepped up simply their engagement and testing and finding mainstream end applications. And so in the second half of this year, it will become more clear, will there be inroads into HBM4E or even HBM3, because there are different end applications, which require a better performance and especially less heat. So all those developments are going on currently, and that's the landscape. Does that answer your question?

Yes, very clear. I have a follow-up on the hybrid bonding orders you got in the past quarter. It sounds like it's all logic applications, 2 repeat customers or 1 hyperscaler. Can you provide a little bit more context, especially around the hyperscaler order because it's interesting because we're not expecting a hyperscaler actually going to buy tools. Can you tell us what is this all about?

Well, what we know, and it's not always clear. There's a lot of development as we have indicated on the longer-term picture on applications, which are in, let's say, wearable products, glasses, for instance, it may well be for that application, but that's not unique. There are many customers or many several in that arena, we know and all of us know very well that the next step in wearables is those glasses. It can also be for other applications.

But then as you rightfully said, it's -- logic is expanding, if you may call it logic in many ways, also co-packaged optics, but that's not the hyperscaler, but co-packaged optics is an application which we expect to grow significantly in the years to come and you need the hybrid bonding technology to connect that into the device. And there are other chiplet architecture devices, which are also developed using hybrid technology.

Maybe last question from me, Richard. Your Q3 guidance, I found this as pretty unusual from a seasonality perspective. The seasonality we're familiar with in past 20 years that Q3 are down, it can be 10% to 25% in some, like, downturn years and Q4, maybe minus 5% sequential. But you're now guiding Q3 up 10% to 15%, a bit unusual from a seasonality perspective. And I think you mentioned about order momentum.

You mentioned about maybe even some recovery in the mainstream part of the market. Do you see maybe looking out a little bit further into Q4 or maybe a little bit beyond, is this -- I mean, above seasonal pattern lasting a little bit longer this year? And what do you see, especially in the mainstream part of the market?

Excellent. Well, if you -- we have this lovely slide in the deck going back to 2006, where you see the quarterly development in the past 2 decades. And you're absolutely right. We're usually first half year loaded as it is called. But that was always because of consumer end market characteristics. So if you take 2 years, 3 years ago, the major part was always communication devices, high-end smartphones. In the last 2 years, that has changed into compute AI data center applications. So that has a different characteristic in terms of seasonality as we were used to.

That is one. The second one is the pattern as discussed in expanding hybrid technology capacity in the world, both in logic and also if all goes well in certain early capacity for HBM. So that may well change the order pattern. But to go with the revenue pattern, and that's nicely depicted in that slide, usually, we have in a strong year, a revenue -- a higher revenue in Q3 than in Q2. So usually, then the orders come down. But as we indicated, we have continued order momentum, but with a different mix. So that's in a, let's say, in a way, you can understand that guidance.

[Operator Instructions] The following question comes from Martin Marandon-Carlhian from ODDO BHF.

Hi Martin.

It seems Martin might have some trouble. We'll move on to the following question. This is from Martin Jungfleisch from BNP Paribas.

Maybe coming back to the hybrid bonding orders in the quarter. In the press release, you mentioned hybrid bonding orders from 2 repeat customers. First of all, can you disclose if these repeat orders were for CPO, logic, memory or even R&D application? And what is your view on the order cadence here from your big Taiwanese customer? There was some news that this customer may actually accelerate CoWoS in favor of SoIC. So just wondering if you see potentially this customer, kind, of slowing down order momentum in the short term here? That's the first question.

Well, to start with the last comment, we don't see that currently. We only see continued, let's say, efforts and pressure on us to expand the capacity. So -- but there's many -- as we said earlier, there are many views shared by different sources, but we don't see any slowdown in that. On the orders received from the 3 customers, as I answered earlier question, the hyperscaler is probably for wearables. The other 2 are -- one is related to co-packaged optics and the other one is simply logic.

Great. No, that makes sense. And then the other question on IDMs. I mean IDM-related orders were up EUR 30 million versus Q1. Can you just disclose was that hybrid bonding? Or was that maybe the mainstream kind of AI power-related application that you were talking about in the press release?

The mainstream, that was not hybrid-related. It was very much our other products, flip chip, also MMA, also conventional. So very strong orders as we explained in the press release, but also in the comments across the board. The only part which is still not very much improving is automotive.

Okay. Interesting. And the Power segment, is that some, I guess, new segment for you here that you would potentially see increasing over the next couple of quarters as the whole 800-volt architectures ramp up?

Yes. That -- although power is, of course, for us a long time part of our business, but we see increased and also a lot of development in applications, which have to do with data center compute modules. And that is, let's say, becoming a much stronger business segment compared to a year and 2 years ago. It's a bit in line with the increased overall demand for data center compute.

The following question comes from Sandeep Deshpande from JPMorgan.

Richard, I have 2 questions on your products. Firstly, on -- I believe that you have broken into the CoWoS market. And how do you see that -- has that been part of the orders in Q2, both in terms of CoWoS-L as well as the original CoWoS at some of the ODM or rather the contract manufacturers who are doing CoWoS now. So how do you see that having trended in terms of the orders? Or is this something which is going to happen in terms of your order book in future quarters?

No, no. That's already a significant part of the orders. We always call that CoWoS-like. And that is -- that already started, you can say, in Q4, Q1, Q2 and should continue to expand.

Understood. And then what about the TC Next? I mean, how do you see TC Next? I mean, is this already part of the order book? Or this is still being tested by the clients that who have bought the early tools and that this will ramp up later in the year or next year?

As we commented, the customer base is expanding. The amount of qualifications is significant. The industry is moving to applications below 20-micron bump pad pitch. And as you know, our system is designed for that segment, so between 10 and 20, also fluxless. And you see increasing engagement in qualifications, but also systems ordered, single systems, not yet major volume. There's one who has several. But that is early days with ever more clear future demand.

And in terms of timing on when do you expect the volume ramp here as well as what products are you looking at the volume ramp like you talked about on hybrid bonding where your demand is coming from?

Combination logic, it's in first instance, it's very -- yes, similar to hybrid. So the critical thing about bump pad pitches below 10-micron to do that with a reflow process and you would also like to do that with fluxless. That is the most intended application. But already, pretty soon, it also found its way into the development of HBM with smaller geometries. And that are both drivers where this system is very much suited and also offering superior cost of ownership results.

The following question comes from Nabeel Aziz from Rothschild & Co Redburn.

I was just thinking about the cadence of hybrid bonder shipments through 2026. I was just thinking, should we expect half-on-half increase in hybrid bonding installations, particularly with AP7 investments ramping?

Well, that's a good question. If all goes according to plan, that may well happen in that sequence. As we know, the, let's say, published total capacity should be roughly double that of AP6. In AP6, there are about 60 bonders. So that still has some way to go to double that. That's always done in incremental steps. we explained the first round orders received in Q1. There may be another round in the second half, we don't know, with a lead time of about 6 months that then explains the -- when it will become revenue.

That's great. Very clear. And then just as a follow-up on CoPoS. Could you talk about what the move to panel-level packaging could mean for the Besi business and what the kind of ramp profile you're expecting for CoPoS over the next few years?

Well, it's becoming ever more clear that the industry is selecting a panel size 310x310. Also larger panel sizes indicated 510x515. But as we know, Taiwan has selected 310x310, which also makes a lot of sense compared to a wafer size currently. Our systems are mostly prepared for panel size 310x310. Also, we have orders to that extent, same with packaging, by the way. So the industry is moving into that direction and not for all applications, for certain applications. So that is definitely part of our, let's say, market going forward.

The following question comes from Martin Marandon-Carlhian from ODDO BHF.

So in the press release, I wanted to understand better a statement. When you talk about increasing penetration of CoWoS and CoPoS, how should we understand it because -- are you talking more about flip chip bonding opportunities or even about TCB there?

Both, and also hybrid. Don't forget that you may well see certain modules assembled in a panel format. So depending upon the size, you can have a certain amount, and you would need those 3 technologies to build those devices. And in the Investor Day material, you see some examples of that. And the most easy way to understand is if you have a round format, you lose corners. If you have a square panel format, you use also those corners. And that increases the efficiency and in the end, reduces the cost of ownership.

Okay. Very helpful. And my second question is on TCB because you had this quite a large order last year in memory. We didn't see really repeat orders since then. And I wanted to understand what's your understanding behind this? Do you think it's because the customer could hesitate with hybrid bonding? Or is there another reason?

No. The reason is quite simple. Every customer is preparing for a next step in miniaturization of chip design. So that is -- always takes time and the development of the next generation can be 2 years, 3 years, 4 years. And in order to be fully prepared, that's how customers buy systems initially. I can explain in many details for many other customers how that unfolds year after year after year.

And the timing is often dependent on many other factors and market factors. Will they use the current technology for a next round, can they stretch that? Or do they need to change to smaller geometries. That's always a trade-off. So that's how you should see that development. But it's very important to be the process of reference at an initial stage. Then if you do it well, you can have the largest opportunity once it becomes mainstream.

The following question comes from Marc Hesselink from ING.

My first question is on the AI power management, which you already addressed a bit. Historically, whenever you start calling elements out in the press release, it typically was the start of a pretty strong period. And I think you discussed this already a bit at the Investor Day. But maybe a bit more detail. Is it -- did you really make now a first real inroad with more significant volume than the years before with one client? Are there more clients to follow? How do you see this ramp pattern into the next few quarters?

Well, as you rightfully concluded, we only mentioned when we feel with high certainty that, that could be a significant mainstream contribution to the total. So yes, there are more customers. As I explained earlier, it's part of the total offering to this data center module end market. And power is a very critical part always. And so that is the background of that statement. We don't mention specific customers, but I can say that it's more than only one customer, it's broader.

So that like with Photonics and the pluggables or with the oS, on-substrates from CoWoS, same like chip-on-wafer, same story. That all is part of the road map of 2.5D, and that gives us a major opportunity. Same like photonics and co-packaged optics, they're all focused on specific end markets, which are mainstream currently or in the future like co-packaged optics. Expectation is very high for that market going into the future. So anyway, you have to see that comment in that context.

And second question is also coming back on an earlier answer. It's on the seasonality. I think you already explained that now the biggest driver is not the communication anymore, but more the data center, which has a different seasonality. But then in the communication part, the memory part, is there still the usual seasonality? And the reason I'm asking is I also read some stories that maybe the cadence of introductions by Apple might be also an introduction in the first half next year that maybe therefore, also the communication is a bit more split out? Or is that something you are seeing?

Well, we've heard that many times in the past, the -- let's say, it all depends on judgment of end markets and consumer markets are very difficult in that sense. So we also only hear at the final moment, usually January, early February, what the decision is of the content change in the next generation to be launched in September time frame.

Sometimes you hear that, that may change to spring time, but we have not seen that. But for us, it doesn't make much difference other than planning of production capacity. So the typical machines used for, let's say, high-end smartphones is, of course, flip chip, MMA for multi-module attach. Cameras is one of the key end products, and you can expect a change in those cameras, which is also indicated. Then you have the whole 5G, but also the power, the -- all the critical components. And they all have, yes, let's say, suppliers in the supply chain for those end products.

We have not seen a change in structure yet. We've heard that, yes, but we haven't seen it yet. So the round we are now, let's say, emphasizing is a clear round, which has to do with the generation, which will be introduced in September, October. And so orders for that we have received in first quarter, in early part of the second quarter. And maybe if that all is successful in adoption in the world, you may see additional orders in the third quarter. So that's typically the pattern.

The final question comes from Robert Sanders from Deutsche Bank.

Maybe if we could just take a step back and talk a bit about the potential for overcapacity building up. I guess, in particular, 2.5D, what worries you more about what you see today in terms of the potential for overcapacity building up? Is it the CoWoS look-alikes? Or is it China? You've been through many cycles. So I'd be interested to sort of get a sense of what you see. And then I have a follow-up.

Well, first of all, thanks for this question. If you look at previous cycles in our industry, typically, an up-cycle last 6 quarters, maybe 8 and then you are bound to have an overcapacity. As soon as you hear new fabs being built, new assembly sites, advanced packaging being built and by different major customers and then also the whole subcontractor universe expanding capacities, you're bound to -- will see at some point an overcapacity.

So what we see in the market clearly reflects the statistical analysis of what has gone on in many cycles. There are many always other prophets who are saying, no, this time, it's different. It will last another super cycle 3 years at a minimum, because of the overwhelming demand. We've also heard that before. So it's a very interesting question. But then if you look at Besi, we can accelerate, decelerate, look at our margins, our ramps and also deramps with increasing margins in this cyclical world, and it remains cyclical. You simply have to follow the demand cycles of the leader end customers.

So are we heading into an overcapacity? Time will tell. We adjust our operations, our supply chains, long lead items 3 times a week, simply following exactly the demand patterns of those customers, as I said. We are suppliers to an industry which today has major drivers. And of course, AI is a game changer in many ways. Will that follow a continued growth path? Nobody knows. So I'm sorry, Rob, I can't give you a better answer.

But do you think that this cycle is potentially more dangerous just because the OSATs, they make 18% gross margin, they can make 40% in CoWoS look-alikes. Doesn't that make it a little bit more dangerous? Or you feel that's always the case that there's some attractive high-margin market that they can go after?

Well, that's the name of the game. So any high margins is bound to result, well, take the memory market right now. Who would have thought 2 years ago that you would ever enter the memory market into a world with margins, which we have right now. So that's the danger in this industry. But as a supplier, you simply have to prepare, you have to enjoy the demand as soon as it is offered, and that's what we do. And then if the tide goes out, simply be able to adjust your cost.

Got it. And just one housekeeping question. I don't know if this was answered earlier, and I apologize if it was. But did you say how many hybrid bonding units were shipped in Q2 and how many bookings numbers you took in units for Q2? Just out of interest, if you're willing to say.

No, we haven't disclosed units. We did disclose the number of customers. We are now over 20, 21 customers. And we updated in detail the progress in HBM adoption with a very clear message that it will be very interesting in the course of this quarter and also the fourth quarter, how much and which end application that will use hybrid bonding.

I will now hand the word over to Mr. Richard Blickman for the closing remarks.

Thank you all for listening in. And in case you have additional questions, don't hesitate to contact us. Thank you again.

Good afternoon, everyone. Ladies and gentlemen, who are the last to come in? Five more seconds and the door closes. kidding.

Welcome, everyone, to our Investor Day presentation on June 18, 2026. Following a tradition in the meantime, every year in June, we try to update you on what's happening, what are the opportunities, what is accelerating, what is maybe not that accelerating, but it's not many today.

So an update on the market and what is it not to be very clear on that. We don't talk about Q2 numbers. You have to be a bit patient in 4 weeks. We will have our Q2 numbers. And that's, of course, focused on the quarter and the outlook for the next quarter. But the purpose of today is to give you a bit more longer-term perspective and how Besi is doing vis-a-vis the market, vis-a-vis the competition, some market share and more importantly, on the fastest-growing segments in our business, we will spend time to try to convince you that those fast-growing parts of the market we serve are leading to a new model, which we announced this morning up from EUR 1.5 billion to EUR 1.9 billion to EUR 1.7 billion to EUR 2.2 billion.

It's interesting that several of you, I won't call names, are a bit disappointed, as I read in the first comments. Well, that's probably in line with the world, which thinks that things will grow forever into whatever skies on Mars. Our models are based on a very careful assessment. We, as you know, some of you test every three years our strategy in a very detailed fashion together with our key customers. Also with the support of a major consultancy firm to have everyone on board and to simply try to be as accurate as possible to what we can expect.

Well, as a nice anecdote last year, we reached a revenue level of EUR 590 million something. If we would have been awarded a bonus for reaching that 2025 target, we would have received nothing because if you look at our budget, which we always make towards the end of the year as every company, about 40% of that EUR 600 million was completely other business than what we expected at the beginning of the year. And it won't surprise you, the AI really started to take off by the mid of last year, Q3.

And the other part of the market, the conventional market did not take off. So that was an interesting development. But just to tell you, we know a lot, but we certainly don't know everything which is moving this market. So a bit of conservatism is what we try to share with you. So this is a long page with all kinds of disclaimers, which you may know very well. Nothing has changed to that page.

But then for today, the agenda, I will give you a brief overview, and then Chris Scanlan will explain the technology part like Chris did last year. And then Peter, on the Sub Micron market developments, so both for Logic, also for HBM stacking. I'm sure you're all very keen to understand how that is progressing. Then also the other applications using a hybrid bonding process. Then we have a short break. And Christoph, after the break on the mainstream die attach. And as you may all know, that is really booming all the 2.5D engagement, photonics. Those are the parts which are driving the bus today, also new modules in high-end smartphones.

Anyway, our engine 1, as we also call that part of the business, Christoph will share the progress, the development road maps. What you can expect going forward. So that is part, if you take the whole model, the now EUR 1.7 billion to EUR 2.2 billion. About EUR 1.1 billion is the Engine 1 part of that. The other part is, of course, the Sub Micron part.

But Christoph will share with you a lot of drivers, market size, very important, Photonics, we often get the question, how big is that going to be?

I forgot to mention Peter, of course, on the co-packaged optics. That's more on the Sub Micron side of the equation. But anyway -- and then we have a summary and an open discussion, and you're most welcome to challenge us in every way because that's what life is all about.

So let's go to the strategic overview. Well, it won't be a surprise, the first sentence, overall market conditions have improved significantly, 50% plus compared to a year ago, which is simply a fact driving the whole AI infrastructure, and that's where Besi has a large part. The second message is equally important, how are we doing in that market? Are we gaining share? Are we losing share? But we'll share with you that we are definitely on the right track. We have the right products. We have the right customers, and that has improved significantly in the last 12 months.

Then if you look at our profile, look at the order growth, that also explains more or less what we've said in the first two sentences. But then also our margins. Many of you are used to Besi margins, gross margins somewhere in the mid-60s. In the down period, it was a bit trending towards 62%, 63%. Net margins, 21% at the bottom. And probably if you look at the guidance for Q2, it should move well again above the 30%. If you take the gross margin and the OpEx guidance. So anyway, our margin structure continues to improve. Hybrid bonding adoption, last but not least, use cases increase in logic, memory, co-packaged optics and also in consumer. Remember, Apple announcing the M5 in a hybrid version.

There are three revenue streams emerging each with attractive growth profiles. So what I mentioned before, the 2.5D assembly equipment growth expanding due to strong data center rollout and photonics demand, the Asian subcontractors, all for the current AI application. 3D assembly gaining traction, next-generation AI use cases and more than more architectures. Traditional mainstream is improving. We see that with all of our end customers, and that also has an impact -- a positive impact on our growth year-over-year.

Multiple drivers converging to accelerate the hybrid bonding growth. As we all know, in Logic, it started. And the reason, again, very simple, bond pad pitch limitations causing the switch from a reflow process to a hybrid bonding process has been the driver in the past 10 years. The first early adopters, AMD and then gradually expanding their product range, but then Broadcom and Apple, Intel, not to forget, they are all moving into hybrid bonded device architectures. Chris will explain much more. NVIDIA Feynman, those of you who have seen that road map issued about 1.5 months ago, clearly tells you that hybrid bonding is the way to go.

Then in HBM stacking, HBM4e, like we shared with you end of April, it's critical in the next few months to understand will that lead to a major mainstream volume? Or will that still be in sort of a niche part of that application. But as every one of those three explains the world, sooner or later, that switch to a hybrid process will become effect.

And then we have co-packaged optics. The COUPE at TSMC is one of the major parts of expansion of our business. We also shared that at the end of April, orders in Q1 received, and that is where we also have a lead in terms of new technologies.

Then adjusting our operating model, a switch to cost. So to be prepared for further growth in this industry, we have always used the simple method, doubling each time. This time, we should probably double -- prepare a bit more than double. That's all in place. And also what we have shared last year, if you share a model, that model should be in place. Otherwise, you're sharing a model which still has to be developed. So that's key to understand. So where we have all organized that operational part is in Malaysia, is in China coming out of Vietnam, supported out of Singapore and ever more support center close to the big end customers, of course, Taiwan. U.S., not to forget, for Intel, but also Micron and others and coming up TSMC. And then you have in Korea, the support for the Korean customers.

The targets, again, increased by about 15% versus last Investor Day, all identified business, identified customers where we also have demonstrated and process of reference positions, and that should help to bring our company into the next phase.

Some nice slides about CAGR. What is to be expected in the IDM and foundry capacity to double by 2030. If you look at the different green colors, you see the bottom end is advanced logic and foundry. That's where we are ever more established. The lighter green part is what is supposed to become far more clear in the second half of this year with enormous growth potential. And then we see also on top of that, the discrete analog and other. So a major market development model and it's all about this build-out has caused a multiple-year investment in infrastructure. Take TSMC's comments, Broadcom, Applied Materials and also NVIDIA. So that is basically supported.

If you look at our model, it is tied very closely to this model, of course. That the industry tide has turned very rapidly. It's very nicely shown in this graph here, where you see from Q1 '23, '24, '25 last year. And then all of a sudden, and that's simply because of memory, an enormous supply and then an offset with not sufficient production capacity. And the same we see here with unit growth, an enormous trend up. But if we look statistically, then there's always sort of a peak, which should follow at some point in time. I won't make any forecast today when, but if history repeats itself, that is what you should be watching out for.

Assembly equipment follows the same pattern in many ways, a bit less aggressive compared to front end. But you also see here from a EUR 5.4 billion market size in '25 towards in '28, EUR 8.3 billion. So enormous increases, and that's where we certainly will benefit from. And we see an enormous list of new advanced packaging fabs being constructed in the U.S., in Taiwan, Korea, India, well, simply name them, Singapore, China, Europe, Vietnam, Malaysia. So enough of opportunities. They won't come all at the same time on stream. But at all of these customers, we are in a very good position to benefit from the investments in those capacities.

This famous slide takes you back to 2006. Some of you have witnessed this trajectory all through the 20 years, and I'm very happy that you take the time to visit our Investor Day once again. And what you've seen here is a wonderful cyclical development, four years, sometimes a bit longer, but that's a typical pattern. And then if you look at the gross margins, where at that time, we were proud to reach 40% in a quarter, but usually below 40% then in the next 42%, 56%, 62%. And as we guided for the second quarter, we should be again above that level. So our next growth period, hold on one second, is in the cards with, as things look now, potentially higher margins. Where does that come from? Number one, we've learned over all those years to focus ever better on advanced packaging opportunities.

So where can you make a difference compared to our competitors and peers in Asia simply focus each time on the next-generation devices and in the markets, so the computing part, communication, automotive. And having done that systematically every three years, redefining our strategy, that has improved our focus tremendously, engaging the customers in the early stages, also customers supporting that development financially. And then at the same time, working on a better cost model. The cost model, multiple sourcing, production in Asia, headquarters in Malaysia production, China, Vietnam coming up and a global supply chain, which is simply based on a concept multiple sourcing, as said, but evaluating that constantly. And that delivers you an average growth higher than the industry and also margins substantially above any of our competitors.

Some more data here. If we compare Q2, Q3, Q4 and Q1 '26, you see the order intake two quarters now to 250 plus. Also, the revenue is certainly climbing. If you look at our cost structure, is intact, and that's what we have guided also for the second quarter. But basically, these financial metrics tell you how we run our company.

Market shares, you all want to know market shares. As we have said many, many times, market share is an end result. We look first at the margin potential because the margin is where it's all about. Return on capital and then ultimately, if you do that right, you see that your market share continues to grow, which is, in a way, very illustrative for Besi.

So if you take the addressable market, the Die Attach, the advanced Die placement market, gradually improving the overall share simply through selecting the best applications. Long-term outlook enhanced by increased hybrid bonding use. This gives you a snapshot, but Peter and Chris will give you much more detail of those customers which have adopted now in Logic to a large extent, but also on the brink of adopting that in memory stacking, some chiplet architecture.

And remember, we now have over 20 customers who have bought hybrid bonders from us for all kinds of applications. Mainstream volume is, of course, in Taiwan, most established than Intel with 30 bonders in 6 automated lines and gradually picking up for other applications.

The adoption, as I said, is expanding. We're now at 20. And we have listed these customers also for the different applications, Logic, Memory, Photonics, R&D purposes. But again, more about that in the presentation of Chris and Peter. Increased R&D spending. Of course, with all these wonderful applications and ever more complex, you see a gradual increase in R&D spends. And you see on revenue levels of [ 25 ], so close to EUR 600 million, 14.8% in 2020, below 10%. But if you look at the trend of order intake, and you simply divide that by the spend in R&D, you can easily calculate that, that in line with the revenue development will change. But the key is, are we investing in the right end applications. And you can easily conclude that if you look at the end margins. So the focus is right. It is gradually increasing, ever more complicated. Look at hybrid bonding, the 50-nanometer, also fluxless TCB, the flip chip multi-module attach. So many applications, which are in the mainstream and will provide us significantly more opportunities to add to the total model we shared the updated version this morning.

That ends my presentation. Chris, it's your turn.

Thank you very much, Richard. I'm Chris Scanlan. I'm SVP of Technology at Besi. And first of all, thank you for joining us here today. Thank you also to everyone on the phone or online listening. I'd like to share with you our view of the market, some of the applications specifically that are driving growth for us and creating opportunities for us. And most importantly, what's changed in the last year since we last spoke together.

First of all, what's changed since last year is the AI market has taken off really faster than we expected for sure. I think last year, when we presented the semiconductor market forecast, we were expecting about $1 trillion in revenue by 2030, which was also impressive. But it looks like we're going to hit that and exceed that already in 2026. Tech Insights anticipating $1.4 trillion already this year. So that's a material change. A lot of that is pricing, of course, Memory pricing is much higher. But underneath, there's also unit growth. So about 25% increase in units month-over-month in April this year compared to the last year.

So overall, this is really driving opportunities for us. And Tech Insights anticipating $2 trillion by 2035. A lot of forecasters, including TSMC, are looking at $1.5 trillion by 2030. pick a number, but what's really driving it is AI infrastructure. So there's a massive build-out ongoing right now. I'll explain why, but it's supporting training and inference using these models for productive purposes as well as all the devices -- AI-enabled devices that interface with these models. So things like mobile phones, computers, physical AI devices like cars and robotics as well.

So how do we measure this adoption rate of AI? And I think the best way to measure is the output of the AI factories. So AI factories produce tokens. Tokens are a little bits of information. A token is approximately a little bit less than one English word worth of information. And this chart from Goldman Sachs shows a forecast for token usage coming out of AI models through 2030. The top line number is 120 quadrillion tokens by 2030 per month. And it's hard to even imagine what that means. But to put in perspective, the library of Congress in the U.S., which is a pretty big library, all the printed works in that library would consume about 3 quadrillion tokens.

And already right now, we're producing more than that per month using AI models. So it's already a really big capacity. and that's going to continue to grow. Why? Because these things are very useful. When we first had ChatGPT in December 2022, we could do some really simple things with it. It's kind of fun, but we couldn't really do much with it. But really, I would say, in the last year, that's really radically changed, and we're all using AI now for really productive purposes like coding, like actually creating products. And this is driving demand.

The other thing that's driving demand is in November of last year, there was an introduction of something called OpenClaw. And that was really the first kind of tool that allowed people to make agents in a very simple way. And what is an agent? An agent is something that can automate the use of AI, and do it in a logical step-wise function. So it kind of amplifies our ability to consume tokens. And that's really what's going to drive the consumption in the future. It's no longer dependent on us humans to take some action in front of the computer.

In order to create that capacity for tokens, we need hyperscalers to invest in CapEx for data centers because data centers are the factories that make tokens. And I'm showing some data from Tech Insights, and there's other forecasters as well anticipating about $1 trillion in CapEx spend per year by 2029. Some of you earlier today in discussions told me, no, it's going to be $3 trillion in 2 years. I don't know, but it's a big number. And all this is having an impact on Besi in our end markets, where do we see our business coming from?

If you would have looked 2 years ago, it was 30% mobile. That was the biggest market for us and probably extending 10 years prior to that, we were really driven by the mobile market. But currently, because of the AI build-out, we're more than 50% computing this year and mobile is only 16.7%. And if you look at what kind of devices and what kind of products we're supporting, it's all the components that go into these AI factories.

So what's an AI factory? An AI factory is something that takes electricity as an input and uses computers to output tokens. That's the output. And these factories are enormous. This is a picture from NVIDIA's keynote at GTC. Some of these factories have over 100,000 servers, 100,000 GPUs. And there are typically multiple of these buildings on a campus. The biggest ones are talking about using 5 gigawatts of power. So it's massive power consumption. Therefore, we have to do everything we can to create our devices to perform as energy efficiently as possible to produce the most tokens per input power.

And what kind of devices are we talking about? It's things like we all know GPUs and accelerators. These are the things that can do really mass parallel mathematics, so made by companies like NVIDIA and AMD and all the hyperscalers as well. And then CPUs, also very critical. These are the devices optimized more for stepwise computing functions following a list of commands. This is becoming more and more important for Agentic AI because in Agentic AI, we're not just calling a model to do something, but we're actually logically creating a stepwise launching of multiple agents, multiple model steps in sequence. And for that, you need CPUs.

And then you have memory, HBM memory, of course, that's what's storing all the parameters for these models as it goes through the sequence of processing and creating the model and then the inferencing for the model. And then, of course, we need to connect all these computers together because they all have to function as one device in this massive building. And for that, you need networking. So you need network switches. These are really high-performance ASICs that direct the traffic, basically the data traffic within the data center and within the rack. And then finally, another aspect of networking is the Photonic transceiver. That's basically the thing that you plug your fiber optic cable into. It takes the light out of the cable, decodes it, makes it an electrical signal and delivers that to the switch.

And Besi is involved in all of these, and I'm going to describe a little bit about what's happening with these devices. One thing that's happening is in parallel with this big demand for AI, there's another phenomenon that's driving our growth, and that's kind of the death of Moore's Law. And if I had to put a time of death, I would say about 2023 based on this chart. So this chart is looking at the maximum size of flagship products coming from NVIDIA and other customers, other high-performance computing IC suppliers. And you see up until about 2023, culminating with the NVIDIA Hopper that was the H100 generation. All these devices were basically fabricated on a single chip.

But in this time frame, they were very close to the maximum possible size that's defined by ASML's reticle size basically. But starting in 2024, all that was no longer possible. It was no longer to scale transistor performance adequately in order to increase the number of transistors needed for the next generations. And therefore, you see every device now exceeding that reticle limit. That means that the devices have to be split into multiple chips. Once you do that, you split the devices into multiple chips, then you start looking for ways to optimize what function do you put on this chip versus that one. We call that concept chiplets. That's splitting devices into multiple chips using different process nodes to optimize cost and performance and then using advanced packaging to put it all together, and that's what we help with.

In this market, the chiplet market is growing at a 69% CAGR. This is a view of what the hyperscalers are doing. So the hyperscalers are not just buying -- of course, they buy a lot of devices from NVIDIA and Intel and so on. But a lot of them are also developing their own chips optimized for their own AI workloads. You see companies like Google, Amazon, Microsoft, Meta, all developing their own chips. And the reason is simply to increase the efficiency and reduce the cost of producing tokens in the data center.

But what they all have in common is they're all very large devices, mostly multichip and all using advanced packaging, 2.5D and 3D packaging. What does that packaging look like? Well, I've shown you a picture like this before. But all these packages, all these devices are using variants of these kind of 2.5D and 3D assembly technologies. And the important thing is that they combine multiple different bonding steps, die attach steps in the same product in order to combine all these devices together. And so there's different types of bonding required depending on which layer of the interconnect you're talking about.

So starting with hybrid bonding, there's multiple use cases inside these systems. I'll draw your attention to the center top. You have the SOIC. So this would be fabricating the logic device itself, for example, a CPU, GPU in a 3D configuration where you have a base die with a certain function, could be SRAM or I/O. And then maybe the processor cores, a smaller chip that's placed on top. And the most effective way to make that connection is using hybrid bonding, and I'll explain why in the next slide.

So this fabrication of 3D ICs using hybrid bonding. The next application of hybrid bonding is in HBM. This is a future application we expect with HBM4e, but hybrid bonding there will allow us to integrate more chips with better thermal and electrical performance and ultimately higher performance for the whole system. And then finally, in the CPO or co-packaged optics application, hybrid bonding is also used.

And let me just explain what CPO is a little bit because I know there's a lot of interest in this technology. Basically, what we're doing with CPO is taking the light from the fiber optic cable and routing it directly to the package. So we have a fiber optic connector on the top. The little blue line represents the fiber. The fiber is attached or an array of fibers are attached to the top of this chip. There's little mirrors etched into the silicon that direct that light down into the bottom chip called the Photonic IC. The function of that chip is to capture that light, turn it into electrons. And then it has to go up into a high-performance processor. We call that an electrical IC. That processor has a job of packaging that information, communicating to the ASIC very quickly. That whole thing is done using hybrid bonding.

So the EIC is connected to that Photonic chip using hybrid bonding. And I'll show you an example of that from NVIDIA in a minute. So once we've done all that, we still have assembly to do, right? So we have the SoIC, we have the HBM. We have to assemble it onto an interposer and test that thing. And that assembly process involves placing all these components onto a wafer. That's why we call it chip-on-wafer. So this blue interposer represents a wafer that our machines place both the HBM and the die onto. And that's a very important process as well.

We have different ways of doing it, either flip chip, which is done with this Chameo Flex machine that Christoph will tell you about or thermal compression bonding using our TC Next. In many cases, in a growing number of cases, you also have components assembled into the interposer. CoWoS L, for example, you might have heard of, L stands for local silicon interconnect, and that requires machines to place those local silicon interconnects very precisely inside the wafer. That's also done with this CamioFlex machine, which we're introducing.

And then finally, we have to assemble all that onto a package substrate also using either flip chip or TCB. So none of this is a surprise to us. We've been working for the last number of years to develop all the equipment necessary to enable all of this. And for our business, what this really means is that the capital intensity, the number of machines customers need to make these products is really increasing rapidly. So if you simply look back 10 years ago, your NVIDIA GPU, your Intel CPU was typically one chip. We call it a system on chip. And every year, we put more transistors in there through Moore's Law, but that started to break down. And the first step was to integrate these devices in multiple chips, call that MCM. And currently, we're now working on 3D stacking, so stacking chips on top of other chips and then further assembling those into the package.

And the next step now is adding optical interconnects into the package, which adds even more intensity to the whole assembly. So more steps, all those steps requiring more accurate equipment, which tend to be more expensive as well, all adding to the capital intensity. Now we have different choices, how we assemble these devices. In the case of 3D-IC, 3D-IC, meaning integrating two Logic devices or Logic and memory together in a vertical way. Really, the process of choice is hybrid bonding.

So in hybrid bonding, we create a direct copper-to-copper interconnect between the back end of line copper structures on both devices. There's no other interfacial material. It's really functioning as if it was fabricated on the same device. Because of that, we can achieve much higher interconnect density so we can drive the pitch down because there's no other materials in there all the way down today to 6-micron pitch, even down to 1 micron in the future. So the interconnect density is much higher. The speed is higher. Ultimately, the energy-efficient performance of stacking using this method is 100x higher than it is for TCB.

So for those high-performance computing applications, this is really the method that is being pursued. And even for memory applications, I'll show you how hybrid bonding helps to improve the thermal resistance, which improves the overall performance of GPU devices. So for these reasons, hybrid bonding is really being adopted as the primary method for 3D-IC.

I just want to share with you now a few examples of the kinds of devices that are using hybrid bonding and how and why. Starting with CPUs and GPUs, which are separating the core compute function in the SRAM and I/O function and putting them back together with hybrid bonding. So the reason they're doing that is because SRAM is really critical in these systems. We talk a lot about HBM, of course. But really, SRAM has a critical part to play in these computing systems also, always has and provides a very high-speed, high-bandwidth connection to the logic that is needed for certain compute functions, not as high capacity as DRAM, but very critical and getting more critical as we scale.

The problem is that SRAM doesn't scale very well as we go to advanced nodes, particularly as we make the transition from FinFET to gate-all-around kind of transistors. This data from Fujitsu is clearly showing that you really don't benefit by scaling SRAM to an advanced node. Actually, you can have performance degradation. And so the solution to that is to keep the SRAM on a trailing node like 5-nanometer or 7-nanometer and then use the advanced node only for the logic cores and then hook them together using hybrid bonding.

So that does a few things. Number one, it improves the overall performance of the device, but it also reduces cost because you only use the advanced node for a smaller portion of the design, in this case, 30% in the Fujitsu design. And the rest of it, you can use a trailing node at lower cost. Intel is using this method. You can see also for their new Xeon 6+, formerly known as Clearwater Forest. In their case, they have the most advanced node being used for just the compute tiles. They bond 12 of those things using hybrid bonding to a base tile. And that base tile is doing the function of, yes, I/O and SRAM.

The next type of product would be GPUs. Hybrid bonding already used by AMD in their GPU products, the MI family, MI350, MI355 and so on. And we see them continuing to expand the use of hybrid bonding and extend the design role capability. So they're already now down to 6-micron pitch, for example, with their interconnect using the best equipment. But we see in the future is other kinds of GPUs that have traditionally been single chip.

NVIDIA now has die stacking on their road map for the Feynman GPU family, which is due out in 2028. In addition to custom HBM, custom HBM means I'll explain a little bit in a minute, but we anticipate that custom HBM will require hybrid bonding as well.

Broadcom also has announced for their custom ASIC customers. So they do a lot of work for hyperscalers and developing accelerators that they have adopted this 3D IC technology along with TSMC, and they have multiple customers in development. And the concept there is very similar, stacking just the accelerator functionality onto an SRAM die or an I/O die and by that method, reducing the total cost and increasing the performance of the system. We expect multiple announcements from them in the coming year or so.

And then there's co-packaged optics or Photonics. So this is taking off faster than we expected last year. I think last year, we showed you this picture on the upper left. This is NVIDIA's Spectrum-X network switch device. And this is a network switch for scale-out networking. That means connecting different server racks to each other or different -- yes, different clusters to each other inside the data center. And what that entails is, first of all, they have a big network switch ASIC in the center of this package, but all the small squares on the outside of the package that I'm pointing to are co-packaged optics chiplets, just like I described in the previous slide.

So they're fabricated using hybrid bonding. So in this one example, there's 36 of those. That means 36 hybrid bonded chiplets associated with network switch. What's new is that it's no longer just NVIDIA. So NVIDIA is in production, but so is Marvell using the same kind of technology. And there's many other technology or switch companies or CPO technology providers working on similar technologies using hybrid bonding. So we think this is a really big market. But even more exciting is the use of CPO and scale-up networking. That means connecting GPUs together within the rack. That's a much bigger, much higher number of connections. And this has been announced for -- also for the Feynman generation of GPUs from NVIDIA. They have a dedicated NVLink network switch or not network switch, but a switch for the scale-up networking.

What this will require is not only the switch device, but also CPO components directly associated with each of the GPUs. So you can easily imagine that this is a much higher volume opportunity. And we expect this, like I said, in 2028.

Next is memory. So we've been discussing memory for the last couple of years, and we still are convinced that the intercept for hybrid bonding for HBM memory will be in HBM 4e, the first adoption and then a broader adoption in HBM 5. There's been a lot more data being generated in the last year since we last talked, demonstrating the performance benefits of hybrid bonding in HBM. One of the main ones is the thermal performance benefit that we get with hybrid bonding. When you stack many of these devices, you have sandwich of 12, 16, 20 layers. You have to somehow get the heat from the bottom ASIC out through the memory to the top. And hybrid bonding is much more efficient because it doesn't have any insulating materials in between the die. When we use TCB, we have the solder bumps, not as good as copper, and then we have some epoxy material that's also filling that gap, acts like a resistor. And the more layers you have, the worse it gets.

This data from Samsung shows with 12 layers, they're able to demonstrate a 30% improvement using hybrid bonding compared to micro pump for HBM. So it's thermal performance, also higher interconnect density, which will become important for custom HBM and electrical performance benefits. So for these reasons, we are convinced that this technology will be adopted. And like I said, HBM4e and then later HBM 5. And then we're now starting to see hybrid bonding percolate into higher volume applications. So this is a consumer application, for example. We have an example now from Apple. This is their M5 Pro, M5 Max family of chips for laptops. So really consumer devices. You can go down to the store and buy this. Maybe some of you already have it. And in this case, what we're doing is placing a separate GPU and CPU chip onto a base silicon layer using hybrid bonding.

And then connecting those two with a very high-density wiring layer in the base silicon die. And you can see the picture on the right, the actual hybrid bond interface, connecting the two devices together. And you'll notice it's basically just copper, copper. If you don't know there was a bonding layer there, it would look like just the back-end line metal and the chip because that's kind of what it is. It functions just like a single chip. But what it allows them to do is two things. One is they can mix and match, so they can take a CPU and combine it with different GPU configurations. That's how you get a Pro and a MAX. So only one CPU tape-out with two GPU tape-outs.

Secondly, it reduces the size of the individual chips, increasing the yield in the factory and helping to reduce the cost of the silicon. So this is really kind of a cool application, and we expect more of this in the future. AMD is also continuing to use hybrid bonding, but they're increasing the use. So they're basically adding more hybrid bonded chiplets into their desktop gaming CPUs, and they're able to do that because they've improved the performance. They don't have any losses like they used to associated with thermal issues and so on. So it's really functioning well for them. And then -- so that's about hybrid bonding, but it's not all about hybrid bonding at Besi.

So AI is offering many other opportunities for us. CoWoS is one of them. So as I mentioned earlier, after we do the hybrid bonding, we still have to assemble into a package. And for high-performance computing, the main method of packaging these different components together is CoWoS, chip on wafer on substrate. And this is really exploding now in terms of demand. There's a 50% CAGR expected between 25% and 30% in terms of the wafer demand.

And one example of these kind of products is shown on the right. This is the next-generation Venice CPU from AMD. And what you'll see there is CCD is basically the CPU core. They combine eight of those together with two I/O die. So they use different nodes, cheaper node for the I/O die, which has bigger area and the most advanced node for the compute die. And then they have some additional chips along the side. So altogether, there's 18 chips that we have to place on to that one unit. So for 1 CPU, 18 bonding steps, plus we still have to bond that wafer assembly onto the substrate, so 19 bonding steps.

So it's easy to see how this drives a lot of demand. And we've developed a very high productive machine for the specific purpose that Christoph will share with you. And then we have the mobile segment. So mobile is projected to decline in terms of the overall market this year. Nevertheless, there's always opportunities where we see technology transitions happening in this market. So one example would be with the application processor. Typically, this has been a vertical structure where the memory is attached on top of the logic device. But because of the increased need for memory and AI-enabled phones and also the increased power dissipation of the application processor with more neural processing units and things like that inside the chip, there's a transition to the side-by-side structure. That offers opportunities for molding for Die Attach and a lot of the different process steps we have at Besi.

Other configurations are wafer-level assembly, which take advantage of the same capabilities I just mentioned earlier. And then in the mobile memory side, we typically haven't played there at Besi, but what's changing now is that the demands on die placements are increasing, meaning that more accurate chip-to-wafer assembly processes are needed in order to create these new kinds of mobile memory structures. And there, again, our equipment comes into play.

And then finally, smart eyewear is kind of a new device type that we see opportunities in. Why? Because it requires very high miniaturization, high accuracy placement. And then in the case of the glasses with display integrated, you need to have a light engine that can also take advantage of hybrid bonding. So still a lot of opportunities there.

And then finally, in our core business, the -- yes, there's growth still happening in automotive, starting to come back. And yes, we see, for example, in the powertrain and electric vehicles, silicon carbide, gallium nitride type modules, being used in ever more complex structures. And we're very much engaged in both the Die Attach and packaging of these kind of structures. Same with data center. So I talked about all the advanced devices in there, but there's a lot of power management in data center. You don't want to have waste in your power management, right, because it's very power hungry. And so there's a lot of unique packages that require things like diffusion bonding where Besi has a leading product offering as well as high precision molding.

Okay. So that's -- I think I'm a little bit over time. So I would just like to leave you with a few things. One is that all these opportunities are really exciting for us. We've been preparing for this for a long time. And I think it's a great opportunity as the market is taking off and growing for us to capitalize on it.

I think I'll leave it at that and turn it over to Peter.

All right. Welcome, ladies and gentlemen, to our this year's CMD presentation. So my name is Peter Wiedner, and I'm responsible for the Sub Micron business. But the ones of you who are following Besi since a while and also participating here, you might more perceive me as Mr. Hybrid. And I can tell you, I had a great last year since our last meeting because wherever I went, people said, you're the right guy, I need hybrid. And if I would recount all the meetings that I had from the last year here, we would be still sitting here tomorrow.

So actually, I want to start with a quote out of a summary from the ECTC '26 just roughly a month ago in the United States, one of the most important semiconductor conferences for that arena, which is saying in the summary, actually memory bandwidth, chiplet to chiplet latency, synchronization overhead and energy per bit now shape overall efficiency in multi-die AI architectures. And bottom line, hybrid bonding directly attacks all those bottlenecks.

And it's also kind of a summary to the explanation, the technical explanation that Chris has given in his section, but that's really the summary of what's going on in the market right now. And we at Besi, we have been a very early believer and for that matter, early adopter of hybrid bonding. And for that reason, already years back, we have brought the very first hybrid machine on the market, the one in the left upper box with stated #1. And this was a 200-nanometer machine. And that was the first machine where hybrid bonding in volume manufacturing did start for devices, for logic devices with roughly 9-micrometer bumpage.

Our second generation, which is the actual generation today, the second one, with increased accuracy, 100-nanometer and relevant for logic devices for 6-micrometer bumpages is the industry standard today. And I'm very proud that also this machine will be the starting point for high-volume manufacturing in memory business, in HBM business. For sure, in HBM, as you are aware, the bump pitch and with that the accuracy is a little bit more relaxed. But in difference to the accuracy, the productivity matters much more. So that's the reason why in the gray bar, we are counting on the productivity or the units per hour that the machine can produce.

And while we are doing that and helping our customers to ramp up their productions, we have been developing already the next generation in the meantime, which we call the N50, N50 for 50-nanometer, which will make it -- will be ready for logic devices with even smaller bump pitches in the area of 3 micrometer. But -- and that's very important. At the same time, this machine generation, where we will start actually the beta test at the customer site this year, will, at the same time, not only be more accurate, but also way faster, which is once again very important for the HBM segment of that market.

And we are not done with that. In parallel, while we start launching this machine, we are already working on the next generation thereafter for even higher accuracy and more speed. And this is why I'm showing you this because this is a unique road map and a unique product lineup actually in a very short period of time that no other competitor of us is matching. While we are still having all these experiences and generating -- improving generation by generation, they are still fighting to make their first inroad. And that's really our big advantage and our strong competitive position.

But that is not all. Because that's what I've shown you, that's the bonder. That's one step in this whole hybrid chain of manufacturing. But as you are aware, there is a lot of machines involved and also the pretreatment of the die and the wafer before it gets to the bonder is a very critical part for a successful hybrid bond in the end. And for that reason, we have teamed up a long time ago with Applied Materials. You are aware of that. And together, we can offer not only individual machines, but really a complete system, which is doing the pretreatment plus the bonding and by the way, plus the measurement after the bonding, so the quality assurance.

And that's the integrated system that you see on the bottom here, which is our shared AMAT and Besi system, which is called Kinex, and why is that important that this is an additional offering as well because the more chips you have to put together, and you heard some examples of Chris, where this is happening, where the packages are getting more complex, the yield is even more difficult to achieve and an integrated system simply can help you to maintain a high yield even in a complex package. And that's the reason why that one is very important as well.

Nevertheless, there is one more aspect to our partnership that we are having with Applied Materials, and that's going a totally different direction. And I would even consider that perhaps even more important than the product offering. We have been since ever we have been engaged with Applied Materials, we have started together a joint development activity, and we call it the center of excellence in Singapore. And for sure, that lab is equipped with everything, our own integrated equipment, two pieces and five bonders, all downstream and upstream equipment that you need because the aim there is really to do research on the bonding -- not on the bonding, but on the hybrid process itself because it has a lot of ingredients. It starts with the material, with the layers that you take there, once again, the preparation and you all need to understand that to make a solid hybrid bond in the end for a volume manufacturing.

And this lab is utilized by us to learn that, to understand also to research to come up with even better hybrid processes. But at the same time, we are engaging in this lab with our end customers who want to do hybrid bond and help them to develop their packages with hybrid bond interfaces inside. And so far, we have been engaging together, Applied and Besi with over 25 customers in our Singapore lab. And you see a nice correlation, by the way, to the Richard's chart where we already have now 20 active customers who have bought machines from us. And that is also driving actually the hybrid adoption into the market much faster with our help and our activities.

Now talking about customers. Let's go to the market. And let's see what happened in the last year in the market. And let's start for that matter with the Logic side. And a lot of things Chris has explained and I can summarize them here a little bit.

So what we definitely do see on the Logic hybrid side is that a lot of use cases have been materializing throughout the last year, and I'll get to that on the next slide. But if I would sum that up, you could say two things. One thing is that hybrid interconnect technology for AI logic devices has become and is confirmed to become the predominant interconnect technology. You can name it, AMD, Intel, NVIDIA, then also the Meta, Google, all the ones who are doing their own chips, and you saw also one example of that in Chris' presentation. So that's very clear, very obvious straightforward. And perhaps also looking a little bit back in history, nobody would have doubted that this is the primary area where hybrid bonding does make a lot of sense.

But I also do remember in some of our previous presentations like that, I got the question, well, what do you think? Will hybrid also enter, let's say, a consumer market, for example, into our laptops? And that was not so easy to answer back in the days because even so the very first hybrid device from AMD was a consumer device, it was a graphics processor. There was not really a lot of activities visible for follow-on. But now this year, with the launch of Apple's M5 high-end processor group, which you, by the way, can already buy now at the market in a laptop, it's very visible that what we thought all along that on the high-end segment, hybrid will penetrate the consumer market that this also has been happening throughout -- yes, in that case, it was -- the announcement was this year.

And once again, coming back to some of the use cases, I want to use and to show you the latest ones. I want to go back to a chart that we have been showing you two years back in this meeting here. And back then, this chart was in Chris section, technically trying to tell you how it typically -- the development of a processor evolves from an SoC system on chip, so one chip to two or four chips, which we call it back the split die in order to get more compute power, but without hybrid interconnect, all the way then to the chiplet architecture where you then use hybrid bonding.

And back then, two years ago, there was only the EPYC from AMD, which you see on the right upper side, which was confirmed and in production. But we said, well, that's a pattern and that will happen to all of them, one earlier, one later. And only Intel back then had announced their internal Clearwater Forest architecture. So that's not an end product, that's an architecture that they are working on that. But it didn't have a sellable product. And really, what happened in the last year was that all of this, what we predicted two years ago came together. So Intel finally launched a real product based on their architecture of Clearwater Forest, Xeon 6+. Then as I mentioned before, in the consumer market, Apple launched the M5 Pro and the M5 Max. And finally, and that was also very often a question, well, what will NVIDIA be doing? Very interesting because it's obviously one of the biggest suppliers. They don't have a product here today, but they announced with the Feynman architecture and product lineup that they will use that as well. So it's very clear on the logic arena to sum it up once again, adoption, the use cases are there. They have been announced. You can already buy things today, and that's just moving along and the adoption rate will increase.

Now let's move from Logic to another segment, which was way more discussed over the last years than the Logic arena, and that was the memory, the HBM. You remember the times when said, yes, hybrid will come and then there was the height topic and yes, you can make higher stacks and everybody said, with the higher stacks, you still can use TCB. And that's not good for hybrid and all these discussions in the past, but if you look at what really happened now and once again, finally this year is that for the pure technical benefits that Chris pointed out, and I just wanted to sum it up in one sentence here for -- because the hybrid enables a lower power and a better heat dissipation, and that's up to 30%. That's such a big advantage that the manufacturers of HBM can then turn either into faster memories or into more memory in one stack that everybody wants to use that.

And also in addition, and Chris explained it already, this custom HBM with custom logic underneath, which is driving the bump pitches down to a smaller level, and we predict roughly around 12 Micron, which is then much closer to a hybrid interconnect than to a TC interconnect. Also that is speaking a language, the same language. But once again, that's just a technical summary.

So what did happen last year or in the course of the last 12 months. So finally, all three leading suppliers are seriously doing research and seriously evaluating hybrid bonding for HBM memory. And by the way, they all do it with our machines. For sure, some started earlier, did start earlier, some a little bit later. So the early adopter or the front runner is already very far in his research, and you can read his announcements also. He is not -- he's also very proudly announcing that. And it's very clear that the first one will adopt the hybrid interconnect in HBM 4e in '27. And it's very visible well, to the outside world, but also internally because he's already starting preparing his factory for high-volume manufacturing. So for mass production for sure in advance so that in '27, he can start to deliver. And that's a very clear signal and a clear message that the question, if hybrid will come for memory, yes or no. That question is answered, it will come.

So now the matter of fact is, well, how fast will it come? And for sure, you can have different models to that. But on a higher level, one thing is very clear. Once again, why do you want to use that because you get a better performance. Now every new HBM generation is defined by better performance data. That's the reason why there is an HBM 5 and then a 5e because it always gets faster and more memory. So that means with every successive HBM generation, you will need -- you will use more hybrid interconnect because it simply helps you to achieve, yes, the target data and the target performance of the HBM Die.

And once again, starting with HBM 4e is the start is just around the corner. So also there, very exciting and great news. But I have to tell you, the most exciting one is the CPO topic that was already mentioned. Because if you remember our last year's presentation, I had in my section a slide where I was showing actually also that picture and a little bit more on the technology side and saying, yes, this co-package optics, there is some advantage and packages are being developed and TSMC is doing this COUPE, they call it COUPE. And so -- but in the end, if you look at our last year's presentation, it was really more telling you an R&D is ongoing in that arena. And now within one year, we are from an R&D stage to volume manufacturing. So because, as Chris pointed out, you can already buy from NVIDIA, the Spectrum X or the Quantum X that's and that's -- he showed the picture as well, you can already buy that today. So it's already there.

So in one year, in no time from zero to hero, that's even for our fast-paced semiconductor industry, extremely fast. And let's have a look at it, perhaps why it went that fast. And what's behind that, not technically perhaps, but in general. And I think there are two major factors. One factor is there are these undisputed advantages for the end customer, not for our customer, but for the end customer who is using the interconnect because as you can see from that conference slide from TSMC, if you take greater power efficiency and 20x lower latency. So it's -- you take less power and you get a higher speed at the same time.

Thinking of the huge effort that the data center today has and the power consumption, there is a huge drive really from the one who is using these devices that he wants to have that. So that's always good if the end customer wants something. So that's giving a big drive number one. But the second drive that is also very driver, which is also very important is the people who have designed these packages. Once again, COUPE as an example, as the most important example, they did a very smart job because they designed a package, which is using a lot of technical ingredients, which are already there today. And that brings that package to a great scalability and a very fast scalability. And if you just take that one interconnect that we are doing with hybrid here, and as you've seen, there is more to that. But if you just look at that one, you do that with the same process, with the same machine like you do a logic hybrid.

So that means you don't need to develop something in addition like you need to do for memory, for example, because the stacking of memory has some special challenges, which need to be addressed in R&D. You just take the same machine, you put that device on it and you bond it done. So that means the ease of scalability is also the important ingredient, which allowed this fast adoption.

Now when I say it's in mass production, and if you look at the right chart here, which is a prediction of forecast, it's saying 0.2 million devices this year. Well, it's the starting year for that device, so no surprise there. But comparably to what is predicted four years ahead in 2030, that's a very low number because the prediction of these market forecasters is that this will be a market of around 60 million units a year. And that's a huge growth rate of a CAGR of over 300%. And this is -- and for that matter, the speed of HBM and also the size that can be achieved with that specific market segment. That's the reason why I'm saying that's for sure the most exciting development for hybrid in the last year.

And with that, for sure, we have also increased our market model that you know every year, we are presenting that. And last year, I had a little bit of CPO in, by the way. So it's not -- it was not zero, but it was very modest, and I also had it in the high case because it was not sure when it is coming, how much it will come. So for that reason, you don't put it in the low case. But now a year later, I can easily put it in the low case and for sure, with much higher numbers behind. And for that reason, the low case, so the logic with all its adoption plus the CPO, we can increase that for 18% versus our last year model.

Then like always, we are adding the memory part to that in order to come to the mid-case. And if you will compare, you will see I did not change -- we did not change that much on the overall, let's say, number of machine assumptions there. I think that because we always have been positive that this is coming, so we had that modeled in already. However, what is the difference is the confirmation that it will come is this year clear why you could have had a question mark last year. So that means also the mid-case is confirmed.

And then -- and I did not talk to any of these subjects or topics in detail. We are adding up the edge devices like smartphones, processes for smartphones, but also smart glasses in order to come to the high case, which then increases overall by 12%. Now just a comment to these edge devices. Once again, I did this year not prepare a presentation for that and spend the time. But I did show you, for example, a lot of activities on the smart glasses last year. Just let me assure you, the R&D activities with the end customers of that is ongoing. We are still engaged with that, but that's a little bit more of a development work to be done by the end customer. So that's taking a little bit more of time. That's the reason also why the high case curve is not kicking in this year, but only a little bit later because these devices still need some development time until they are getting to high-volume manufacturing.

So all in all, a great increase up to a cumulative 2,200 machines and also the low case up, as I said, up to 1,150 machines. So all in all, as I said, I had a great year. But I'm not only responsible for TC -- for hybrid, also TC. And we are also doing a lot there.

And I want to give you at least a small overview of TC as well. We do have an offering not on the basic TC processes, but on the so-called fluxless, which is, let's say, the most modern style of thermocompression process. And we did expand our adoption last year -- over the last year to five customers, which includes customers for memory and CPO manufacturers. And the interest in this fluxless TC interconnect is really increasing, especially in the last three months. We get a lot of -- yes, customer requests. And with a new technology, you get a request for all kind of market segments in the end. But I want to point out four relevant market segments for this future technology.

And that's on one hand, memory. And for sure, today, TC is normal in memory, but that's the NCF, the standard thermocompression process. But we are engaged with one memory supplier to develop a TC fluxless interconnect package style. And if you now think -- but Peter, you just told me four slides ago that goes to hybrid. Think of it, I said not everything will be hybrid and don't think only of HBM memory. There is more different memories out in the market than only HBM. So you have to think on the whole memory market where this may play a role.

Then for sure, the CoWoS, also mentioned and explained in detail by Chris, which is today flip chip processes. But when the CoWoS tiles are getting very big, you need to switch from flip chip to thermocompression because of the size. Once again, we need to consider actually the optical market with the CPO, but also the transceivers because also here, these are very complex packages and the transceiver has up to more than 10 product interconnect steps that you need to do. And also here, some of them are poised to become TC. So that's an interesting market.

And ultimately, even so I agree that, that's farthest out, but we are doing R&D with certain companies who are engaged with that already. And that's for, let's say, I don't know if I can call it a semiconductor then, but also these quantum computing, let's call it, tiles need to be packaged in whatever in a way and need to have electrical contacts to the outside. And unfortunately, you cannot use, let's say, usual materials like copper, you need to take very special materials like Indium, which are hard to make them connect. But that's also some arena where fluxless and inert gas TCB machines play a role, and we have already made samples for that -- yes, on an R&D stage. So these are the four, let's say, most relevant segments that we are looking at when it comes to fluxless TC. And my counter is at zero here, and I'm at the summary slide, so that fits perfectly together.

And so you can -- we can have a break and a coffee now. But before that, let me just finalize it. So we have a perfect product lineup for hybrid. On the market side, the use cases for logic has been confirmed. On the memory side, the implementation, the first runners are around the corner. And as a third one, the CPO has developed in one year into volume manufacturing and all these positive signals are driving up our market potential. And for that matter, we have also increased our market forecast.

Yes. I think with that, I can -- well, not close the session, but with that, we can go to the break. And after that, Christoph will continue with the mainstream business. Thank you.

[Break]

All right. So welcome back, everyone. Yes. I hope you're still with us after the break. I'm Christoph Scheiring, Senior Vice President for Die Attach, and I'll walk you through the main updates for the mainstream Die Attach, as this is my responsibility. As my colleagues, I'll focus on the changes. You're following this Capital Market Days for years already, so you don't need the full story, only highlight what has changed over that last year.

So the message for today regarding the mainstream Die Attach is clear, Mainstream Die Attach is no longer only driven by the traditional semiconductor cycles. It is increasingly supported by structural AI growth. That's the clear goal to make that transparent for you. What I will do is I will show you the drivers. I'll talk a little bit about technology changes and certainly about the product portfolio and the positioning and why we believe we are able to capture those growth opportunities.

And I'd like to start here with a chart showing the end markets for the mainstream Die Attach and talk about the scale of the AI impact. So basically, what we see is strong growth expected across various segments. And this means it's not just about compute anymore. It's about the complete infrastructure that is driving growth from Photonics to power applications and others. On top, we see also a second order growth vector into smartphones and AI devices as those units or AI models are increasingly being performed on those devices at the edge. Altogether, means an increase in advanced packaging and more accurate, more complex, more sophisticated die bonding tools. And this is, at the end, where Besi plays a main role.

China, important. China remains a demand pillar. This has to do, especially with the effect that advanced packaging is even more relevant in markets where you are not constrained, right? So what does that mean for the Besi, for the machine, the equipment market?

So what we see is that we are, over that last year, the Die Attach intensity, as I call it, is rapidly increasing. So if you look at those AI systems, what you realize is that we have more packages per system and even more, if you look into the package, there are more dies that need to be placed package. All that creates an additional structural AI demand on top of the traditional semiconductor cycles.

So because of that, we have updated our model the addressable market model to a EUR 1.6 billion level by 2030, which is an increase of about 8% as compared to the numbers I have presented last time. And what's also clearly visible is that the growth is concentrated on the advanced segment, which again plays well into the strength of Besi.

Talking about the portfolio. We are having a known slide. In the upper part, you see the volume drivers, the Multi-module attach machine as well as the Flip Chip lineup. In the lower part, Epoxy and Soft Solder. So what we see is that the growth is not evenly distributed. Obviously, stronger growth is seen in the upper part for MMA and Flip Chip, while the lower part, the Epoxy and the Soft Solder remains important, no doubt about that, but it is a more moderate growth. Competitive positioning remains strong for all the segments, whereby the Besi strongholds are always via accuracy and process capability.

So if you look at the big picture, the most relevant applications for us going forward are here in the Flip Chip part, the 2.5D CoWoS market. For the MMA, it's clearly the Photonics part. And a bit scattered across the platforms, it is, yes, smartphones or in general AI devices that are driving the business. And those are the three applications I'd like to dive into more detail in the subsequent slides.

And I'll start here with the 2.5D market, which over that last year really has emerged as one of the major growth drivers for us. What we see is -- sorry, what we see is a strong growth going forward of about 50%, same as Chris already has shown, driven by clearly AI data centers. On top of that, we see modules growing very, very large. And because of that, a technology transition to so-called CoWoS L formats or architectures which is seen as the dominant architecture going forward. This CoWoS L is based on RDL interposer wafers, which require additional die bonding steps to place so-called bridge dies in order to create local interconnects, which means additional die bonding steps. This, together with the increased complexity, has a very strong and positive impact on the number of , yes, die bonding steps per system, and at the end this creates a driver for equipment demand.

In more detail, is a slide that explains this equipment demand or that driver. If we look at the next-generation accelerator package, like shown here, we basically start with an interpose wafer with only a handful of components of units on it because of the sheer size of those modules. And then going into the module itself, you see there is a number of chiplets that need to be bonded onto one or two or even more bottom dies. Those dies later on go onto the interposer wafer. But that, together with a larger number of surrounding dies, HBM dies as well as chipless, then some dummy dies. And as mentioned already, we see also quite a large number of these silicon bridges that need to be placed and those are the dies that create the interconnect later on. And at the end, everything needs to be bonded into the package substrate.

So what we see, while we had in the past, a single-chip process flow. Now we look at very complex architectures with several layers of die bonding steps needed, in this case, it's about -- you can count it, it's about 40 different steps, whereas one step in the past. And this obviously creates a lot of demand for us. So Besi is engaged and aligned with the needs on all those layers. We start at the upper end with our Hybrid bonder to do the most advanced bonds.

We have introduced new machines for the middle layer, our 8800 Chameo Flex for the mass reflow. As well as for the highly accurate Die Attach steps that are needed in the stack, we have introduced our 8800 TC Next, for the TC related Flip Chip bonds on to wafer level, as Peter explained in his presentation. And we have at the bottom here, an important platform, which is tool of reference for the interposer attach into the package substrate. So quite strong presence in the market already.

So clearly, from a mainstream perspective, this platform, Chameo Flex is the most relevant development over the last year. That's why I'd like to show you a bit more in detail what it is and what it can do going forward. So it's basically building on the success of our leading Chameo Advanced, which is the tool of reference in today's CoWos like markets. But it addresses additional markets such as Bridge die, which I explained, but also panel level applications. It's, yes, designed to provide best-in-class productivity on the one hand side, together with 1 Micron accuracy. And these are basically the key ingredients needed for next-generation 2.5D structures.

Over the last year, we made enormous progress. We have launched the platform. We have started engagements with several foundries and OSATs and are qualifying the system. OSATs are specifically relevant going forward. As you may have heard, TSMC decided to outsource their internal chip-to-wafer operations to their OSAT partners. So also that portion will go to OSATs. And as you probably know, Besi Mainstream has a well-established relationship with those OSATs and many of them are using Besi tools. So that's a good news for us.

Importantly, we have already been able to complete one of these qualifications, and I'm expecting the first purchase order any moment. So you clearly see that platform is developing rapidly from product introduction into a commercial validation. And we believe that with the help of this machine, we will increase our share of wallet in this 2.5D market and capture additional markets, such as the Bridge die attach.

I -- with this slide, I'd like to turn it then to the next market to Photonics and specifically to Transceivers, those units here, which basically sit at the end of every fiber that goes through the data center from rack to rack and so on. Those Transceivers are -- because of the data center build-out are seeing an enormous growth.

Indicated in the chart here, '25 at a number of 100 million units already and going forward, showing or expected to grow significantly overall with a CAGR of about 30%. What we also see, besides the market growth, we also see a transition to 1.6 Tera Transceivers, which are faster, higher transmission speeds. And this is, and I will show that in a minute, increasing the Die Attach intensity, which again is a positive factor for us.

And lastly, we also see new units coming up, so-called LPOs, Linear Pluggable Optics, which is a response to the ever-increasing constraints in AI -- power constraints, I should say, in AI factories. Those LPOs are from a packaging perspective, very similar to Transceivers. However, their assembly requires higher accuracy, which is again positive news for us as this is our stronghold.

So Transceivers represent an enormous growth opportunities for us already on an already high level, which we are seeing at this moment. Besi is one of the leading suppliers of assembly equipment for that market. And our 2200 Evo advanced is the real workhorse there used by all of the leading suppliers in this market that are listed here.

So going forward, what we see, first, clearly, the market expansion. Secondly, the transition to this 1.6 Tera Transceivers, which is increasing the Die Attach intensity and a graphical representation of that can be seen here. While a typical 800-gig Transceiver currently used in many of the data centers coming from Innolight is comprising 4 lanes of 200 gig each. We see in the next generation, the 1.6 Tera, the number of lanes is doubling. And again, that also means that the number of attachments per Transceiver is almost doubling.

Basically today, and I tried to indicate that here in red, is already covering quite a number of different process steps. I don't read it out here, but it's a few of them. On top of what I just explained, we are targeting with a new development, Evo ONE, enabling higher accuracy as well as Eutectic bonding. With that, we are addressing in the future even more steps, extending the scope as well. So basically having an increased scope, more Die Attach intensity and this in a market that is exploding, as Chris said, is a clear, demand driver, a structural demand driver for Besi. And that's why we are excited about that Photonics market.

And lastly, let me turn it from AI infrastructure to AI devices. AI is increasingly being executed on at the edge on those devices. The market itself expected to grow significantly also. The main variant, the dominant device for the time being is smartphone and will be smartphone. However, others like robotics, automotive or AR/VR are emerging as well. And the one thing they have in common is that those AI devices are rich in semiconductor content. That basically means more advanced processors, more memory, better memory and also sensors are typically also being better and more complex.

From a packaging perspective, that means that, yes, we are looking at integrated functions on more complex functions on tighter accuracy requirements, which plays well into the strength of the Besi portfolio, namely our Evo, which is used in smartphones, but also increasingly in AR and VR devices into our Epoxy platform, which is heavily used in the sensors, in power management, power amplifiers for smartphones.

And lastly, in the Flip Chip portfolio, which is used in the processor packaging, in the memory packaging and recently being introduced even in the camera package. Talking about camera, we are also seeing camera innovation at an inflection point. We are seeing the introduction of new functions. And since we have been able to qualify our tools again as tool of reference for those applications, this will add to the overall growth story.

So in summary, and you can read it out, we see a structural driver coming from AI that is also impacting the mainstream business in -- across a broad range of end markets. I think that's very important and also impacting various basic products. We see that this increased Die Attach intensity is a multiplier on equipment demand, as I have pointed out for Transceivers, but also for 2.5D structures. Because of all that, we see our addressable market growing by a factor of 2 until 2030, which is an up compared to what we had in our models last year of about 10%. We are very well positioned in those highly growing markets like 2.5D Photonics and Edge AI devices and our product developments, which we have launched and are getting track now and are helping us to accelerate the market growth and help us hopefully also to increase market share.

That brings me to the end of my presentation, and I hand it over to Richard for the summary.

Thanks. So a lot of information. I hope it didn't surprise you. It's all following crystal clear road maps, but the adoption rate always varies. But let me go to the summary. So as we shared this morning, an upgrade of our target model, new target, EUR 1.1 billion Engine 1, so crystals part mainly, but also adding Die Attach and several plating solutions to that. And then we have Peter's Engine 2 Submicron. Submicron up from EUR 0.5 billion to EUR 0.9 billion to EUR 0.6 billion to EUR 1.1 billion, simply as explained because of continued adoption in Logic, the range because of expected adoption in HBM that moves the needle either faster to the upper end or in a somewhat 1, 2 years slower trend, but it will move to that upper end, we're convinced, but also other developments, the chiplet architecture.

And one of the key factors which many of you ask us is can Besi maintain its leadership position in this submicron world because what we haven't said, I think somebody counted there are over 10 Hybrid bonders announced in the market so far. How many, Chris, did we add up in total? Yes, yes. Well, there are some Chinese as well. So -- there are 2 then questions which arise, why is everyone developing a hybrid bonder?

Well, there's only one clear answer. In the near term, but certainly the long-term future, the direct copper-to-copper way to connect devices will become the mainstream. So that opportunity leads to all investments in these, whatever, 20 hybrid bonders. But then the question, how can Besi maintain its leadership position?

Well, we have explained to you several things. We were there very early. I don't say the first because Shibaura was also there. But we've been very successful so far to understand clearly what it takes and translate that into production tools. The same time, on the road maps of the key customers, early adopters of Hybrid bonding. But then the process know-how, which we built with Applied Materials help over the past nearly decade, seven years in our formal relationship. And as Peter explained, what we do in Singapore, what we do jointly in Singapore, but also in Radfeld coming up in the U.S. is second to none. And that gives us a sustained engagement with customers in developing of all kinds of new processes.

Whether we have the best bonder is still the challenge. It looks pretty good today, but you have to be always sensitive, careful, is somebody else building a better bonder. But anyway, that's our entrepreneurial life blood, which definitely will take us further.

So that's key to the upgrade in the model. And why did we increase the operating margin? Very simple. If you look at the success of advanced packaging and then not only Submicron, but also the part in Engine 1 is moving to ever more complex devices. And if you do your homework right, you have higher margins. And that you will see translated in the end into higher operating margins. And that's the reason why we instead up a bit in line with our strong markets development in the past couple of years, but certainly at this very moment. So that's the new model.

And then if we take all these messages, many have been touched upon already. you see a significant increase, however, words you choose, due to AI infrastructure build-out, and that will continue over the next couple of years. You may argue, if you look at this industry in a historic perspective, it's never a straight line. There may be, let's say, periods where the adoption takes some time, building new fabs takes time, building advanced packaging facilities takes time, not only to build, but also to train people. But more importantly, end markets always take time. So that's also why our nature has always been a bit on the conservative side.

But if you look at the drivers, they are enormous. We are at the very beginning of a significant up cycle driven by this whole AI change in the world. So we're in an excellent position to benefit from that. Second comment is in line. Currently, you can say there's a deficit in capacity. Besi is well positioned. The new up cycle, as we explained to you, began in the second half of last year. I won't repeat the anecdote of our close to EUR 600 million revenue last year. That has surprised, by the way, everyone in this market. It's very interesting to see that phenomena time and again in this industry. It's conservative in a way. But on the other hand, it's most leading-edge technology you can imagine. But the timing of these cycles remains very, very difficult.

And we are moving ever more into, let's say, the decision arena at customers, also helped by Applied Materials. And although the big picture is very clear, the immediate demand changes are often, let's say, not according to a clear schedule. But anyway, it's turned second half, and you've seen that in Besi. As we announced our first quarter numbers, the guidance for Q2, fourth quarter already. That definitely proves that we are in the middle of that. Significant progress achieved, multiple drivers converging. I hope that was a bit more clear this time on what's happening in this hybrid bonding arena. And it's a lot where we are in the core process development and in many cases, process of reference, increasing our targets and adjusting our operating model.

Although everyone takes for granted that we have gross margins above 60%, many ask, why is it only 60%? Are your targets high enough?

Well, that's also why we increased the targets. I can share with you, if you would know all what goes wrong in our company, there's still a lot to improve. So those targets increasing is a very natural way to manage the business.

So with that, we'd like to conclude what we have prepared to share with you, and we would now like to open the audience for some questions. And Peter, Christoph and Chris, why don't you come forward, take your chair along, and we'll be happy to answer your questions.

Who wants to sit where? Did we make -- please take a seat. Peter in the middle, you're the most important.

You're the boss.

Anyway, who has the first question? May I choose on the front row, please.

One question I have is about your capacity. You talked, I think, on one of the slides that you're going to increase your capacity to 35 tools per month, if that is what you're saying. Is that going to be sufficient to supply this demand that you're talking about by 2030? Or is there going to be a continuing increase in your capacity to be -- in terms of being able to supply the market?

And my second question is, clearly, I mean, now there seems to be -- you seem to be closer to volume adoption of hybrid bonding in HBM. In terms of what needs to be still overcome to go from here to high-volume manufacturing? Is it something that you need to do or is something customers need to do? So what are the steps between now and, say, next year or whenever that high-volume adoption takes place?

Well, 2 excellent questions. Number one, if you do a quick math, we are now at a level of building 300 bonders per year. So theoretically, if you multiply that by 2.5 million as an average price and you look at our model, it's close, but we also have shared already that the next phase is already prepared. So this capacity expansion we do in steps, this should satisfy the first part of the 5-year model. But then once it takes off much more, we are able to expand with locations already in place.

So one of the simple comments is Vietnam. We have started with high-precision tooling, and we've also shared by the end of this year, we will build the first bonders in Vietnam. They move out of Malaysia. So we simply have more space in Malaysia, is as easy as that. So don't worry, if that accelerates, we are certainly able to address that. And customers on an ongoing basis come and test our capabilities because that same question is on their table and is busy, ready to deliver once that is required.

Your second question, what does it take high volume? We mentioned in one of the slides, we are expanding our support footprint in Taiwan. We are also expanding that in Korea. In the U.S., we have expanded it significantly in bringing Intel up to speed. And that requires a service support along a front-end business engagement model. And that has taken us some time. We had enormous help from the two big customers who have a certain mainstream volume. That is now in place, the people, different people, training levels.

Also, what we mentioned is the spare part infrastructure that's completely different supporting a front-end fab compared to a high-end assembly facility. So in a joint effort also with Applied Materials, we have the most brilliant benchmark available that we bring this ever closer to front-end requirements in the market. But so good so far.

Next question behind you.

Francois Xavier Bouvignies, UBS. So I have 2 quick questions, sorry. The first one is on the road map for hybrid bonding. So you mentioned the 15-nanometer another 25-nanometer. So when are we going to see the 25-nanometer tool? And I don't want to look too greedy, but what is the road map beyond that? I mean, how far can you go in the accuracy for your tools do you think, maybe in the more 5 to 10 years view, if you have any road map would be great.

And secondly, you mentioned as well the accuracy and the throughput improvement that you are delivering despite this improvements. How should we think about the pricing of that? Because if you have a massive unit increase throughput plus accuracy on top, I mean, I guess it gives you good pricing power here. So you mentioned 2.5 million average ISP. I mean, how should we think going forward given the big advantage you provide to the industry?

Well, I'll start with the last question with the pricing because that's easy to answer. The effort of developing ever more accurate machines and that might be a bonder or ASML for EUV and so on is exponentially increasing on the R&D effort side and the bonders are getting more complex and the machines are getting more complex. So the price level is definitely going up with each increased step of accuracy. So that's without any doubt. And that needs to be considered if you are looking forward.

On the accuracy, for sure, the accuracy increase will move on, and we will see and we will adopt to what our customers are needing. Currently, the customers are telling us that they are targeting and depending on which customer you're asking, bump pitches of 1 micron and some are, say, 0.8 microns. So -- but it's in the ballpark, a little bit below 1 micron. And that can be addressed with our developments that are currently ongoing, which will be there in time when the market needs it before the end of this decade.

And -- and then we need -- not then in parallel, we are always watching out what is the next step of the customer. And if they say they go down to 0.5, then we will simply follow with our road map to that. So not everything in our industry can be already predicted years ahead. But in any case, as accuracy has always been and independent of this Hybrid bonding or any other bonding step has always been a very vital and important part of our DNA, we are anyway having our pathfinding R&D departments that are working very generically on how to improve accuracy even further in die placement machines.

So that's what we are anyway do independently. So we will be prepared whatever comes along.

Yes. And to add to that, in very close cooperation with the key customers, as Peter said, the engagement in their road map is what it's all about. So they regularly test what progress we are making. And you must realize that the timing, and we've seen that in the past, is also not very clear to the customers, even the largest customer on this planet. So we, as a supplier, have to switch gears unexpectedly faster and sometimes also somewhat later.

But anyway, to underline, we are always connected to the most critical road map for accuracy in the market. And that's in Logic. It's not in HBM. Memory is always a bit less critical, but still very critical. And that determines our long-term future. So your question is very important. Thank you. Next...

I just have on the CPU front. So you talked about like very encouraging to see Intel Clearwater Forest announcements recently and AMD has obviously adopted Hybrid Bonding as well. So I'm curious about your penetration in the ARM CPUs, so the NVIDIA Vera, Axion, Google and AWS Trainium and the propensity for them to adopt hybrid bonding as well.

And then I guess the second question was just kind of following up from an earlier question about the level of service and support that you're giving. Do you see that your service business could accelerate faster than the group revenues over the next few years given the intensity is rising? Or do you kind of see it rising with group revenues?

Chris, the first question?

Yes, I can address the first question. I think for all the ARM CPU suppliers, none of them yet are using hybrid bonding. That's true. So our involvement there is supporting the current processes that they do use, which tends to be CoWoS, for example. Sometimes in the case of, I think, the Amazon chip, it's just chip on substrate multichip module. But in any event, our equipment can support those process steps. The question is when do they convert to Hybrid Bonding. I think the good news is that ARM has a design toolkit now to enable that for their customers for both CPUs and GPUs. And that was only done, I think, maybe 2 years ago and still being designed into. So I expect to see that, but I don't know -- I can't tell you exactly when.

And the second question, Peter?

That was about the service revenues. Service revenues are definitely accelerating as we speak, simply because of the reason that -- and that is coming back to what Richard already said that hybrid is placed in the front-end area of our customers and not like we are used in the back end. And there is simply -- in the front end, there is simply a different style of working with. And yes, it reflects on spare parts, but it also reflects on the service because in the back end, you normally only are going to the customer if you ask for a specific service, while in the front end, typically, there is an ongoing presence of the supplier, and that's for sure then covered with service contracts.

So yes, definitely, the service part is accelerating as we speak. And, and it is growing along with our engagement in the front-end arena.

Next.

Simon from Barclays. Maybe it's a relatively simple question, but we get a lot of questions from investors on how to think about Engine 1. You -- we've seen the small increase or nice increase for the guidance for that. How do you think about the mix of Engine 1 in your target operating model? 50%, say, of the group today is computing. So is it even bigger in the future? How do we think about spares and services within that? I think that would be very helpful for investors just to picture how that grows from here given it's implied doubling, I'd say.

And then second question would just be on Photonics. Last year, you gave us, I think it was a 60 Hybrid Bonding unit TAM for Photonics. You sound very bullish about it. It's clearly a big opportunity for you guys. How do we think about-- how you think about that TAM, but also the dynamic between some customers will use Flip Chip for a while and eventually shift over and others are already using Hybrid Bonding today? That would be very helpful as well.

Christoph?

Yes. So your question about the Engine 1 split and the driving market. What we see is basically, as described, a strong push coming from the AI, which is going to compute 2.5D CoWos as the main driver right now. But the Photonics portion is seeing an increase, which will reflect in higher shares of Photonic business for overall Engine 1. At this moment already very relevant, but expected to increase over time in the next years. That is a bit on the expense of mobile, which has been very large in the last up cycles and is still relevant.

And as I explained, with AI-enabled phones kicking in, again, seeing kind of a revival, but not at the level we are -- we have seen it, and we expect the Transceiver business to be. So both Transceivers and 2.5D compute will be the major contributors going forward. And if you translate that to product lines, it is Flip Chip for the compute and Evo mainly for the Transceiver business.

Martin Jungfleisch, BNP Paribas. The first question is really on China. I think you have talked about much about it. How much of an opportunity do you see there? And is that included in your TAM? So the Hybrid Bonding TAM, does it include the bonders that are potentially going to China? That's the first question.

And the second one really is if you could split up the TAM for Hybrid Bonding for the high and low end. I think last year, you gave us these individual market groups by HBM, AI Logic, et cetera. So just is it fair to assume that the majority of the increase would come from AI Logic part of it today?

Well, let me answer that question. So your first question about China, yes, we are considering China as we are selling to China today. So it's also considered in our TAM. And the second one is the main portion actually from today's model that I have been presenting, the increase is coming really from the CPO section actually. It's for sure, the AI Logic section, as I said, and also the computing Logic section is confirmed. But once again, that section was the first one to be manufactured in the world. So that was the segment, which was also years back, the easiest one to predict.

So that was always -- has always been there and has always been refined year-on-year. So that's not the segment which kind of is giving a bit the big surprise. So the -- because, yes, we know that segment best. But really, the CPO, as I said, which was really skyrocketing within 1 year from R&D to real manufacturing and now has a tremendous outlook, which is definitely supported if you dig deeper and look into what is happening in the data centers and how copper interconnects are being replaced by fiber interconnects for the pure reason of speed and especially less power needs for fiber interconnect.

So you can also translate that to the end customer needs. So it's clearly the CPO, which is the driving factor.

Martin Marandon from ODDO BHF. My first question is on Hybrid Bonding competition. I'm trying to understand how much of an edge you have versus competition, notably with the new 50-nanometer accuracy tool. So maybe what could be helpful is that maybe giving an example of current products, which would be based on 15-nanometer accuracy hybrid bonding tools where maybe you are alone in the qualification process if is that relevant?

And the second question is on memory. I think I saw a slide where you talked about HBM5 and the fact that, it will be in the majority hybrid bonding. I'm just trying to understand there, if you talk about the majority of chips, we have hybrid bonding in them or you also are talking about the hybrid bonding assembly market being larger than TCB for HBM5?

Let me answer the first part. Peter, you the second part. On the competition, key is, of course, to find the mainstream adopters. And in that mainstream, any production of any good in the world looks for a process window, reliability of that window, and then the outcome is the yield. So that is ultimately where you can achieve a volume sale, which then if you do it right, brings your margins and you can pay your R&D cost and your support.

So the focus on mainstream application is where it's all about. If you look at the competitive landscape, certainly, the highest volume is expected in memory. And that's also not difficult in a rule of thumb, there's a relationship 1 to 4 or 1 to 6 between logic and memory in a very simplistic way. So most competitors are looking at that part of the market. You also heard earlier that the accuracy requirement for memory is a bit less than the accuracy required for logic because you have more IOs, which are denser designed.

So if you look at that whole landscape, and we look at it every single day, you also try to understand which solution is best suitable along those criteria in the next generation because it's always about the next generation. Today, we have this 100 as more or less an industry standard. Also, as said, 100 is now enough for logic and also for memory, stacking. And that's fantastic because you can make that platform ever better.

You can improve the accuracy, you can improve the speed, you can improve the reliability and for that reason, the yield. So that's in full swing every single day. And you can bet your life, customers are beating us up that it should be better and measured. And we have these calls where I'm also involved following the progress. And as long as that's happening, it's difficult for competition to get in. Competition will get in with the next round.

So the next round, 50 on the logic -- we have several prototypes now running, one focused on the accuracy, the other one focused on the speed. So that platform, as Peter explained, you can use with slightly less accuracy, but then the expectation and already data shows that you can increase the speed because the machine is far more stable and that improves the cost of ownership. And also along those and why am I explaining this, customers are always looking if somebody else having a better product. That's the open competitive world.

So in these different applications with different requirements, you're carefully understanding through customers, but also through other context what's happening in this world. If you ask me, what is your competitive lead timing that's always difficult to answer. You can assume today that is a very decent lead. Will that last through to 5 years? Well, on the high end, it's pretty, let's say, safe to expect that there is not immediately 15-nanometer tool available that can be a revolution.

On the more lower, let's say, accuracies and that's where competitors are aiming you may have in 5 years, a segment of the market, which is less accurate, but also immediately, the margins will be far less attractive. So then the discussion, as always, where should you put your best assets. So your people should focus always on a next generation, and that's our strategy DNA and not to defend a market, which is in a lower application offering lower margins.

So in a bit of a longer answer, that's the way we are looking strategically and every week in our teams at what is happening in this market. Chris, from a technology road map angle, Peter and all the support people from very much how are we doing today. And also with the help of Applied, who have a much deeper, let's say, position in this overall market. So you try to gain as much intelligence every single day. But anyway, just to share with you how we look at that. Peter, next part.

Yes. If I understood your second question correct, it was about HBM5 and what we are thinking about, yes, hybrid demand versus TCB demand, yes. So well as I said in my -- during my presentation, so with every sequential generation, we are expecting higher hybrid needs simply in order to achieve the specifications of these new generations.

And if you also look at the forecast chart that I have been showing is when you look at the years where HBM5 is showing up, then you can see that in the total market, the growth only is then coming from the -- and that's normal from the newest generation anymore. Now if you then say, well, more than 50%, definitely far more than 50% will be hybrid. That means from a market, from a machine market perspective, going forward, that will be mainly a hybrid market versus a TCB market.

And that's different than today because today, all the growth rate, the nice growth rates that HBM has experienced in the last 2 years and 3 years, well, as TCB was or still is the standard yes, that was for sure all on TCB bonders. But that will drastically change then with the generations to come. That's what we are thinking about, how that will develop.

Marc, ING. Two questions. First, wafer-to-wafer hybrid bonding. I think we discussed it in the past, and I think there were clear technological and reasons why wafer-to-wafer would be less attractive for the long run. I think Peter, you discussed already a bit.

But I think ASML was quite vocal at the last Imec event that they are going into that direction. Is there something that really changed there in the discussion with the clients, do you think there's some reasons why wafer-to-wafer could be a bit more of a viable alternative to die-to-die or die-to-wafer?

Yes, I can take that one. There's definitely applications where wafer-to-wafer bonding makes a lot more sense and is more useful. So one good example of that is CBA DRAM, which is an emerging memory technology that requires the funding of different portions of the DRAM device using wafer-to-wafer bonding and that's expected to be a very high-volume application. It's not an application where, at least today, where we see die-to-wafer bonding making sense because of the fact that you simply can't test and evaluate the goodness of either side of that sandwich before bonding and they're anyway, both the same size. So it's a perfect application for wafer-to-wafer bonding.

There's also other applications like image sensors that have been using it for a long time for similar reasons, same die size, difficult to evaluate the quality before bonding. Those are the cases where wafer-to-wafer makes the most sense, in my opinion. And -- but there are high-volume applications for those use cases. What it cannot address is use cases where you have, let's say, multiple smaller die being bonded onto a larger one, which is most of the logic and memory combination kind of applications that we shared with you. And so that's where we don't see a threat coming from wafer-to-wafer bonding.

And second question is I think in the introduction, you said you only talk about confirmed clients. I mean I think there's quite a bit of talk about Terafab. And is that -- and some of your competitors said, okay, we take it seriously. Is there anything that you can say about this? Is there any discussion you already had? Or is it indeed potentially a large opportunity for you? Or how do you look at it?

It definitely is also for us a large opportunity, the Terafab, and we have started discussing with the team, which is setting up the Terafab up in the United States. They have confirmed that this will not only be a front-end facility, but also packaging will be included. And so yes, out of that, that's a business opportunity. And as it said, Terafab obviously, a very big one.

And every chip needs to be assembled one way or the other. So that is directly to confirm what Peter said. And the good thing is, again, with our Applied relationship, we are engaged from the very beginning, same with Rapidus in Japan. Same with new fabs being built in the U.S. So just to confirm that we are very much engaged and that is an understatement.

Ruben Devos from Kepler Chevreux. I just wanted to have another question on HBM. I think you've now mentioned being NVIDIA Feynman and CPO being included in your low case. HBM is still in the mid case. So that means you're -- I mean, you've brought a very convincing case, let's say, why it could be adopted, but you haven't done it in a low case. I guess there's maybe some hurdles yet that you're not entirely confident about.

I think just in general, apart from the fact that the process costs have to come down. You're not -- you don't have all of these parameters under your own control, probably also your partners, your customers are having some of that having impact there. So what could really unlock adoption, I guess?

Well, that's an interesting question, actually. And you're totally right with that question, but if you would see us preparing the presentation, then it would make clear because we were discussing on -- not on that specific question, but in general, to shape the graph this time, totally different to give kind of a different kind of transparency because it's really in these days, as you are pointing out, rightfully not necessarily more -- only more a low and a high case and something in between.

And in the end, and we already had it all different. And in the end, we said, well, you're the audience here is so used to this and also used to, okay, you have more -- in a nutshell, you have the logic in the low case and then the memory and then the rest in the high case that we said, well, in the end, let's simply keep it.

But not because we are not convinced it's not coming, but simply in order to not confuse the history because then we are consistent with the charts. And so that's the reason why we just kept it like that. But it's interesting that you're asking that perhaps we should have still gone to changing the whole graph and showing it differently.

Yes. And I guess you could probably apply with the same reasoning for the fact that you now included in NVIDIA Feynman right in the low case. I mean you talked about CPO being a significant contributor to your uplift. But I guess, last year, you talked about sort of 500 units that you expect to sell in AI logic, you didn't repeat those targets anymore, but is that also a bit what we should think about that you're very conservative still there?

No, actually, the -- what you should consider is that in -- also in the older forecast, the last year or the year before, as I think I said already in one of the other, we already -- we did not only consider in this forecast, let's say, the companies that already had hybrid manufacturing. Otherwise, it would have in the beginning only an AMD forecast, and that's it. So we have, from the beginning, consider that, also this customer will change, also this customer will change one earlier, one later.

So also from that angle, we had factored in a certain portion of NVIDIA, for example, because you're asking Feynman specifically, already in our previous forecast. For sure, we have now refined the numbers and time lines and so on according to the information, but it's not that this is a total -- that this is coming on top totally. So with having that said, as a generic explanation, I can say that, yes, also the logic part was in our forecast was increased a little bit with this positive development.

But there, I have to say also when I did the update, our -- let's say, our view from last year was already pretty solid. So that's the reason why there was not so much change in that section, simply because we had yes, the best visibility anyway there. And so the big portion of the change is really coming from the CPO section.

[ Sam Jada ] with [ Galileo ] Investors. I had a question on Edge AI. So we saw NVIDIA introduced large models being put on device. Apple similarly with the smartphone is going to have some on-device models. Given the capability increase on models and the need for lower inference costs, latency improvements, all that stuff, do you see an update in terms of your long-term penetration of smartphones and just edge AI devices in general of hybrid bonding?

Well, I would say along these lines, I would see 2 things. So number one, there is definitely -- well, today, we know it's all done, yes, not on the edge device. It's just passed through and done in the data center. So still, we are seeing activities in the smartphone section to make the processors more powerful. And for that reason, we know some, let's say, projects where yes, where they are trying to incorporate or planning to incorporate hybrid funding for these application processors. Actually, I see the biggest drive for that coming out of China, by the way, as is on a side note.

The other thing which is going along the line and which is very interesting lately is that [Technical Difficulty] what we have not seen last year, for example, and the year before is compute power and on an edge device -- and that's on car processes for automotive for cars. And there, we are seeing currently a trend starting, I would say, that really they want to make the processors in the car, way more powerful in order to utilize AI.

But there, as we all know, you can -- because of latency, you cannot rely on yes, whatever on the data center and to get the data back because, yes, a self-driving car has to react immediately. And so lately, we are seeing something in that arena which is then not handheld device, but which is still edge AI, but in that case, automotive edge AI. Not sure if there is something to add from you, Chris?

Yes. I would just add. I think outside of the [ Engine 2 ] hybrid bonding side of the business, I think in the [ Engine 1 ] side of the business, it is having -- or presenting opportunities, like I mentioned in my presentation. So what it means is higher processing power on the AP, means higher heat dissipation and the need for a lot more memory capacity, LPDDR next to the processor. And that's actually driving different kinds of packaging, which I mentioned.

So there's opportunities in chip-to-wafer, die attach using mass reflow flip chip. There's opportunities in high precision placement for memory stacking, which is something new in the memory market where Besi can participate or before we couldn't and even in things like molding. So we do definitely see some relevant changes there that present opportunities. Yes.

Yes. Let me add one more. You mentioned APs processors. You mentioned memory, but you have not yet mentioned these sensors, many of those new emerging AI devices come with more sensing functionality in order to sense your surrounding environment and give you context-based feedback. So we also see quite a lot going on, on the sensor side, which is again very much supporting parts of the mainstream business because we have a strong position with many of the sensor suppliers.

Michael, Degroof Petercam. And first one is about data centers. If somebody designs a data center fully loaded with photonics and co-packaged optics, how many more bonding steps are they needed compared to one with copper connections? Is that factor 2 or 3?

Well, we've heard numbers from NVIDIA speakers at conferences talking about like 1 million CPO connections in a data center. But we haven't really -- at least I haven't done the math on the ratio of total bonds. And all the other packages and other components in the data center versus that, but it's a pretty big number.

And it's much more intensive, so positive.

Yes. Yes, definitely. I mean, I showed you the network switch. The Spectrum-X switch, the switch itself, the processor has 7 individual chiplets I didn't talked about, but that's already a chiplet package. So there's 7 bonding steps there, plus one to attach that to the substrates, 8. And there's 36 hyper bonded chiplets around that. But -- so it adds the hyper bonding steps inside the chiplet plus a flip chip or TCB step on to the substrate. So that's a big multiplier on that particular component.

Now when you talk about GPUs, it's going to be a very similar story, not as many per GPU, but now you have a GPU and an associated switch for every GPU. So it's -- yes, maybe we'll do that math and present it next year.

Okay. And is there -- are data centers also being retrofitted with photonics and co-packaged optics? Or is that only going to happen in 7 years when the first AI centers will be burned out?

I think it's very difficult to retrofit, yes.

The second question is a follow-up question on services. The annual report always provides segmentation between equipment sales and services sales. And I noticed that services have been flattish over the past 5 years, even though you sold more than EUR 2 billion of equipment to your customers, which must have boosted their installed base. Typically, when installed base goes up, your services sales also go up, but it didn't happen. So I was wondering what has been happening.

Well, number one, to answer that, it's -- you need to see the timing because if you sell a machine, you are first having a warranty period and only after warranty period, then you sell the service. And sometimes, customers are even in advance buying, let's say, instead of the usual 12 months, 24 months with an increased machine price already.

And so -- and as I said on the other answer before, the service increase is mainly coming from -- yes, from my business unit because that's more than front-end related. And so the huge numbers of sales and if you go now back 1 year, 2 years, my numbers were relatively speaking, to Christoph's numbers, still the way smaller ones. So you -- so we are only starting now with the service contract because of the service of the warranty period. And -- but now as the numbers are increasing and increasing and always with the delay of the warranty period, you will see that climbing up. So it's what you see or what you are observing here is more or less the introduction phase, which has a certain delay relative to warranty.

Okay. But again, it's about a 5-year period, and that seems to be that the delay will eventually kick in. But...

5 years back, we didn't sell hundreds of hybrid bonders. We did sell hybrid bonders, but they were very, very limited and in number back then because it was only starting. So you cannot expect a big number out of that.

No, but I mean the flip chip units have also been so -- I'm talking about total...

Normal flip chip bonders, you don't sell service contracts with that. That's a classic back end equipment and you install it and that's it. Nigel?

A question on China. I think most of the slides very detailed by the way, thank you for that. I've been on end markets, but -- can you discuss how you are addressing what I believe to be strong growth and push in terms of advanced packaging in China?

It seems you have a really good market positioning. And I think, Christoph, you've mentioned as the demand driver or pillar you said, but how do you see this in the next couple of years, both from a competitive perspective, but also in terms of growth rate across, I guess, 2.5D, photonics consumer, et cetera.

Yes, as you rightfully said, China is important for us, remains important for us, especially for the mainstream product portfolio. We see basically very similar developments as in all the other countries. We see photonics being very strong there, we see not directly CoWoS, but CoWoS like packages, which are from a bonding perspective, very similar to the TSMC application. We bond to that demand by having our own facility there, right, with Leshan manufacturing, China for China there and not only manufacturing, but also having a dedicated supply chain for the Chinese market.

What we see is that the competitive intensity there is very high, and that has not so much to do with our typical competitors but more with local Chinese competitors also driven largely by government funds and government policies to introduce local sources whenever possible. So that's kind of a limiting factor for us. However, so far with our advanced technologies, we have been able to escape those competing fields and find our sweet spots where we can offer technology that's simply not available in China and gives basically really a lead in supplying equipment into those fields.

As I said, transceivers is one of those fields. And honestly, there have -- we have seen a trend that some of the lower-end processes were moving over to Chinese competitors. But yes, this is anyway then a business we are not interested in as it is very, very price sensitive and margins we can achieve there are not as expected.

So yes, we focus on the high end part of the business is always, yes, that's how we look at China. You have to be careful. You have -- we don't develop anything there, right? All the core developments are driven out of the European headquarters. We have engineering in Singapore, and we have in China, basically, the application team that is bringing the machines to the customer, but not exposing ourselves to the IP risk and helping our competitors to get up to speed.

Thank you. Actually, I had a question on exactly that. But more on Peter's side, you have been shipping hybrid bonders to China. I was wondering -- well, 2 things. First off, aren't you afraid that those are copied by local competitors?

And also, if not, you said that it's part of your TAM. Is there a point maybe in a few years where these tools, I can imagine are shipped for R&D purposes today inflect in terms of volume applications?

Well, if you don't want, I will answer it. Happily, well, to add to Christoph's part first, you have to also understand which customers do we support and there are basically 3 types of customers in China. We went to China to support non-Chinese customers setting up their operations in China. In the very early days, it was Motorola and then TI and Intel and yes, in the whole supply chain for high-end smartphones. So that's one part.

At the same time, using Chinese subcontractors and many customers of ours, European customers have built capacities at and therefore typical ones who have capacities established for non-Chinese customers. And then you have the third part, Chinese customers. For many years, that was more, let's say, the middle and the lower end. But recently, as explained, the 2.5D modules [ goal ] was like using our flip chip, also driven by their end customer. So the end customer who builds those modules with Chinese suppliers, dictate which equipment to use.

So our strategy for China, and that brings me also to the hybrid camp is driven by end customers and is driven, of course, by margin, not by market share. In order to protect ourselves in current situation already, but in the future, that may well be more stringent, that we have less U.S. components in our machines, we have them all very carefully analyzed not to make any mistakes. And that may well, in the end, bring that part of the business in line with what's happening in geopolitics to whatever lower levels. But that the world demand is the world demand that will then be manufactured outside of China.

So you see an enormous increase in the countries around China in Vietnam, for instance, more done in Philippines. Thailand is up and coming for the more mid- and lower-end, India. So simply follow the customers. And those customers are simply evaluating day by day, what is the best tool of reference? Is that a basic tool or somebody else? And that's, in a nutshell, how we organize our business. Does that answer your question, Nigel?

Yes.

Any next question? Have we answered all the questions? It can't be -- yes, finally.

Richard, actually, following on to that question, I mean, you've seen in the press about the new Chinese scaling, which is not Moore's Law based, but based on some folding structures, and that would probably include hybrid bonding. Has there been any discussion with governments on this? Because the risk remains that there will be further risk. You don't have restrictions on hybrid bonding into China at this point. But whether there will be restrictions given how the industry there is trying to overcome their lack of access to EUV, as such.

And then secondly, there has been speculation in the press in the last few months about BESI being acquired. I mean, any thoughts -- I mean is Besi a big company in its own right now as such really. So does it really need a partner like that?

Well, let me first answer this wonderful development [ Tau ]. Maybe Chris, better answer on this stacking, you're the technology man, so...

Yes. I mean, from what we know about this Tau scaling, first of all, the -- what you're referring to is -- because the access to advanced lithography is not there. One solution to that is to simply use more silicon area and fold so-called folding the circuit on top of each other. And frankly the same argument that we use for all the other hybrid bonding use cases, right?

Our understanding is that at least the first implementation that they're trying to accomplish there is using wafer-to-wafer bonding so that they don't test anything beforehand. They just sandwich it together and then test it later. But we'll see how it develops. They very well could adopt die-to-wafer hybrid bonding for that use case as well. But we don't really -- at least I don't see that as being fundamentally different than other kinds of logic-to-logic hybrid bonding use cases. Yes.

Yes. And the second question we simply refer to 2 press releases in the last 2 years. When those rumors arise, and I hope we convince you a little bit today, we have great confidence in our own strategy. And we have a wonderful opportunity to execute on that strategy. And rumors we don't respond to.

Also the partnerships, and we highlighted that today and in several comments with the Applied Materials is very important for us. Partnerships in this industry are becoming ever more also for the simple reason that due to geopolitics, the M&A situation is what it is nonexistent nearly, so partnerships is the way to go. So that's what it is.

So I see 6 zeros in front of me. We've run out of time. I hope we answered all your imminent questions. If you have more questions, don't hesitate. Thank you all for coming, and those participating online also, thank you very much. And let's end it for today. Thank you.

So maybe what could be helpful is that maybe giving an example of current products, which would be based on 15-nanometer accuracy hybrid bonding tools where maybe you are alone in the qualification process if is that relevant?

And the second question is on memory. I think I saw a slide where you talked about HBM5 and the fact that, it will be in the majority hybrid bonding. I'm just trying to understand there, if you talk about the majority of chips, we have hybrid bonding in them or you also are talking about the hybrid bonding assembly market being larger than TCB for HBM5?

Let me answer the first part. Peter, you the second part. On the competition, key is, of course, to find the mainstream adopters. And in that mainstream, any production of any good in the world looks for a process window, reliability of that window, and then the outcome is the yield. So that is ultimately where you can achieve a volume sale, which then if you do it right, brings your margins and you can pay your R&D cost and your support.

So the focus on mainstream application is where it's all about. If you look at the competitive landscape, certainly, the highest volume is expected in memory. And that's also not difficult in a rule of thumb, there's a relationship 1 to 4 or 1 to 6 between logic and memory in a very simplistic way. So most competitors are looking at that part of the market. You also heard earlier that the accuracy requirement for memory is a bit less than the accuracy required for logic because you have more IOs, which are denser designed.

So if you look at that whole landscape, and we look at it every single day, you also try to understand which solution is best suitable along those criteria in the next generation because it's always about the next generation. Today, we have this 100 as more or less an industry standard. Also, as said, 100 is now enough for logic and also for memory, stacking. And that's fantastic because you can make that platform ever better.

You can improve the accuracy, you can improve the speed, you can improve the reliability and for that reason, the yield. So that's in full swing every single day. And you can bet your life, customers are beating us up that it should be better and measured. And we have these calls where I'm also involved following the progress. And as long as that's happening, it's difficult for competition to get in. Competition will get in with the next round.

So the next round, 50 on the logic -- we have several prototypes now running, one focused on the accuracy, the other one focused on the speed. So that platform, as Peter explained, you can use with slightly less accuracy, but then the expectation and already data shows that you can increase the speed because the machine is far more stable and that improves the cost of ownership. And also along those and why am I explaining this, customers are always looking if somebody else having a better product. That's the open competitive world.

So in these different applications with different requirements, you're carefully understanding through customers, but also through other context what's happening in this world. If you ask me, what is your competitive lead timing that's always difficult to answer. You can assume today that is a very decent lead. Will that last through to 5 years? Well, on the high end, it's pretty, let's say, safe to expect that there is not immediately 15-nanometer tool available that can be a revolution.

On the more lower, let's say, accuracies and that's where competitors are aiming you may have in 5 years, a segment of the market, which is less accurate, but also immediately, the margins will be far less attractive. So then the discussion, as always, where should you put your best assets. So your people should focus always on a next generation, and that's our strategy DNA and not to defend a market, which is in a lower application offering lower margins.

So in a bit of a longer answer, that's the way we are looking strategically and every week in our teams at what is happening in this market. Chris, from a technology road map angle, Peter and all the support people from very much how are we doing today. And also with the help of Applied, who have a much deeper, let's say, position in this overall market. So you try to gain as much intelligence every single day. But anyway, just to share with you how we look at that. Peter, next part.

Yes. If I understood your second question correct, it was about HBM5 and what we are thinking about, yes, hybrid demand versus TCB demand, yes. So well as I said in my -- during my presentation, so with every sequential generation, we are expecting higher hybrid needs simply in order to achieve the specifications of these new generations.

And if you also look at the forecast chart that I have been showing is when you look at the years where HBM5 is showing up, then you can see that in the total market, the growth only is then coming from the -- and that's normal from the newest generation anymore. Now if you then say, well, more than 50%, definitely far more than 50% will be hybrid. That means from a market, from a machine market perspective, going forward, that will be mainly a hybrid market versus a TCB market.

And that's different than today because today, all the growth rate, the nice growth rates that HBM has experienced in the last 2 years and 3 years, well, as TCB was or still is the standard yes, that was for sure all on TCB bonders. But that will drastically change then with the generations to come. That's what we are thinking about, how that will develop.

Marc, ING. Two questions. First, wafer-to-wafer hybrid bonding. I think we discussed it in the past, and I think there were clear technological and reasons why wafer-to-wafer would be less attractive for the long run. I think Peter, you discussed already a bit.

But I think ASML was quite vocal at the last Imec event that they are going into that direction. Is there something that really changed there in the discussion with the clients, do you think there's some reasons why wafer-to-wafer could be a bit more of a viable alternative to die-to-die or die-to-wafer?

Yes, I can take that one. There's definitely applications where wafer-to-wafer bonding makes a lot more sense and is more useful. So one good example of that is CBA DRAM, which is an emerging memory technology that requires the funding of different portions of the DRAM device using wafer-to-wafer bonding and that's expected to be a very high-volume application. It's not an application where, at least today, where we see die-to-wafer bonding making sense because of the fact that you simply can't test and evaluate the goodness of either side of that sandwich before bonding and they're anyway, both the same size. So it's a perfect application for wafer-to-wafer bonding.

There's also other applications like image sensors that have been using it for a long time for similar reasons, same die size, difficult to evaluate the quality before bonding. Those are the cases where wafer-to-wafer makes the most sense, in my opinion. And -- but there are high-volume applications for those use cases. What it cannot address is use cases where you have, let's say, multiple smaller die being bonded onto a larger one, which is most of the logic and memory combination kind of applications that we shared with you. And so that's where we don't see a threat coming from wafer-to-wafer bonding.

And second question is I think in the introduction, you said you only talk about confirmed clients. I mean I think there's quite a bit of talk about Terafab. And is that -- and some of your competitors said, okay, we take it seriously. Is there anything that you can say about this? Is there any discussion you already had? Or is it indeed potentially a large opportunity for you? Or how do you look at it?

It definitely is also for us a large opportunity, the Terafab, and we have started discussing with the team, which is setting up the Terafab up in the United States. They have confirmed that this will not only be a front-end facility, but also packaging will be included. And so yes, out of that, that's a business opportunity. And as it said, Terafab obviously, a very big one.

And every chip needs to be assembled one way or the other. So that is directly to confirm what Peter said. And the good thing is, again, with our Applied relationship, we are engaged from the very beginning, same with Rapidus in Japan. Same with new fabs being built in the U.S. So just to confirm that we are very much engaged and that is an understatement.

Ruben Devos from Kepler Chevreux. I just wanted to have another question on HBM. I think you've now mentioned being NVIDIA Feynman and CPO being included in your low case. HBM is still in the mid case. So that means you're -- I mean, you've brought a very convincing case, let's say, why it could be adopted, but you haven't done it in a low case. I guess there's maybe some hurdles yet that you're not entirely confident about.

I think just in general, apart from the fact that the process costs have to come down. You're not -- you don't have all of these parameters under your own control, probably also your partners, your customers are having some of that having impact there. So what could really unlock adoption, I guess?

Well, that's an interesting question, actually. And you're totally right with that question, but if you would see us preparing the presentation, then it would make clear because we were discussing on -- not on that specific question, but in general, to shape the graph this time, totally different to give kind of a different kind of transparency because it's really in these days, as you are pointing out, rightfully not necessarily more -- only more a low and a high case and something in between.

And in the end, and we already had it all different. And in the end, we said, well, you're the audience here is so used to this and also used to, okay, you have more -- in a nutshell, you have the logic in the low case and then the memory and then the rest in the high case that we said, well, in the end, let's simply keep it.

But not because we are not convinced it's not coming, but simply in order to not confuse the history because then we are consistent with the charts. And so that's the reason why we just kept it like that. But it's interesting that you're asking that perhaps we should have still gone to changing the whole graph and showing it differently.

Yes. And I guess you could probably apply with the same reasoning for the fact that you now included in NVIDIA Feynman right in the low case. I mean you talked about CPO being a significant contributor to your uplift. But I guess, last year, you talked about sort of 500 units that you expect to sell in AI logic, you didn't repeat those targets anymore, but is that also a bit what we should think about that you're very conservative still there?

No, actually, the -- what you should consider is that in -- also in the older forecast, the last year or the year before, as I think I said already in one of the other, we already -- we did not only consider in this forecast, let's say, the companies that already had hybrid manufacturing. Otherwise, it would have in the beginning only an AMD forecast, and that's it. So we have, from the beginning, consider that, also this customer will change, also this customer will change one earlier, one later.

So also from that angle, we had factored in a certain portion of NVIDIA, for example, because you're asking Feynman specifically, already in our previous forecast. For sure, we have now refined the numbers and time lines and so on according to the information, but it's not that this is a total -- that this is coming on top totally. So with having that said, as a generic explanation, I can say that, yes, also the logic part was in our forecast was increased a little bit with this positive development.

But there, I have to say also when I did the update, our -- let's say, our view from last year was already pretty solid. So that's the reason why there was not so much change in that section, simply because we had yes, the best visibility anyway there. And so the big portion of the change is really coming from the CPO section.

[ Sam Jada ] with [ Galileo ] Investors. I had a question on Edge AI. So we saw NVIDIA introduced large models being put on device. Apple similarly with the smartphone is going to have some on-device models. Given the capability increase on models and the need for lower inference costs, latency improvements, all that stuff, do you see an update in terms of your long-term penetration of smartphones and just edge AI devices in general of hybrid bonding?

Well, I would say along these lines, I would see 2 things. So number one, there is definitely -- well, today, we know it's all done, yes, not on the edge device. It's just passed through and done in the data center. So still, we are seeing activities in the smartphone section to make the processors more powerful. And for that reason, we know some, let's say, projects where yes, where they are trying to incorporate or planning to incorporate hybrid funding for these application processors. Actually, I see the biggest drive for that coming out of China, by the way, as is on a side note.

The other thing which is going along the line and which is very interesting lately is that [Technical Difficulty] what we have not seen last year, for example, and the year before is compute power and on an edge device -- and that's on car processes for automotive for cars. And there, we are seeing currently a trend starting, I would say, that really they want to make the processors in the car, way more powerful in order to utilize AI.

But there, as we all know, you can -- because of latency, you cannot rely on yes, whatever on the data center and to get the data back because, yes, a self-driving car has to react immediately. And so lately, we are seeing something in that arena which is then not handheld device, but which is still edge AI, but in that case, automotive edge AI. Not sure if there is something to add from you, Chris?

Yes. I would just add. I think outside of the [ Engine 2 ] hybrid bonding side of the business, I think in the [ Engine 1 ] side of the business, it is having -- or presenting opportunities, like I mentioned in my presentation. So what it means is higher processing power on the AP, means higher heat dissipation and the need for a lot more memory capacity, LPDDR next to the processor. And that's actually driving different kinds of packaging, which I mentioned.

So there's opportunities in chip-to-wafer, die attach using mass reflow flip chip. There's opportunities in high precision placement for memory stacking, which is something new in the memory market where Besi can participate or before we couldn't and even in things like molding. So we do definitely see some relevant changes there that present opportunities. Yes.

Yes. Let me add one more. You mentioned APs processors. You mentioned memory, but you have not yet mentioned these sensors, many of those new emerging AI devices come with more sensing functionality in order to sense your surrounding environment and give you context-based feedback. So we also see quite a lot going on, on the sensor side, which is again very much supporting parts of the mainstream business because we have a strong position with many of the sensor suppliers.

Michael, Degroof Petercam. And first one is about data centers. If somebody designs a data center fully loaded with photonics and co-packaged optics, how many more bonding steps are they needed compared to one with copper connections? Is that factor 2 or 3?

Well, we've heard numbers from NVIDIA speakers at conferences talking about like 1 million CPO connections in a data center. But we haven't really -- at least I haven't done the math on the ratio of total bonds. And all the other packages and other components in the data center versus that, but it's a pretty big number.

And it's much more intensive, so positive.

Yes. Yes, definitely. I mean, I showed you the network switch. The Spectrum-X switch, the switch itself, the processor has 7 individual chiplets I didn't talked about, but that's already a chiplet package. So there's 7 bonding steps there, plus one to attach that to the substrates, 8. And there's 36 hyper bonded chiplets around that. But -- so it adds the hyper bonding steps inside the chiplet plus a flip chip or TCB step on to the substrate. So that's a big multiplier on that particular component.

Now when you talk about GPUs, it's going to be a very similar story, not as many per GPU, but now you have a GPU and an associated switch for every GPU. So it's -- yes, maybe we'll do that math and present it next year.

Okay. And is there -- are data centers also being retrofitted with photonics and co-packaged optics? Or is that only going to happen in 7 years when the first AI centers will be burned out?

I think it's very difficult to retrofit, yes.

The second question is a follow-up question on services. The annual report always provides segmentation between equipment sales and services sales. And I noticed that services have been flattish over the past 5 years, even though you sold more than EUR 2 billion of equipment to your customers, which must have boosted their installed base. Typically, when installed base goes up, your services sales also go up, but it didn't happen. So I was wondering what has been happening.

Well, number one, to answer that, it's -- you need to see the timing because if you sell a machine, you are first having a warranty period and only after warranty period, then you sell the service. And sometimes, customers are even in advance buying, let's say, instead of the usual 12 months, 24 months with an increased machine price already.

And so -- and as I said on the other answer before, the service increase is mainly coming from -- yes, from my business unit because that's more than front-end related. And so the huge numbers of sales and if you go now back 1 year, 2 years, my numbers were relatively speaking, to Christoph's numbers, still the way smaller ones. So you -- so we are only starting now with the service contract because of the service of the warranty period. And -- but now as the numbers are increasing and increasing and always with the delay of the warranty period, you will see that climbing up. So it's what you see or what you are observing here is more or less the introduction phase, which has a certain delay relative to warranty.

Okay. But again, it's about a 5-year period, and that seems to be that the delay will eventually kick in. But...

5 years back, we didn't sell hundreds of hybrid bonders. We did sell hybrid bonders, but they were very, very limited and in number back then because it was only starting. So you cannot expect a big number out of that.

No, but I mean the flip chip units have also been so -- I'm talking about total...

Normal flip chip bonders, you don't sell service contracts with that. That's a classic back end equipment and you install it and that's it. Nigel?

A question on China. I think most of the slides very detailed by the way, thank you for that. I've been on end markets, but -- can you discuss how you are addressing what I believe to be strong growth and push in terms of advanced packaging in China?

It seems you have a really good market positioning. And I think, Christoph, you've mentioned as the demand driver or pillar you said, but how do you see this in the next couple of years, both from a competitive perspective, but also in terms of growth rate across, I guess, 2.5D, photonics consumer, et cetera.

Yes, as you rightfully said, China is important for us, remains important for us, especially for the mainstream product portfolio. We see basically very similar developments as in all the other countries. We see photonics being very strong there, we see not directly CoWoS, but CoWoS like packages, which are from a bonding perspective, very similar to the TSMC application. We bond to that demand by having our own facility there, right, with Leshan manufacturing, China for China there and not only manufacturing, but also having a dedicated supply chain for the Chinese market.

What we see is that the competitive intensity there is very high, and that has not so much to do with our typical competitors but more with local Chinese competitors also driven largely by government funds and government policies to introduce local sources whenever possible. So that's kind of a limiting factor for us. However, so far with our advanced technologies, we have been able to escape those competing fields and find our sweet spots where we can offer technology that's simply not available in China and gives basically really a lead in supplying equipment into those fields.

As I said, transceivers is one of those fields. And honestly, there have -- we have seen a trend that some of the lower-end processes were moving over to Chinese competitors. But yes, this is anyway then a business we are not interested in as it is very, very price sensitive and margins we can achieve there are not as expected.

So yes, we focus on the high end part of the business is always, yes, that's how we look at China. You have to be careful. You have -- we don't develop anything there, right? All the core developments are driven out of the European headquarters. We have engineering in Singapore, and we have in China, basically, the application team that is bringing the machines to the customer, but not exposing ourselves to the IP risk and helping our competitors to get up to speed.

Thank you. Actually, I had a question on exactly that. But more on Peter's side, you have been shipping hybrid bonders to China. I was wondering -- well, 2 things. First off, aren't you afraid that those are copied by local competitors?

And also, if not, you said that it's part of your TAM. Is there a point maybe in a few years where these tools, I can imagine are shipped for R&D purposes today inflect in terms of volume applications?

Well, if you don't want, I will answer it. Happily, well, to add to Christoph's part first, you have to also understand which customers do we support and there are basically 3 types of customers in China. We went to China to support non-Chinese customers setting up their operations in China. In the very early days, it was Motorola and then TI and Intel and yes, in the whole supply chain for high-end smartphones. So that's one part.

At the same time, using Chinese subcontractors and many customers of ours, European customers have built capacities at and therefore typical ones who have capacities established for non-Chinese customers. And then you have the third part, Chinese customers. For many years, that was more, let's say, the middle and the lower end. But recently, as explained, the 2.5D modules [ goal ] was like using our flip chip, also driven by their end customer. So the end customer who builds those modules with Chinese suppliers, dictate which equipment to use.

So our strategy for China, and that brings me also to the hybrid camp is driven by end customers and is driven, of course, by margin, not by market share. In order to protect ourselves in current situation already, but in the future, that may well be more stringent, that we have less U.S. components in our machines, we have them all very carefully analyzed not to make any mistakes. And that may well, in the end, bring that part of the business in line with what's happening in geopolitics to whatever lower levels. But that the world demand is the world demand that will then be manufactured outside of China.

So you see an enormous increase in the countries around China in Vietnam, for instance, more done in Philippines. Thailand is up and coming for the more mid- and lower-end, India. So simply follow the customers. And those customers are simply evaluating day by day, what is the best tool of reference? Is that a basic tool or somebody else? And that's, in a nutshell, how we organize our business. Does that answer your question, Nigel?

Yes.

Any next question? Have we answered all the questions? It can't be -- yes, finally.

Richard, actually, following on to that question, I mean, you've seen in the press about the new Chinese scaling, which is not Moore's Law based, but based on some folding structures, and that would probably include hybrid bonding. Has there been any discussion with governments on this? Because the risk remains that there will be further risk. You don't have restrictions on hybrid bonding into China at this point. But whether there will be restrictions given how the industry there is trying to overcome their lack of access to EUV, as such.

And then secondly, there has been speculation in the press in the last few months about BESI being acquired. I mean, any thoughts -- I mean is Besi a big company in its own right now as such really. So does it really need a partner like that?

Well, let me first answer this wonderful development [ Tau ]. Maybe Chris, better answer on this stacking, you're the technology man, so...

Yes. I mean, from what we know about this Tau scaling, first of all, the -- what you're referring to is -- because the access to advanced lithography is not there. One solution to that is to simply use more silicon area and fold so-called folding the circuit on top of each other. And frankly the same argument that we use for all the other hybrid bonding use cases, right?

Our understanding is that at least the first implementation that they're trying to accomplish there is using wafer-to-wafer bonding so that they don't test anything beforehand. They just sandwich it together and then test it later. But we'll see how it develops. They very well could adopt die-to-wafer hybrid bonding for that use case as well. But we don't really -- at least I don't see that as being fundamentally different than other kinds of logic-to-logic hybrid bonding use cases. Yes.

Yes. And the second question we simply refer to 2 press releases in the last 2 years. When those rumors arise, and I hope we convince you a little bit today, we have great confidence in our own strategy. And we have a wonderful opportunity to execute on that strategy. And rumors we don't respond to.

Also the partnerships, and we highlighted that today and in several comments with the Applied Materials is very important for us. Partnerships in this industry are becoming ever more also for the simple reason that due to geopolitics, the M&A situation is what it is nonexistent nearly, so partnerships is the way to go. So that's what it is.

So I see 6 zeros in front of me. We've run out of time. I hope we answered all your imminent questions. If you have more questions, don't hesitate. Thank you all for coming, and those participating online also, thank you very much. And let's end it for today. Thank you.

Welcome to the BE Semiconductor Industries Q1 Conference Call. I will now give the word to Richard Blickman. Richard, go ahead.

Thank you. Thank you all for joining this call. I'd like to remind everyone that on today's call, management will be making forward-looking statements. All statements other than statements of historical facts may be forward-looking statements. Forward-looking statements reflect Besi's current views and assumptions regarding future events, many of which are, by nature, inherently uncertain and beyond Besi's control. Actual results may differ materially from those in the forward-looking statements due to various risks and uncertainties. including, but not limited to factors that are discussed in the company's most recent periodic and current reports filed with the AFM. Such forward-looking statements, including guidance provided during today's call, speak only as of this date. Besi does not intend to update them in light of new information or future developments nor does Besi undertake any obligation to update the future forward-looking statements.

For today's call, we'd like to remind -- we'd like to review the key highlights for our first quarter ended March 31, 2026, and update you on the market, our strategy and outlook.

First, some overall thoughts on the first quarter. Besi reported strong first quarter results and advanced packaging orders in an improving industry environment. Revenue of EUR 184.9 million increased 28.3% versus the first quarter of 2025 due to higher shipments for high-end mobile and 2.5D AI photonics and data center applications. Q1 '26 orders of EUR 269.7 million more than doubled versus the first quarter of 2025 due to broad-based growth across all Besi's end-user markets, with particular strength in hybrid bonding, mobile and photonics applications.

Orders increased 7.7% versus Q4 last year due primarily to a significant increase in bookings for hybrid bonding systems from multiple customers and end-user applications. Increased revenue growth this quarter favorably influenced Besi's profitability.

Net income rose 20.6% and 63.8% versus Q4 '25 and Q1 '25, respectively, with net margin increasing to 27.9% versus the 21.9% in the first quarter of 2025. Improved profitability this quarter was due primarily to enhanced revenue growth, disciplined expense management and the benefits of operating leverage in Besi's business model. We realized a gross margin of 63.5% in the first quarter this year as increased prices helped offset increased component and energy cost inflation.

In addition, our liquidity position improved significantly with net cash growing by 186.9% versus the fourth quarter last year to reach EUR 103.3 million. Growth in our net cash position reflected improved profit and cash flow generation from operations of EUR 93 million in the first quarter 2026, which more than doubled versus the comparable period of the prior year.

During the quarter, Besi repurchased approximately -- for approximately EUR 14.2 million of its shares, which brings the total purchases to EUR 25.5 million under the current EUR 60 million buyback program.

Next, I'd like to discuss the current market environment and our strategy. We've noticed an important improvement in market conditions since our last report, driven primarily by strong growth in AI demand and to a lesser extent, additions to mobile and automotive capacity. The latest TechInsights forecast calls for 21% assembly market growth in '26 and 75% between 2025 and 2030. We expect to significantly exceed such projected growth rates given our leadership position in advanced packaging and wafer-level assembly, particularly in flip chip, multi-module die attach, hybrid bonding and next-generation TCB systems.

Favorable order trends in the first quarter of this year reflect the strength of Besi's advanced packaging market position, particularly for next-generation 2.5D and 3D AI applications. Unit orders for hybrid bonding systems more than doubled versus the fourth quarter last year and exceeded the prior quarterly peak reached in Q2 2024 with respect to total units and order value. Growth was due primarily to a larger-than-anticipated capacity build this quarter by a customer and to a lesser extent, repeat orders from a memory customer for HBM applications.

In addition, we shipped 2 evaluation tools to a second memory customer for HBM applications and adoption increased to 20 customers overall. Progress also continued on our TC next agenda with 2 new orders received and adoption increasing to 6 customers. Besi's business prospects for 2026 were also enhanced by renewed growth for high-end mobile and automotive applications in this first quarter.

Our business strategy is currently focused on supporting customer adoption of our wafer-level assembly and 2.5D AI product portfolio and ramping the supply chain and production personnel necessary to meet increased order levels. We are also developing additional Vietnamese production capacity for mainstream assembly applications in order to free up incremental capacity in Malaysia for wafer level assembly production. Further, Besi is increasing its service and support efforts in Taiwan and Korea in anticipation of increased hybrid bonding activities in such regions.

Our favorable outlook for hybrid bonding growth in 2026 is also supported by a series of new products and use cases announced this year for logic, memory, co-packaged optics and consumer applications. Such announcements suggest that the pace of hybrid bonding adoption is increasing as we approach the timing for the introduction of many new AI-related products anticipated in the 2027 to 2030 period.

Now a few words about our guidance. Based on our backlog and feedback from customers, we anticipate that Besi's Q2 '26 revenue will grow by 30% to 40% versus the first quarter of this year as strong revenue and order growth continue versus the prior year period. In addition, gross margins are anticipated to increase to a range of 64% to 66% Operating expenses are anticipated to be flat to up 10% due to increased revenue and customer support activities. As a result, we anticipate a significant expansion of our net income and profit margins relative to Q1 '26 and Q2 2025. As a result, we forecast for H1 '26 that revenue will increase by 49% versus the first half of 2025, assuming the midpoint of our second quarter '26 guidance with a substantial improvement in operating and net income.

That ends our prepared remarks. I would like to open the call for questions. Operator?

[Operator Instructions] Our first question comes from Didier Scemama from Bank of America.

Richard, can you hear me?

Yes, I can hear you Didier.

Sorry about it. Just on hybrid bonding, those orders in Q1, you mentioned that you're a bit surprised by those orders. I think it was not really expecting that they would be as significant as they were. Does that change anything about the profile of the ramp for this year at your main customer? Or like is that leading to higher deliveries already this year because of AP7? Just give us your thoughts on this. And then of course, I've got follow-ups on HBM.

Well, what is happening, you can follow easily in the bigger picture provided by Taiwan customer is that we see an acceleration in adoption of hybrid bonding and the orders scheduled for installation in the first round in AP7 has been pulled forward somewhat from Q2 to Q1. In addition, the program has been enlarged for 2 reasons. One, for the overall, let's say, time line to fill in the anticipated 100 bonders. That number we are told may be significantly higher, plus orders placed for co-packaged optics. So overall, you can say good news and acceleration of placing orders and to some extent, also an outlook for increased number of bonders required.

Understood. And so on that front, I think you mentioned in the past in previous calls that 2027, you could start to see some new AI logic customers coming on board. I mean, have you got sort of line of sight on that?

Well, we all know that AMD was the first to adopt hybrid bonding for several families, and they continue to do that. But then we also know Broadcom and one of the positive developments was also the Apple M5. So we see broader adoption. At the same time, we've heard or we've been told road maps from another very big customer in the data center modules that we can expect more hybrid bonding adoption going forward. So that is why we make the statement that we see accelerated and broader adoption. Also the number of customers. Remember, a quarter ago, it was 18, now we are at 20. So on the logic front, the adoption is broadening and increasing.

Okay. Makes sense. And then on the HBM front, I think you mentioned that you had repeat orders from, I think, a memory customer. I think that customer, if I understand correctly, was the one sort of in sort of final trial phases for HBM4E 16 high adoption. And I think he was expecting some form of results in shipping those samples to their large customer. Is that validation of your view that HBM 4E is the really insertion point for hybrid bonding? And any idea as to the volume opportunity there?

Well, the let's say, evaluation programs, the customer engagement end customer has also increased very well in this first quarter. That has resulted in several more orders. And if all goes well, that should lead to mainstream adoption for certain HBM devices. It's still following the time line which we have understood that this year will be a major qualification year. And then based on the success of the qualification, setting up production capabilities towards the end of this year for mainstream volume production in '27. That road map stands, and it's being supported ever more by orders by publications in the public domain of the progress. Also, one of the end customers is very clear also on their website on their adoption strategy of hybrid bonding for HBM. So that pace has picked up in the quarter.

And just a final question. I think last quarter, you said that the rule of thumb was 150 hybrid bonding system deployed by logic customers that will mean the TAM for HBM could be [ 600 ]. I mean anything that would sort of make you change your view either positively or negatively?

No, that still stands. So if you compare a capacity of 50 bonders for logic, and you simply look at all these beautiful websites and materials about building these 2.5D modules, you can easily see a processor surrounded by 3 or 4 memory stacks. And that explains you already one ratio. The other ratio when you have 16 dies in a stack, you need to do at 16x as opposed to 1 logic device. So you need much more capacity for HBM than you need for logic. But that has always been the case. So the rule of thumb is intact and also supported by customers demonstrating capabilities. One of the interesting recent documentation from TSMC is about the advanced packaging road map. And I invite everyone to look at that, published on CNBC.

The next question comes from Alexander Duval from Goldman Sachs.

Congrats on the strong orders and progress on hybrid bonding. Just wanted to ask a couple of regional questions. Firstly, when we look at the regional trends, it looks like U.S. was comparatively low relative to some other regions. So just curious to what extent it would be reasonable to expect an increase in the coming quarters as hybrid bonding orders for logic expand beyond your Asian customer base and into the U.S. customers? And then secondarily, understanding on China, it looks like robust orders there. I wondered if you could help delineate what are the key factors that were driving this?

Excellent. Well, U.S. currently at the levels where [indiscernible]. Remember, we had a big round for the initial capacity for hybrid bonding received already about 1.5 years ago. And that capacity is being filled in, is being qualified, is being tested. And based on the results of that customer, one can expect more bonders to be required or not. So that success is depending upon customer adoption. At the same time, we have the onshoring programs, one from TSMC to the U.S., one from Amkor, also Micron. And if all goes well, one can expect a shift from capacity built in Asia to more capacity onshore in the next years to come.

What we heard is that in the next 2 years, so '26, '27, preparation, building fabs and then as of '28, volume production. We are, of course, engaged in those programs. And timing, again, is, according to those customers' information, volume production as of '28.

And your second question about China, yes, there are several robust orders from Chinese-based customers. Number one, what's hot is the 2.5D coals like capacity expanding at the same time, photonics, all the pluggables and also a recovery in modules for high-end smartphones, so mobile, and carefully tide turning for industrial automotive. So that's the picture of China. But I can also share that more and more future capacities are built outside China. So you see more in Malaysia, Philippines, Thailand and also coming up more strongly Vietnam. And that's where we have our facility building currently tools by the end of this year, the first bonding system, not hybrid, but proxy bonding. And then you see a market opening up in India. 5 major customers are setting up production capabilities for mid- to lower-end devices, mostly power right now, but also modules for high-end smartphones and also other devices more in the mid-market applications. So China, although you have to segment also a China local market, which is also expanding, but the non-Chinese manufacturing in China, you see a clear change to countries outside of China.

Our next question comes from Ruben Devos from Kepler Cheuvreux.

I just had a follow-up on the second memory customer regarding the 2 evaluation tools. Just curious around your thoughts, whether you could help us understand a bit what they're testing at this stage? Like is this a full sort of tool of record type of evaluation? Or is it more of a focus qualification around the specific application, more configuration? And how would that conversion maybe from evaluation to pilot lines look like in terms of time line?

As I mentioned, the time line is '26, '27, '26 development, setting up certain pilot although small volumes for end market qualification purposes and then more production expected for mainstream market adoption '27 onwards.

Okay. And that would be a full tool of record type of evaluation, right?

Yes, of course.

Okay. And then just a second one regarding agentic AI. I think we've been hearing about agentic AI as a strong driver at the CPU level. Just interested to hear your thoughts whether that would have a different packaging intensity versus maybe the cycle that has so far been GPU-led? Also, have you seen any shift in the approach of your U.S. logic customer on advanced packaging with you in recent months? Is that CPU angle showing up in discussions?

No, not in those details. I can't help.

Okay. And just a final one. I mean, I think about like 6 weeks ago, there was some chatter around the potential relaxation of these JDEC thickness standards for 20 high. I mean they were talking about moving from 775 micron towards 825 or even 900. Yes, of course, curious how you read those discussions? And has that changed the conversations you're having with your memory customers at all?

No. First of all, it does not change the advantage of using a hybrid process over a reflow process. The benefits are more and more demonstrated that you have a faster circuitry, you need less power, and that means less heat. The only reason we understand that this height should be available is for a process for 1 of the 3, which simply requires that height. The other 2 are not impacted by that change.

So as we said end of February already with our year-end numbers or third week of February, and that is confirmed in the rest of this quarter, we see an increased engagement and activity and also announcements, and again, look at the Samsung website about hybrid bonding for HBM. That has not changed the adoption pace or rate of adoption because of the benefits.

And you could also add those benefits are every day more proven in the logic application. And you see a broadening adoption, higher volumes, pulled in capacity requirements. So that supports also the adoption of hybrid bonding in HBM stacking.

The next question comes from Charles Shi from Needham & Company.

First off, really congrats, Richard, I think hybrid bonding has been a 10-year work for you and for the company by now and glad to see it finally coming into fugitation. But I have a few very important clarification I want to make with you here. You said the 2 evaluation units is going to a second memory customer, but I thought you already have 2 memory customers. So is this actually going to the third memory customer?

No, you're right. We have -- we had 2. One is the U.S. and one Korean who started in a lab to develop a similar hybrid solution already 2 years ago, I think. But the change is that they have moved this to the forefront. So -- and that customer has 2 applications, one is logic, the other one is HBM stacking. So on the memory front, adding the third one, we now have all 3 who have our hybrid bonders to further evaluate and define the adoption of hybrid bonding for HBM stacking.

Got it. So the time line you provided to a previous question regarding that customer who just took your 2 evaluation units, 2026 qualification, 2027, maybe transitioning into production. Is that still the right time line to think for that -- I mean, the third memory customer who actually came in a little bit late?

Well, the time line is '26. And as we explained several quarters, that has not changed. The first customer aiming towards the mid of this year, June, July, and the second one, a bit following behind, which could be end of Q3, Q4. So that will determine the adoption of volume for '27.

And the third customer?

The third customer is ready to go, but they are all evaluating along the same, yes, let's say, parameters for one specific end customer, the whole world knows. That customer has invited all 3 to have these hybrid bonded stacks available by the end of '26 to be used in end market applications in '27.

Richard. That's very encouraging. So maybe I want to ask you one more question on the memory evaluation in general. We know your leading foundry customers sticking with the stand-alone tool configuration. But what's the landscape there for your memory customers between integrated and stand-alone? Which route do you think they are going to -- going after? And one of the very frequent questions I got from investors is whether there is any difference in terms of the economics, in terms of the revenue dollars you get from integrated tool setup versus a stand-alone on a like-to-like basis, meaning same configuration, same customer, are there any difference?

Excellent. I'll start with the dollar numbers first. So we sell bonders and AMAT sells Kinex automated lines. And they both have a sales value. And the extra which you have is the handshake between the bonder and the Kinex tool. Customers currently, and we have shared that several times, and you also said that in Taiwan, still the overwhelming majority is stand-alone because of the initial phase where we are in. So you have multiple customers, different die sizes, different process requirements and for flexibility reasons that customer uses stand-alone.

It's undisputable that in an integrated line, you achieve better process requirements, particles, also timing between the steps and the integrity overall of die-to-die and wafer-to-wafer. Those advantages are used in front end for over 3 decades in the so-called cluster [indiscernible] concepts. So the industry is evaluating the 2 aspects. Number one is the hybrid bonding process. And number two, what is the best total solution to produce devices using hybrid bonding. And as we all know, the hybrid process is very sensitive to particles, so 0 particle requirement. And by definition, in an automated line like the Kinex, you can achieve the best process environment specifications.

So the verdict, you can say, in a way is on the one hand, towards high-volume production of specific devices with minimum changeover. Once you have more changeover and you require more flexibility like the Taiwanese customer, at this moment, that is still in stand-alone. But that is very likely in the future to change to an automated line concept simply because of process requirements. But for us, back to the dollars, it doesn't make the difference for the bonder. The bonder has a certain value, cost of ownership value, and that is the same stand-alone compared to integrated in a line.

Richard, if you want to make a call today for HBM, is it more likely to be integrated or stand-alone? That's the last question.

Integrated because HBM is dedicated high-volume production. And then it's more likely to do that automated than stand-alone.

The next question comes from Robert Sanders from Deutsche Bank.

Just a question on HBM again. Can you talk about the yield numbers that you see at the moment in development? And what does the end customer need to see in terms of yield before they go ahead, whether it's the memory guys or the end customer? And the second question would just be, is this going to be a partial transition at HBM 4E if all things go well? Or would it be a wholesale transition? I guess the reason I'm asking is because there's a lot of installed TCB capacity that the companies would continue to like to reuse. So I'm just interested to think it will be a sort of bin chip. So only the best chips will be hybrid bonded and the rest will be used PCB and then maybe at HBM 5, it will become a full inersion? Or do you think it could be quite rapid?

Excellent. Thanks, Rob. Number one, what we hear is clearly, and this is forever, there's always a quality difference over a wafer on the certain devices. So you could, like we've had historically, end up with quality classes, the highest and hybrid bonders, et cetera. On your yield question, we do not receive detailed yield numbers. But what we know as a rule, if a process is not well up into the 99.9%, then it becomes a very difficult long-term perspective. So yields in interconnect are at those levels.

Where are we today? Well, we should be able to achieve those levels. Otherwise, it would make no sense to do these evaluations and qualifications. So the confidence is certainly that we should reach those levels. We reached them, as we all know, with logic already quite some time where the yields are very well up in the 99.99-something percent. So the hybrid bonding process can achieve that. And for HBM stacking, one still has to prove that, and that's what's currently happening.

Is HBM more difficult than logic? On the one hand, you have far less I/Os, so it should be more simple. On the other hand, you have vertical stacks of 6 devices. They are per definition thinner. So the process is different. And in certain ways, on the one hand, easier than logic because of bad pitch is less critical. But on the other hand, the vertical stacking. So they both have their specific, let's say, issues to deal with. But again, coming back to why are these customers, all of a sudden, and which we expected from the very beginning, putting far more effort into the adoption of hybrid bonding is simply because the proven performance upgrades are driving that adoption by an end customer, and that's a very big customer.

And any idea when the TCB market could see a downturn because of this in memory, for example?

No one should see. if you would imagine a certain volume to be produced, it will be less TC if one uses hybrid. So that's an offset in the same end volume. We should see that by the end of this year or as we have said already in the middle of this year, we should see more confirmation from the Samsung side. So in the end, it's the number of devices produced and whether you use Process A or Process B results in a different number of machines.

The next question comes from Martin Marandon-Carlhian from ODDO BHF.

My first question is on hybrid bonding use in GPUs. I mean the biggest GPU vendor and recently some aspect of their design for their next GPU coming out in '28, saying they would use 3D stacking. So first, do you expect it to be hybrid bonded? And second, if that's the case, when do you think you will have visibility on the timing of the ramp-up? And I have a follow-up.

Excellent. Well, that's exactly one of the major game changers, and it is forecasted to be hybrid bonded. So that all fits into the acceleration, which we explained at the very beginning, anticipating on the adoption of hybrid bonding in that family of next-generation products. And the timing for that is more equipment to be ordered and installed in '27, ordered in '26 for volume production as of '28.

Okay. Great. And a second one would be on the chip-on-panel packaging. Do you think that the shift to square panel could somewhat open new opportunities for TC -- next or hybrid bonding?

Well, the 310 by 310 panel is a very clear development coming to market also in the next year and 2 years. We already received orders for certain applications. So our bonders can handle the 310 by 310. Hybrid is a bit early, but we see it for many other applications being anticipated because it saves quite some waste. And you can expect with larger die sizes more module type of designs, 2.5D, but even more 3D, that panel will be used as a carrier more and more. So that trend is clearly visible. We will share some more information on the Capital Markets Day or Investor Day mid-June to give you some more examples, but that is certainly happening.

Okay. Great. And the last one for me, just on the cost for change. I mean the guidance for the next quarter is OpEx up around mid-single digits sequentially, while sales are up 30% to 40%. So can you share a bit more color on what you did there to maintain that kind of discipline on OpEx, that would be helpful?

Simply, it's controlling costs. That's our job. No, but there is no change as such in our structure. But with increased revenue, you have an enormous operating leverage, if that is also your question.

Our next question comes from Nigel van Putten Putter from Morgan Stanley.

I've got a question on photonics actually, even before moving to co-packaged optics. I think you're already seeing quite a wide range of applications in terms of your tools like hybrid bonding, I think TCB, flip chip and multi-module attach can all be involved here. But in terms of sort of focusing on the near term, so actually before CPO, are you already seeing more of a benefit as the market moves to silicon photonics and also and/or, I guess, higher throughput pluggable devices? Yes, that's my first question.

Well, as we have reported, started middle of last year, a significant expansion of that market segment with multiple customers building those pluggables. And also in the pluggables, you have the next generation, which requires more bonding steps. So that unfolds in a very positive way for us. You see that also in the numbers and the details we provide.

You should not mix that with co-packaged optics because that's another application and a different process. Also on that co-packaged optics, we have made significant progress. And for instance, the [indiscernible] process, we delivered the hybrid bonding for accomplishing those kind of contacts. Also there, we will spend more details on background and development road maps in the Investor Day. But again, it's certainly an extension of the hybrid bonding applications. into this rapidly developing market.

Helpful. Sorry, go ahead.

Sorry, does that answer your question?

Well, I had a follow-up, but I'll wait until the Investor Day then I look forward to receiving more detail. I want to ask my second question on order intake, which has clearly been very strong last 2 quarters. I know you don't really disclose the backlog, but I calculate around EUR 400 million by March end. So that seems you could do with some digestion on the order side while still growing revenue very comfortably. However, on the other hand, I presume the backlog is for a narrower set of applications around 2.5D and hybrid bonding, while you're also now flagging mobile and automotive picking up. So essentially, how should we think about order intake in the current quarter relative to the last 2?

Well, we mentioned continued momentum, a continuing trend. Don't forget, we are in an up cycle and up market. So as long as there's no signs of saturation in the end market, you can expect that to continue. We have been able to ramp our capacities in past up cycles significantly, 50% quarter-on-quarter. You see that now again ramping as well. So that's as much as we can. Yes, so far this quarter, we have seen no change.

The next question comes from Nabeel Aziz from Rothschild & Co Redburn.

I just had one on your service business. You talked about raising your presence in Taiwan and Korea for your service professionals in terms of preparing for greater hybrid bonding shipments. Have you seen a pickup in recent quarters in your service revenues in 4Q and in 1Q? And how do you see that trending through this year?

Well, certainly, number one, when the tide turns positively, clearly, customers' production lines are loading and they need more support. They need more spare parts, they need more service, upgrades. And then for hybrid bonding, but also for certain refill processes, you need more specialized support to reach the 24/7 production requirements. And that simply is following a model used in front end where, within 4 hours, a defined list of spare parts for hybrid bonders it's close to 900 need to be available. And that's all in place. So you see a broad increase in the demand of service spares and retrofit kits.

Okay. Yes. That's very clear. And I think on -- in recent years, your service revenues have been pretty stable around 15%, 16% of group revenues. So as we look forward with a greater proportion of hybrid bonding in the mix and your hybrid bonding installed base growing, should we expect the service intensity to reflect more a front-end mix kind of towards 20%-ish range of group revenues?

Absolutely. So what I just explained in a few words, the level of support we have to provide to hybrid bonding front-end type of environment is significantly higher than in the back-end environment. So that 15% may very well move up towards the 18%, 19%, 20%. For front end, it's typically somewhere between 20% and 25%. Also the long-term contracts in service and support are standard in front end. So that increases and changes the model altogether.

And then just last one. So yes, on a margin perspective, do you see the greater requirements being either a headwind or a tailwind to gross margins?

A tailwind, certainly a tailwind. So support is certainly, if you organize it right, of course, but that's with everything, is potentially a higher-margin business.

The next question comes from Martin Jungfleisch from BNP Paribas.

Congrats on the strong results. The first question is really on capacities and lead times. In the press release today, you talked about freeing up incremental capacity in Malaysia. Can you just disclose what your current hybrid bonding capacity is in terms of tools per month or year and where the expansion could potentially get you to? And also, if you could provide some updates on lead time for hybrid bonders now that the order momentum is picking up quite a bit.

We were at 15 bonders theoretically per month. So that leads to about 180. With the increase in floor space and adjusted to a model required by several customers, we can now expand that to 250 per year. So that is a significant increase altogether. You won't see that for the number of bonders produced in the year, but how typically orders are placed and expected delivery by customers with a lead time for now the 100-nanometer of 6-month standard. We can satisfy any model presented to us by the big 5 using the current expanded capacity.

On top of that, you need more people in the field to install to support. I mentioned earlier the spare part model supporting operations. We have put that all in place. So the infrastructure needs to be ready to support that higher volume as well. So it's not just the production floor. But that is all part of our overall model, the EUR 1.5 billion to EUR 1.9 billion in the next 3 to 5 years, which is a prerequisite to support organization for growing revenue to those levels, which is roughly 2.5 to 3x what we have currently. And for the hybrid bonders, it's significantly more.

Right. And the other question is maybe on EMI from Inter. There's a bit of news flow on increased demand for packaging. Can you just disclose like what kind of relevance this business has for you and what kind of your prospects are, where you think this could go to in the future?

We are involved since the very beginning in placement of the EMI modules that could be a positive business impact. But as things with Intel develop as they do, we first need to see more evidence. But they have a significant capacity installed, which we delivered the systems placing those modules. But it's good news when it increases. Any next question?

We have time for one more question. The last question comes from Madeleine Jenkins from UBS.

I just have one quick question on China. I know they're building out a lot of capacity at the moment on 2.5D. I was just wondering on 3D or hybrid bonding. Are you in any discussions with them about this technology? And are they indicating that they might order tools kind of in the coming years? And when would you sell to China that equipment?

Number one, we only sell to China, what we and that with any country, what we are allowed to sell. So we follow very strictly the regulations in this case, by the U.S. government, and we have that tested every 6 months. And we are allowed simply with the current levels and the current ingredients in the die bonders and in the hybrid bonders, it's not much different. So that is open for use in the China market currently.

There's, of course, development going on and applications are still distant. There could be a philosophy to use hybrid bonding in 3D stacking to lengthen the node size life, so to increase the performance of those devices with a 3D hybrid bonded structure. We are, of course, in development of those kinds of modules. but that is still in very early stage. So the current big market in China is 2.5D mass reflow flip chip for us, which we also disclosed in previous quarters, which is more or less standard equipment, but very, very much advanced. Our flip chip has absolutely the best cost of ownership also in China. But you can expect that they will develop certain local Chinese device structures using a hybrid process.

Perfect. And just on that, so in terms of timing, is it a few years? Or obviously, China, they do things very quickly over there. So could it be sooner than that?

Yes. They are -- as I just said, they are engaged in development, also very aggressive in a sense, in positive sense to study carefully the benefits of a hybrid process. They are much more driving that. And it's also very easy to understand. The world outside China is very much trying to extend the life of a mass reflow process because we all know those processes. So the hurdle to move to hybrid takes time. In China, it is more because they can overcome that they are not allowed to invest in the next generation with smaller device geometry. So then to solve a using hybrid process, which could be a very significant market.

Thank you. And with that, I will now turn the call back over to Richard Blickman for any final remarks. Richard, go ahead.

Well, thank you all for taking the time and asking questions. You're most welcome if you need to understand some more details we're happy to provide. Thank you. Bye-bye.

Good morning, good afternoon, ladies and gentlemen. And welcome to Besi's quarterly conference call and audio webcast to discuss the company's 2025 fourth quarter and full year results. You can register for the conference call or log into the audio webcast via Besi's website, www.besi.com.

Joining us today are Mr. Richard Blickman, Chief Executive Officer; and Mrs. Andrea Kopp, Senior Vice President, Finance. [Operator Instructions]. As a reminder, ladies and gentlemen, this conference is being recorded and cannot be reproduced in whole or in part without written permission from the company.

I'd like to remind everyone that on today's call, we'll be making -- management will be making forward-looking statements. All statements other than statements of historical facts maybe forward-looking statements. Forward-looking statements reflect Besi's current views and assumptions regarding future events, many of which are, by nature, inherently uncertain and beyond Besi's control. Actual results may vary materially from those in the forward-looking statements due to various risks and uncertainties including, but not limited to factors that are discussed in the company's most recent periodic and current reports filed with the AFM. Such forward-looking statements, including guidance provided during today's call speak only as of this date, and Besi does not intend to update them in light of new information or future developments nor does Besi undertake any obligation to update the forward-looking statements.

I would now like to turn the call over to Mr. Richard Blickman. Please go ahead.

Thank you. For today's call, we'd like to review the key highlights for our fourth quarter and year ended December 31, 2025, and update you on the market, our strategy and outlook.

First, some overall thoughts on the fourth quarter. Besi's revenue, gross margin and operating expense development in the fourth quarter '25 exceeded the favorable end of prior guidance. Revenue of EUR 166.4 million and orders of EUR 250.4 million, increased by 25.4% and 43.3% versus the third quarter of '25, due principally to a broad-based increase in demand by Asian subcontractors for 2.5D data center applications, renewed capacity purchases for photonics applications and a significant increase in hybrid bonding orders.

Net income of EUR 42.8 million increased by 69.2% versus the third quarter of '25 due to higher revenue, increased gross margins from a more favorable product mix and lower-than-anticipated operating expense growth. Besi's progress in 2025 reflected the favorable influence of increased AI infrastructure spending on our business development. Orders of EUR 685 million increased by 16.8% versus 2024 due to strength in AI-related 2.5D demand for data center applications by Asian subcontractors and renewed capacity purchases for photonics applications.

Growth accelerated in the second half of the year, with orders increasing 63.6% versus the first half of '25. Orders for AI applications represented approximately 50% of our total orders in '25 and revenue from Besi's computing end user market grew by approximately 40% of revenue in 2024 to 50% in 2025. For the year, revenue of EUR 591.3 million decreased by 2.7% versus 2024 due to lower shipments for mobile, automotive and industrial end user markets as a result of ongoing weakness in overall assembly markets.

We continued to maintain attractive levels of profitability with gross operating and net margins realized of 63.3%, 29.3% and 22.3%, respectively. Given profits earned in 2025 and our solid liquidity position, we will propose a cash dividend of EUR 1.58 per share for approval at Besi's April AGM, which represents a 95% payout ratio.

Liquidity remained strong at year-end with cash and deposits of EUR 543 million and net cash of EUR 36 million, increasing by EUR 24.4 million and EUR 43.8 million, respectively, versus September 30, '25. We distributed EUR 254.8 million in the form of dividends and share repurchases in 2025, roughly equal to levels of 2024.

Next, I'd like to discuss the current market environment and our strategy. Tech insights currently forecast relatively flat assembly market growth between '24 and '25 driven by a push out of the anticipated assembly upturn from '25 to '26. However, they expect growth of 74% between '25 and 2030. Based on increased AI use cases and infrastructure spending, new product introductions, new fabs coming online and a recovery in mainstream assembly applications, we expect to significantly exceed such projected growth rates given our leadership position in advanced packaging.

We are pleased with our operational progress in 2025 as we completed a comprehensive strategic plan review with enhanced revenue and profit targets and organized additional production capacity and infrastructure to help support that growth. We also experienced progress on our wafer level assembly agenda as hybrid bonding adoption expanded to 18 customers cumulative order grew to 150-plus systems and new use cases were identified for cold package optics, ASICs and consumer applications.

In addition, 6 integrated hybrid bonding production lines were installed at a leading logic customer incorporating 30 Besi-hybrid bonders in collaboration with Applied Materials. The first 15-nanometer placement accuracy prototype system was also completed and available for customer qualification. Our position in the TC market was further enhanced as Besi's TC NXT adoption expanded to five customers for logic, memory and photonics applications. In addition, our Flip Chip and multi module die attach systems gained significant share in the market for AI-related 2.5D assembly structures addressing the rapid growth in demand for data center and photonics capacity.

Further, we successfully introduced a variety of next-generation die bonding and packaging systems for each of our traditional mainstream markets as we prepare for the next market upturn. We see market conditions improving in overall mainstream assembly markets based on favorable semiconductor unit growth trends and a significant reduction of excess semiconductor inventory.

Green shoots are appearing after an extended downturn of nearly 4 years in each of our principal end user markets. Customer road maps also point to expanded adoption of wafer-level assembly over the next 2 years related to hybrid bonding and TC NXT adoption in HBM 4, 4E, co-package optics, ASICs and new high-performance computing and mobile introductions. In addition, recent announcements of substantial AI-related infrastructure investments are expected to increase demand for advanced packaging. Increased AI investment has created capacity shortages for 2.5D packaging which has caused producers to secure increased production for many Asian subcontractors. Further, many new advanced packaging fabs are planned globally which should increase demand for our advanced packaging portfolio.

Now a few words about our guidance. We entered '26 with increased optimism based on strong order momentum experienced in the second half of '25, which has continued to date in the first quarter of 2026. Our current optimism is based on anticipated growth in 3 promising Besi revenue streams, 3D wafer level assembly, AI-related to 2.5D capacity and more traditional mainstream assembly applications. Our optimism also relates to the significant increase in demand from Chinese subcontractors as the country builds out its AI infrastructure.

For the first quarter '26, we anticipate that revenue will increase between 5% and 15% versus the fourth quarter of last year with gross margins ranging between 63% and 65%, aided by improved revenue and a more favorable advanced packaging product mix. Operating expenses are anticipated to increase by 10% to 15% as we maintain discipline in overhead growth while continuing to increase development spending to support long-term growth opportunities.

That ends my prepared remarks. I would like to open the call for some questions. Operator?

[Operator Instructions] The first question comes from Madeleine Jenkins from UBS.

My first one is just, Samsung has publicly said that they'll be dual tracking hybrid bonding and TCB 4E in HBM and the samples are being sent to customers. I was just wondering if you could kind of help us understand from a customer's perspective, what would make them choose the hybrid bonding version versus the TCB and vice versa? And then on that, just generally, when are you expecting the first high-volume orders to come through for hybrid bonding for HBM?

Well, excellent, Madeleine, happy to share some more background. 2026 will be a very important year to understand the adoption of hybrid bonding for HBM stacking. As is publicly shared by Samsung in particular, keynote speech last week in Korea at the SEMICON is a very clear road map to adopt hybrid bonding for very important reasons and that is performance and also heat. And that, with all kinds of tests in previous years should be superior to using a reflow process to build these stacks. We are currently in the evaluation process, customer sample and qualification process.

And as was published by that customer in the course of this year, early Q2, maybe Q2, May, June time frame, it should become clear how that inroad of hybrid bonding stacked in HBM 4, but also in the previous three, the 12 stack should find its way into the end markets. That is Samsung. As we all know, our other memory customer started already much earlier in testing and sampling hybrid bonded stacks and they are ready as soon as the market demands these technologies used for either HBM 4E or other stack devices. Also the 12 supposedly shows much better performance using a hybrid process than a refill process.

And last but not least, the #3, the largest of all the memory -- the three memory producers, has also announced that it will start qualification of the hybrid bonding process in the second quarter of this year to also come towards the end of this year to the conclusion whether this is a technology used for high-volume mainstream in the generation of HBM 4 or whether that is in preparation of the next generation, the 20 stack.

So all these tests, we will update you every quarter on the progress. Also, there's a lot of press coverage and those companies share that with the community also in conferences. So a very important year for hybrid adoption in the memory space.

That's very helpful. And then just my second question is on China. They're clearly adding a lot of AI capacity. Kind of how sustainable do you see this demand as being? Is it multiple customers? And also how high is your market share in this region for the AI bit?

The market share is very high. To our surprise, we would have expected and, let's say, solid market share as we have with mass reflow for a long time, but our share has gone up significantly also among the Chinese. How sustainable that is? Well, the answer is that the world expects an enormous increase in building data centers. So for that 2.5D, some qualified that we are only at the beginning. Our position, as I said earlier, is very strong with a very solid market share. And that you can also derive from our margin, our ongoing margin, gross margin but also net margin development.

The next question comes from Sandeep Deshpande from JPMorgan.

Richard, my question to you is regarding the logic market and the foundry market. I mean you've seen that some orders coming through in the last quarter on the foundry side. Do you expect that these orders from the foundry side continue into the first quarter? And when you say on your release that the order momentum remains strong in the first quarter, how would you quantify it? I mean, are we going to expect a strong sort of orders in the first quarter like you saw in the fourth quarter?

Well, the answer is, first of all, yes. So as we guided, continued momentum, that also means that we expect more orders in the logic space for hybrid bonders. And as we all know, the program in Taiwan entails several steps to build out a complete new factory. The first install start in June and operators have to be trained, maintenance has to be organized and that's all underway. And you can expect, as was also the case with the current factory, the AP6, that over the course of several quarters, that capacity will be built because supposedly demand is building. So that looks very promising.

Then following up on that -- on the logic side, do you expect in the logic business this year that is '26 will be much better than in '25 because when you look at how your order intake was at the end of '24, you had about 100 cumulative hybrid bonding orders, you've had 150 at the end of '25. So there was a slight slowdown in terms of the order intake. And if this could accelerate now into '26? I mean, clearly, memory will also contribute to that, but will logic itself accelerate?

Well, as I just explained to your first question, if all goes according to public shared plans, it should increase because already AP7 is supposedly twice the size of AP6 and that's only one customer. So the adoption for logic is continuing. We saw that in the whole of '25. Again, we now have 18 customers, of which most are the far most are logic oriented customers with all kinds of different device designs. Remember, the first was AMD which has expanded its family throughout. And then we have many others now following.

The big question here is when will the largest end customer have a product line using this technology, that should be on the horizon. So that then will create a significantly higher demand than what we have witnessed in '25. But that's according to the road map we've shared forever. There's a nice slide in our deck where we see a development in the past 5 years and an expected significant growth in the next 5 years.

As we've said many times, that line of growth, it can have several variations, especially as you said, the adoption of memory will change that landscape significantly in terms of total volume required but we're still on track on that, let's say, road map, we, ourselves derived from what is happening in the market in the past 4 years, which we update every year. So that's in a nutshell, the overall picture, we should or we could expect.

The next question comes from Didier Scemama from Bank of America.

Richard, I have a couple of questions. So first question is on HBM. If everything goes according to plan and your two lead partners decided to put the trigger on TCB or TC NXT and hybrid bonding, can you give us a sense of the magnitude of orders sort of the volumes that would be required to create a production line? I've got a follow-up.

Well, as a rule of thumb, typically, one needs a factor more memory supporting a logic device. So when you take the rule of thumb of a factor of 4, then with the installed base so far for logic, which is now over 130 systems, shortly coming up 150. Then if you multiply that, then you know how much capacity you would -- or how many machines you would require to support the capacity for memory. It doesn't work exactly like that, but the factor for number of machines capacity required is significantly higher than for the logic. So that's a major step up what we can expect when that adoption occurs. But that, again, you see in that picture we share on the adoption scenarios.

Understood. Very clear. My second question is on mobile. So if you remember, like, obviously, a few years back, very high-end smartphone adoption bonding. Can you just give us a sense as to, first, whether we should expect the traditional order intake in the first quarter related to high-end smartphone, new features, cameras, et cetera? And then if you look a bit further out, how this is shaping up to be in terms of hybrid bonding adoption, whether it's '27 or further out in at least your best guess?

Well, this year, as we already shared a quarter ago, we should see some improvements or new updates on features on high-end smartphones. On the camera front, there are some new developments. But also maybe foldable versions are, let's say, on the road maps, and that requires also different solutions inside those cameras. So those are developments, which we see.

But then the next question is what kind of computing power will need to support AI functions? And that is a big, let's say, question and that could have a significant impact and whether they are built with a reflow process or with a hard bonding process, chiplet architectures. As we've shared many times, there's a lot of development going on and certain road maps indicate those new major inflection points in technology, either already in '26 or certainly in '27. That is how it develops and there is no change in that road map. Does that answer your question?

Yes. Just had a quick follow-up. For your third quarter guidance, I just wonder why your order conversion is quite a lot lower than it normally is? So I think it's about 100% plus or minus. So why on EUR 250 million in Q4, you're sort of guiding to only EUR 185 million or so at the midpoint?

Well, a very easy answer. The orders were -- or let's say, the order placements were very much to the end of the quarter and the manufacturing throughput time for many of these orders, so take the high-end Flip Chip machines, the CHAMEOs, but also the multi-module attach, which is very much also for photonics, they take 12 to 16 weeks. So you simply can't physically arrange the shipment in the first quarter. So the answer also implies that you should see a significant impact on the second quarter.

The next question comes from Charles Shi from Needham & Company.

The first one -- I want to go back to the comment about the hybrid bonding cumulative revenue orders, it was 150 plus by the end of last year. And if I do the math, and it looks like last year, the number of orders you got was actually more or less comparable with 2024. So the question here is what about 2026? What's the overall sense where the cumulative order number will go? 200 seems possible. I mean that basically assumes, I mean, flattish number of orders you're going to add this year versus 2025. But can you go to 250? Can you go to 300? And I mean, to go to higher numbers, what do you think needs to happen for -- yes?

Two things need to happen. Number one, the adoption of hybrid bonding for mainstream applications for logic devices next to what is already now using hybrid bonding. So think about the big AI providers, which are still building certain modules using mass reflow, using TC, if they switch to hybrid bonding, that could change the landscape dramatically. And number two is as we discussed to earlier question, is the adoption of hybrid bonding for memory stacking.

Got it. But -- okay. So maybe I'll just go direct into memory. Now Samsung HBM 4E, that is the fact that it's happening. But I mean, on the other hand, if we understand correctly, the other 2 HBM customers have not even have a order from you -- hybrid bonding order from you. Why the hesitancy? That's the question I believe top of the mind for a lot of people here. And what's delaying them? And could they start getting some orders this year?

Well, it's -- the other two. One of them with the U.S. base. They have ordered already several hybrid bonders to develop HBM stacking for about 3 years now. It's also known publicly that the other one, the Korean, will start evaluating the hybrid bonding process in April-May time frame, we are invited for that, and they have publicly shared that their end customer demands them to have hybrid bonded version available by the end of this year. So although cost is higher using a hybrid process, performance is better in two ways. Number one is speed and number two is heat. So it is gradually moving from TC solutions for stacking to a hybrid version. And the big question, will this move in '26 or certainly in '27? That's how we read the inputs from all three and the biggest end customer driving, ultimately, the change in specification for these end products. Next question.

[Operator Instructions] The next question comes from Robert Sanders from Deutsche Bank.

Yes. Good afternoon, Richard, maybe the first question would just be around the situation with Apple, which seems to be moving to using SoIC-X for the M5 Pro and M5 Max. Is that beachhead do you think going to be swiftly followed by other SKUs? Or is it going to remain do you think a relatively niche use of SoIC-X for high-end notebook type situations? And I have a follow-up.

Well, that's precisely put. So they're all preparing from a production and technology readiness to be able to adopt this technology shift when decided. For us, what we can do is continue to offer whatever qualification samples, testing to be ready for that. There is one big question out there, how much additional computing power will be required for AI functionality?

And that is what we hear, still the question to be answered. How does that impact the choice of the technology used in these mobile devices? Because in the end, it will increase the cost but then the functionality is significantly more advanced. So that open debate, you follow at conferences directly from customer engineers, and also, we had our technology advisory board meeting 2 weeks ago in Taiwan, where there were also technology persons from the community sharing road maps and thoughts exactly on this subject.

And just quickly on China. Maybe you could just discuss a bit about what's happening in China. I mean I think it was 27% of your sales in the first half, but it has been higher than 45%. I mean it sounds like it's going to go up to close to 50%. Is that fair? And how do you think about the sustainability of that spending?

Well, so far, we've always had this typical mix. You have non-Chinese customers producing in China. Since 30 years, all non-Chinese customers have set up assembly capabilities in every technology and that is still today the case. Although there's a lot new established outside China and Asia, in Vietnam, in Thailand, in Malaysia, Philippines and then India, but that's still slow and coming. So to be less dependent upon China.

And then you have the emerging Chinese technology, which is growing year by year. And as we said for the 2.5D modules, we are very much engaged in these Chinese versions. So supposedly, the cost of ownership using our equipment is beneficial for local compared to local alternatives. Don't forget, we built all these machines in China. We have a wonderful facility in Leshan, which is expected this year to surpass the peak it achieved in '21. So although there's a lot of expectation that, that will become less, we don't see that at all.

But we are expanding in Vietnam. As many of you know, we have set up a factory 3 years ago. We're expanding that significantly this year. By the end of this year, we're also able to build one of our die attach systems in Vietnam. And then as I said earlier, the expansion in Thailand, in Malaysia, the whole Pacific Rim, is preparing to have next-generation products produced in those countries rather than establishing more capacity in China. But the Chinese market itself is growing rapidly.

Any next question?

The next question comes from Daniel Schafei from Citi.

Basically, the first one would be on TCB NXT. You mentioned 5 players. I was just wondering, just to clarify, this is a testing or are some of them already high-volume manufacturing? And then if hybrid bonding will take longer for some customers, what is your expectation now going forward for TCB, especially given you are now gaining traction within TCB NXT? That would be helpful to understand.

Well, number one, our system is designed for bond pad pitch below 20 micron. The world today is still above that, 25, 30. So the preparation with these five customers is to be ready once technology moves to smaller bond pad pitches and stretch the life of using a reflow process because the reflow has many advantages compared to hybrid bonding. One of them is simply cost. We have mentioned several times that our system has demonstrated even to be able to bond successfully at 10-micron bond pad pitches, and that comes very close to the crossover point with hybrid.

So we cover the space between mass reflow Flip Chip and as mentioned earlier, very successful at this moment. And then the TC space where it becomes difficult for TC and then beyond that, the hybrid bonding. So gradually, always the industry moves to smaller geometries, and that is where exactly this TC NXT is aimed for.

Your question, how much in high volume? Not yet. It's in the early stages in qualifications and in two areas. So in the logic space, so single die, but also one of the major memory producers is using TC NXT to prepare for the next generation. And that is what we mentioned last year when we received the order, 5 systems ready to go once that becomes the mainstream.

Perfect. And just as a follow-up, then you mentioned also earlier the adoption of hybrid bonding within '26 or '27. Just to understand what your expectations are right now, do you see hybrid bonding being adopted between all the HBM layers or only within certain layers? Yes, that would be just interesting to understand.

Well, it can be a mix. There are different road maps showing a combination of a certain hybrid part of the stack and also a reflow part. So one has to go into a bit more detail to understand all of the road maps, but that is also why we have this 2-track development strategy that you have to cover both.

Okay. Is it then dependent on the HBM structure itself, where I would say the mix is more a hybrid bonding. Basically, the taller you go. My question...

The reason -- sorry to interrupt you, is simply performance. If you connect direct copper-to-copper you have less heat in operating such as stack. And that allows you to get a higher power out of that stack and the higher the stack, the more, let's say, loss of power you have due to the heat. So a mix can already help in that performance.

The next question comes from Nabeel Aziz from Rothschild & Co Redburn.

So the first was just on hybrid bonding tool maturity. So I was just wondering if you could provide an update on the hybrid bonding tool maturity and progress that you're making on throughput and yield improvements?

Well, we've come a long way that after 4 years, you certainly can see enormous progress. And where do you see that progress is, number one, the predictability of any application. So understanding the right preparation time required and the preparation processes, remember, cleaning, tracking, wet, clean plasma. And that is the most, yes, let's say, process technology, which sets us apart from many others. There are many bonders in the world which can place accurately. But exactly that bond process is where it's all about.

Where are we right now? As I said in the beginning, we certainly have -- but there's still a long way to go. The process itself is, each time you could say every day, improved. One of the issues is always throughput, so the time required to place the die accurately. And the faster you can do that, you have more output of that machine and that influences the cost of ownership. So that [ battle ] is identical to what we have gone through with mass reflow Flip Chip for 25 years every year, either focus on accuracy improvement or focusing on throughput. And that combination is exactly the same challenge we have with now over 130 hybrid bonders operating in the field for larger die, smaller die, stacking dies and that's where we are.

Very clear. And just a quick follow-up on that. A lot of your competitors are starting to develop hybrid bonding solutions of their own and in some cases, shipping R&D tools. So I just wondered how you see the competitive landscape in hybrid bonding evolving? And how competitive are your peers' tools with your own?

Well, what we did share end of October was the simple fact that for the next round in Taiwan, that was based on the outcome of a complete landscape evaluation where -- because the orders were placed with us and are placed with us, the outcome is what it is today. But if you look at the whole landscape, everyone understands that hybrids sooner or later will become the mainstream technology for advanced packaging. So that's why every bonder company is focused on this market.

How can you maintain your leadership? Because after 10 years nearly where we started this development with that big Taiwanese customer, it's all what I just said along the accuracy and speed. So today, the 100-nanometer is sufficient covering the logic and the memory requirements as it looks today. In the next year, we have to move down 250 because of the next-generation technology. And then the accuracy and speed combination is what sets us apart from others.

Also, what is very important is the partnerships in this change of technology inflection from the assembly reflow space to hybrid bonding, hybrid bonding has to occur in front end. And front end requires complete different support structure than what we have in back end. Through the partnership with Applied Materials, now for 5 years, we have come at the levels that is supporting the highest end customers in the industry. And that combination is unique. So that support, so not only having a successful bonder, but also how to support customers 24/7 in a front-end environment is a complete different challenge than in the back end.

So we see certainly competitors trying to participate in this market as well. But there is a very clear challenge for us to maintain in that lead.

The following question comes from Ruben Devos.

I just had one on your prepared comments where you talked about new hybrid bonding use cases that were identified for co-packaged optics. I was curious, is that mostly referencing sort of the material you presented at the Investor Day in June? I think you talked about sort of NVIDIA Spectrum X, which requiring 36 hybrid bonding tests per device. I think earlier in this call, you talked a bit about the factor difference between memory and logic, but how does that shape up for maybe co-packaged optics?

And yes, I think the mid case was also somewhere around 50 cumulative units through 2030. But with the prepared comments around new use cases, might that really be contributing this year or next year?

Well, that's a very, very big question. Number one, co-packaged optics is still in early days and a lot of development is going on with the use of hybrid bromine because the accuracy is required. So as shared in the Capital Markets Day or in the Investor Day, that is going on, that's continuous development. How many systems that entails? I can't tell you at this very moment. So that's -- you could qualify that as the next step in technology. So we first have the interconnect and co-packaged optics is a step beyond.

Okay. And maybe something unrelated, talking about the mainstream market, basically, that's what I was thinking about. It's -- I think you also talked about green shoots, right, after a full year downturn. I think you mentioned smartphones in the mobile market, obviously, automotive and industrial are two other end-user markets. 50% of your business might already be computing. But so yes, a bit more color on maybe what you're seeing across the industry? What are maybe the die bonder utilization rates at this point? That would be very helpful.

Well, we have seen, as we said, green shoots. We see some of our main customers for years in automotive and industrial after a long time showing signs and having new programs where equipment will be required, which gives a positive outlook for '26. That is referenced to Techinsights, which also expects the market to carefully improve in '26, more sizable in '27. So that's how our comment is also based on.

What we have seen in these 4 years of very modest capacity increase only for new devices. We have seen development of many new devices ready for next-generation electronics, especially power devices for automotive, supposedly for hybrid application. So hybrid cars, I mean, not hybrid bonding. So in power, there's a lot happening. But still, the overall picture is not recovering to the extent which we are used to.

But the growth in the other areas is so significant that, that offsets the -- yes, usually, our revenue, we've shared that forever. Automotive was between 15% and 20% of revenue. Now it's somewhere around 10% to 15%, depends on which quarter. It probably will drop in the next quarters or it has to turn. But that's about -- well, it is 10% to 15% of revenue.

The following question comes from Marc Hesselink from ING.

Yes. Can we have a bit more view on the 2.5D photonics opportunity? I think that is sort of a momentum that's been building up throughout the year in '25. And I think with an extremely strong end with the order intake towards the end of the year, can you maybe see -- I would assume that in this business, maybe the -- because it's strategic investments, the visibility is a bit higher than in your usual mainstream product portfolio. So can you maybe see -- how do you see this ramping the capacity? Is it just a few quarters? Or is this a longer-term trend? Is this going to accelerate from where you are today? It would be very helpful if we get some extra detail there.

Well, there are two growth drivers. Number one is simply the data center, let's say, capacity built in the world. So the connectors to connect those computers inside those, yes, data center units, that is what we have been involved in for the past -- over 10 years. But the second driver is that there is a technology step and that will require twice the amount of steps, the interconnect steps in these connectors than the current generation. So there is definitely more growth ahead of us for these two drivers. So not just the growth in data centers, but also in technology.

I mentioned already, 10 years -- that started, well, over 10 years ago with Cisco modems. And these connectors, it's a set of 5 main customers, and they all produce for the very big end customer. And as long as that is growing as the world expects, we are directly linked to that. Does that answer your question?

Yes, it does. And maybe as a follow-up on that. Now that you're also seeing a lot of that volume coming from the OSAT. Is it then fair to assume that implies that it becomes even more mainstream and even more adoption beyond what you just mentioned?

Yes, certainly. Certainly. And that's also publicly known that the IDMs, as usual, they offload more mature products to the subcontractor space and the more complex the technology, the more attractive that is for the subcontractor space. And we know the big two leaders, both starting with an A. But then there are many subcontractors who are also involved in this expansion into mainstream for data center computing applications.

Any further questions?

The following question comes from Martin Marandon-Carlhian from ODDO BHF.

My first question is on the new fab of your Taiwanese customer, the AP7. Do you think the vast majority of the demand there will come from new customers adopting hybrid bonding? Or do you also expect AMD to be a big contributor since the announcement of its deal with OpenAI? That's the first one.

Well, what we hear, and I was just there 2 weeks ago, there are several big companies, and we all know the names, either for high-end smartphones or for data center computing who are supposedly on the brink of changing from reflow process designs to hybrid bonded designs in whatever end products. And that is the driver for a factory, which is twice the size of what is currently the AP6. But then there is a next plan, AP7 is not the end. So there are major plans.

So look at the model, which is shared by many of the front-end companies, what they expect in the next 3 years, the demand for AI translated into capacity that similar model you can use for the advanced packaging. So the next note plus an enormous expansion in end market demand. Whether that will -- as presented, we all know this industry but anyway, that's the picture driving the programs in Taiwan.

Okay. Very clear. And the second one is a bit of a different one, is on high-bandwidth flash HBF, some expect HBF to be necessary to improve the memory capacity in future AI chip packaging. So my question is just what do you think about this? Do you think it's a driver for hybrid bonding or TCB? Is it part of the current discussion with your customer? Or is it not really relevant in the near future?

Well, I can't answer that. Yes, simply, I have no, let's say -- if I would, I would answer it, of course, for you, but time will tell, and we are certainly following that closely.

Ladies and gentlemen, we have arrived at the end of the presentation. I would now like to hand the word over to Mr. Richard Blickman for any closing remarks.

Well, thank you all for joining us today. And in case you have any further questions, don't hesitate to contact us. Thank you. Bye-bye.

Ladies and gentlemen, you may now disconnect.

Good morning, good afternoon, ladies and gentlemen, and welcome to Besi's conference call and audio webcast to discuss the company's 2025 third quarter results. You can register for the conference call or log into the audio webcast via Besi's website, www.besi.com.

Joining us today are Mr. Richard Blickman, Chief Executive Officer; and Mrs. Andrea Kopp, Senior Vice President, Finance. [Operator Instructions] As a reminder, ladies and gentlemen, this conference is being recorded and cannot be reproduced in a whole or in part without permission from the company.

I will now hand the word over to Mr. Richard Blickman, Mr. Rich Blickman, go ahead.

Thank you. Thank you all for joining. I'd like to remind everyone that on today's call, management will be making forward-looking statements. All statements other than statements of historical facts may be forward-looking statements. Forward-looking statements reflect Besi's current views and assumptions regarding future events, many of which are, by nature, inherently uncertain and beyond Besi's control. Actual results may differ materially from those in the forward-looking statements due to various risks and uncertainties, including, but not limited to factors that are discussed in the company's most recent periodic and current reports filed with the AFM. Such forward-looking statements, including guidance provided during today's call speak only as of this date. Besi does not intend to update them in light of the new information or future developments nor does Besi undertake any obligation to update the forward-looking statements.

For today's call, we'd like to review the key highlights for our third quarter and 9 months ended September 30, 2025, and update you on the markets, our strategy and outlook. First, some overall thoughts on the third quarter. Besi reported Q3 '25 revenue and operating results within prior guidance in an assembly equipment market showing early signs of recovery. Order levels improved significantly Q3 '25 with bookings of EUR 174.7 million, increasing by 36.5% and 15.1% versus Q2 '25 and Q3 '24, respectively. For the quarter, revenue decreased by 10.4% and 15.3% versus Q2 '25 and Q3 '24, respectively, reflecting continued weakness in mainstream assembly markets, particularly mobile and automotive applications and lower hybrid bonding revenue.

Operating income was at the high end of guidance, reflecting higher-than-anticipated gross margins and operating expense developments, slightly better than forecast. The improved order outlook this quarter was principally due to a broad-based increase in die attach bookings by Asian subcontractors for mostly 2.5D data center applications and renewed capacity purchases by leading photonics customers. We also noticed improvement in more mainstream electronics and automotive applications. A push out to Q4 '25 of certain anticipated hybrid bonding bookings limited even stronger order development during the call -- during the quarter.

Besi's results for the first 9 months of 2025 reflected similar trends experienced in Q3 '25 with revenue of EUR 425 million and orders of EUR 434.6 million, decreasing by 6.4% and 6.5%, respectively, versus the comparable period of the prior year. In general, weakness in mobile and automotive applications this year has been partially offset by significantly increased die attach orders by Asian subcontractors for AI-related computing applications. Year-to-date '25, net income of EUR 88.8 million decreased by 27.6% versus the comparable 2024 period, primarily due to lower revenue, lower gross margins realized principally due to adverse ForEx effects, and higher interest expense net related to our senior note issuance in July '24.

Liquidity remained strong with cash and deposits of EUR 518.6 million, at September 30, increasing EUR 28.4 million or 5.8% versus June 30 this year, due to cash flow from operations more than doubling versus the second quarter of this year. In addition, we completed our EUR 100 million share buyback program, October '25, and authorized a new EUR 60 million program with an anticipated completion date of October 2026.

Next, I'd like to discuss the current market environment and our strategy. TechInsights currently forecasts assembly market growth of 1.8% in 2025, which is below last quarter's forecast of 9%, driven by a push out of the anticipated assembly upturn to 2026. Forecast growth is focused primarily on AI and data center logic and memory applications. TechInsights now expect cumulative growth in the period 2026-'29 of 42% based on continued advancements in AI use cases, new product introductions in the 2026-'28 period, and a cyclical recovery in mainstream assembly applications. We expect to exceed market growth rates given our leadership position in advanced packaging.

The semiconductor market has shown signs of normalization with inventory to booking ratios improving from above 2 in 2022 to below 1.5 currently. In addition, interconnect unit growth has also rebounded, improving from a low of roughly minus 20% in October 2023 to approximately plus 7% currently. These indicators point to a more positive assembly equipment environment as we look ahead to 2026.

Besi continued to make progress in its wafer-level assembly activities in the third quarter, securing new customers and orders for both its hybrid bonding and TC Next systems. Hybrid bonding adoption expanded with the placement of orders in the third quarter '25 by a new foundry customer. Progress also continues on integrated hybrid bonding production lines with internal operating 6 Kinex lines with 30 hybrid bonding tools. Future hybrid bonding demand is also supported by recent announcements from AMD and Broadcom in collaboration with OpenAI.

In addition, high-level discussions with major memory players are ongoing as HBM4 assembly processes start to take shape. TC Next progress continued with the new order received from a fourth customer. The outlook for Besi's business in the second half of this year has improved based on third quarter order trends and continued order momentum to date in the fourth quarter.

The improved outlook reflects increased demand for advanced packaging capacity necessary to support the rapid expansion of data centers, software and next-generation semiconductor devices required by the industry-leading AI players. Advanced packaging is one of the key ways to achieve AI system differentiation, develop innovative consumer edge AI devices and provide the most energy-efficient data center performance.

Now a few words about our guidance. For the fourth quarter this year, we anticipate that revenue will increase by approximately 15% to 25% versus the third quarter of this year, due to increased bookings levels. Besi's gross margin is anticipated to range between 61% and 63%. Operating expenses are expected to increase by 5% to 10% versus the third quarter due primarily to higher R&D expenses.

That ends our prepared remarks. I would like to open the call for questions. Sorry, operator.

[Operator Instructions] Our first question comes from Didier Scemama from Bank of America.

I just wanted to ask you a bit more about 2 things. The usual questions really. On the hybrid bonding and TCB Next side, maybe just give us a sense of your conversation with foundry customers, but also DRAM customers. How you're thinking about the bookings for those type of tools in the fourth quarter? And then also related to the point you made earlier on this recent announcement by OpenAI and AMD. We know that AMD has been a major customer of your hybrid bonding systems via TSMC. Can you tell us a little bit more about that whether the capacity is in place or whether you expect a material improvement in orders from TSMC to support that ramp?

Thanks, Didier. If you allow me not to go into specific customers, I'm very happy to answer a bit more in general terms. First of all, the hybrid bonding adoption for logic is continuing quarter-by-quarter. And we see that with adding another customer with several machines. And also as we guide for the fourth quarter, we expect orders. One is a larger one, as we have discussed also in the previous call and that may be related to those end products, which you just referred to.

At the same time, the adoption for HBM stacking is all pointing towards a critical evaluation year 2026. The big 3 in that market -- all 3 of them have publicly announced that hybrid bonded devices should be available in their program by the end of next year. So that will be, as we have expected for many years, 2026-'27, that should be the adoption time using this hybrid bonding technology with all its advantages versus the TC solutions, so reflow and clarity, we will share as quarters go on, and that should result in probably first orders for some initial capacities.

The orders received so far, as we shared in previous quarters are from 2 of the 3, who are evaluating in many different designs using the hybrid bonding technology stacking the 12 and 16, and they even go up much further, simply to achieve data on a comparable basis, on performance and of course, on cost. So that's the bigger picture in those 2.

Then the chiplet architecture, adding more different devices and different structures is also continuing. And we see ever more customers. So if you look at the total count now, around 16, having hybrid bonders for different type of applications, and all developing applications in early stages apart from the major volume in Taiwan and what we also discussed the capacity having been set up in the U.S. by one customer, but the others are testing, qualifying and publishing data on using the hybrid bonding. So that's for the hybrid bonding.

If we look at TC Next, the key issue in TC are basically there are 2 issues. One is going to a fluxes solution. Our system is prepared for adding that to the system. And at the same time, more accuracy required for bond pad pitches below 20 micron. Recent additional data has been published by IMEC in Belgium, that on our system, successful products have been refloat down to below 10-micron bond pad pitch, even 7-micron bond-pad pitch. So that should fill the gap between the necessary hybrid bonding and the reflow process as the world is using today above 20-micron bond pad pitch. And as you can see, another customer has been added, they all prepare for those 2 criteria required for next-generation TC.

Very well. As a follow-up, I just wanted to check also another thing. So my understanding is that this OpenAI AMD chip and let's forget the name of the customer, but it's 3, if not 2-nanometer design. So are you ready to ship your 25-nanometer accuracy system in support of that customer? And I've got a quick follow-up as well.

Well, for hybrid bonding, the current, let's say, the majority of systems shipped so far is 100-nanometer accuracy. And the 50 will be shipped for evaluation, qualification towards the end of this year. And that is in preparation for design structures below 2 nanometers as we understand from customers. So that's still some time out. Today, it's all 100-nanometer basically the benchmark technology used for many applications and not just for 1 customer. So we will see in the course of the coming quarters, a broader adoption for different types of devices and one of them has been announced publicly and the MI450. There's also a next one above 500, and they all use hybrid bonding as far as we are informed.

Perfect. And then my final question, Richard, at this time of the year, it depends sometimes it's in Q4, sometimes it's in Q1, you start to get a feel for some flagship smartphone design upgrades, which typically leads to orders for you guys? Any feel for what it could mean for the new models that come in the later part of '26. Could that be orders for you in Q4 or in Q1? Or is it too early to say?

No. The typical pattern for these new models is ordering Q4, Q1 with then market launch in September. So delivery of systems in June, for qualification July, August. So we should understand much more in Q4 and when we released the numbers in Q1 end of February, where this is heading. So if you follow the public domain, there's a lot of information about what will happen in this next generation, probably different cameras, also foldable. That means different design of the infrastructure in these units, and that could lead to a next round. But also this year in the generation for 2025, many let's say volume related, but also slightly new versions in these modules in these phones have led to a very positive business, albeit at lower levels than at the peak years 2021, '22.

So the key is to understand what really changes, our new machines required. And as soon as new machines are required, that means an extra round. Currently, you can simply conclude that a lot is being manufactured on systems already installed. And in many cases, retrofits have done the job in bringing successful the latest models to market.

Our next question comes from Sandeep Deshpande from JPMorgan.

My question is regarding your business with older customers, such as in the smartphone market, the autos market, et cetera. There seems to be some signs of life in the smartphone market. But the question I have is that will those signs of life translate into orders for you given that sometimes your orders are very much related to next-generation product and whether that needs to wait until the new product comes out next fall? Or is it likely to happen because of the volumes? Or is that too early to say because it can happen because of the volumes. And I have one quick follow-up.

Some are related to volume. So with the success which we have read in the public domain recently, that means, yes, more volume and that means shortages in certain areas and that is definitely helping. But as you said rightly, the key is to understand what are the real changes for the next model. And that typically becomes more clear towards the end of Q4, Q1, and then we have a much better understanding is there next year, going to be a ramp in those applications? Or is it at similar levels? That's typically how it has developed over the past many cycles.

Automotive, the developments in automotive are mostly at new power modules also quite complicated modules. They're all for hybrid cars. That's what we hear. Volume is still, yes, let's say, moderate, not any expansion to mention, but that is also clear from the announcements of the major customer in that space. So there you can say still the turn of tide is to be expected early next year. Let's hope so.

But yes, our success is always in these new technologies. So automotive, we mentioned also last quarter, we don't see further decline. We see a stabilization, and we see new products where we are included also new technologies for instance, in soft solder in bonding those high-powered devices. So that's typically what the status is in automotive.

One quick follow-up on the comment you made in your opening remarks on TC Next. You said you've got -- you've got an order for another customer in TC Next, I mean can you help us understand totally how many customers you have on TC Next? And is this, I mean, based on how you are seeing it play out. I mean clearly it's early days yet, but could this become a major new revenue stream for you?

Yes. Well, let's hope so that, that will be the case. That's what we are aiming at. There are 2, again, markets. You can say the logic markets, and that is where you first reach the smaller bond pad pitches, so below 20 micron. And that is where the concept the TC Next is aiming at the first place. But at the same time that system is uniquely capable to stack those dies in HBM application. And also it's prepared to add the units required for fluxless application.

The development is in both directions. So time will tell. And as I said earlier, '26 is going to be a critical year for adoption of hybrid bonding in HBM stacking, depending on what -- how that split will be at some point for HBM4. Most will be as everyone expects in refill process DC, but there are different variations in those processes. As we all know, the 3 use a different process, but that's less critical. So the machine typically for HBM3 is not -- that's why it's in an early stage, but an important year ahead of us to see where these applications can lead to major volumes.

The next question comes from Ruben Devos from Kepler Cheuvreux.

I had one on photonics. These orders, are they tied to I guess optic pilots? Or are they for, let's say, the pluggables inside the AI data centers? And...

Sorry, you had more part to your question or...

Yes. Well, just a follow-up on the photonics like these customers have resumed capacity purchases, I understood. Like is it for new platforms? Or is really expansions on the installed base? Yes, that's the first one.

Well, it's for pluggables. So the connectors in the data centers. And it's partly add-ons, but it's 5 customers as we explained in the previous quarter and they're all ramping up and very much on our systems. So we have a major market share in that area. So -- yes, we expect that also to continue in Q4 because it's all tied to data center expansions.

Okay. Just thinking about sort of the mix shift that has taken place since, let's say, 2021 when you sort of had the peak in mobile, I think it was 40% of your business and 20% compute and now approaching the end of '25. How do you think of that mix from what you've seen already so far? And particularly now in Q3, you see more momentum with orders obviously up particularly the OSATs ordering. Like how would you characterize maybe that shift mix today? And do you see, in general, like how do you assess the investment appetite basically from the OSATs now for compute as what it was maybe a decade ago in mobile?

That's a very good question. In a big picture, the world clearly for the last decade was very much focused on mobile. Every year, next generation, every 3, 4 years, a major, let's say, new, whether it's from 4G to 5G and before that 3 and then also the cameras and the movies and all that has been a constant driver. And that still is today. So you can expect the 6G, but also the connection to wearables. And what we haven't mentioned yet is the increasing development in wearables in the glasses. It started with Google glasses, now Meta glasses, where we are also very much involved. So you see that developing along -- yes, let's say, the development I would sometimes characterize in mankind using those devices.

Now we are in an AI phase, and that's more data using, again, yes, data in whatever more intelligent ways. And that is shifting then the percentage of revenue. Already last year, 43% was related to computing and data center high-power computing as opposed to many years before that, it was somewhere in the mid-20s. So that's all a very positive development. We were, for many years, characterized that we were very much dependent upon the high-end smartphone cycles. Currently, that is far less. We have more -- we have major drive in the whole AI world and with different technologies. So we look upon it in that sense with continued engagement in the forefront of the development of communication devices. And then we have automotive, which has dropped to below 20%, which was the average level, somewhere between 15% and 20%. So that's in a broader brush how our business is developing.

The next question comes from Charles Shi from Needham & Company.

Congrats on the pretty strong guidance for fourth quarter. Maybe I want to go back to the major hybrid bonding order you expected that could arrive in Q3, but now it looks like it's going to be a little bit later, given the push out. So the question was -- question is this, how much confidence you have in getting that particular large order in the current quarter because last time, I think, quite frankly, we were a little bit disappointed by the push out, but really hope that this time it's real and it's coming.

Well, you can also qualify that a bit in our own success in building machines. So at the very beginning, the throughput time to build these machines was over 9 months, closer to a year. And now since these 100-nanometer machines have become more standardized, we can turn them around in a 6-month period, which has the benefit for customers to align that more closely with their end customer demand and also the logistics. So what we understood is that the initial delay because of certain manufacturing or building construction issues at that customer. And that is why the placement is somewhat delayed. That is our current view supported by all the information directly from the customer.

Got it, Richard. So it sounds like 2 factors there, customer clean room delay and also the fact that you will improve the manufacturing cycle time. So they are not really -- they don't really need to place order well in advance, did I understand correctly.

Yes. So we are currently installing machines at that same customer and those machines have been built in 6 plants. And that is also one of the factors.

Got it. So maybe a little bit of a more technical question. Regarding the Gen 2, the 50-nanometer accuracy tool, well, you have been very consistent. I think over the last 2 years that you expect to deliver the tool, maybe the end of 2025, that time line hasn't really moved. But at the same time, people have high expectation about Besi and probably were wondering why the schedule didn't move up. And was that more of a customer road map issue or more of a little bit of technical challenges on your side? Can you kind of shed some light on what's happening there with the 50-nanometer tool.

And I think on a related question, I think when your schedule didn't really move in, do you worry about a little bit increased competition. I think on the HBM side, competition -- the landscape -- competitive landscape is well-known, lots of regional players there. But in logic side, do you see any increased competition there, especially at the leading foundry customer?

Well these are very good questions. The first question on the timing of the 50-nanometer requirement, that's purely customer road map. And that road map has not changed. The road map is '27 onwards. And so that tool has to be ready by the end of '26 as we have shared, that is not being pulled forward. The adoption of hybrid bonding is ever more confirmed and we see that with additional orders, additional customers. It's definitely logic oriented because that is where the most critical and the smallest geometries are requiring this technology. On HBM stacking, it is a bit less in a sense, the bond pad pitch is not that of a great issue. But there, it's more the heat factor, so the performance of the device, which is driving using hybrid as opposed to a reflow process.

On the competitive landscape, let's -- there from the beginning, a Japanese competitor has been already for 8 years sort of side by side. So far, our concept is certainly leading with a market share of over 80%, even some people say 90%. There has not been a change in that landscape. We have successfully moved the generation from 1 to 1 plus, so 150, 200-nanometer down to 100-nanometer. As far as we know, also from a cost of ownership, throughput, our system is certainly in the lead. On the HBM, it's a different competitive lens, more Korean based. Exciting will be in the course of this year, how the evaluations will, let's say, develop in terms of side-by-side comparisons that will take place in Korea at the 2 major Korean customers. So that will give us a better understanding of the competitive landscape. So that's in a nutshell, Charles, where we're at.

Got it. So in logic, no real change. In HBM, it's always a little bit of -- in some flux. But thanks for the color.

The next question comes from Andrew Gardiner from Citi.

I wanted to come back to the market slide that you put up every quarter in your deck. You've highlighted that tech insights have reduced expectations for this year. And I can see as well for next year, those have come down. I fully accept Besi is going to outgrow the market given your positioning in some of these areas. But expectations are pretty high out there at the moment for your revenue growth into next year. I'm just wondering if you can shed a little bit of light on how you are seeing things. You've talked about orders in the near term, but any indications from customers as to their thoughts into next year and what could help you to drive such outsized revenue growth into next year?

Well, always a very good reference is the order run rate. So if you look at the last quarter, third quarter, also our guidance in broad terms for the fourth quarter, that leads to levels, which, yes, quarterly run rates give an indication on a yearly model. If you look at our revenue, let's say, if you take the guidance for revenue Q4, you take the midpoint and you add it up with the first 9 months, and then you look at the run rate in orders.

And also, let's say, where those orders are coming from and you -- I think you also shared it in that sense. Then we are benefiting from a part of the market, which grows significantly more than the average assembly equipment market. So the TechInsights numbers are for the overall markets, and it could very well be that the mainstream market for less, let's say, complicated devices is growing far less than for the advanced, which has always been the case.

So based on -- yes, the current run rate, one -- yes, should see that development. And also with the adoption of hybrid bonding gaining more traction more broadly, and also TC for that matter. Yes, that's a bit different than the forecast, which you see from the TechInsights. But in this industry, I've never seen any forecast, which is on the dot. It's usually either much too high or it is too low, it's a difficult time.

If you also see this in respect of what's happening in the whole industry, and that in relation to the world, there -- yes, it's not that straightforward. Well, it's never been that straightforward. But anyway, so my message is our statements, we expect based on the current evidence and trends that we should be able to outgrow what is currently forecasted for the market.

The next question comes from Timm Schulze-Melander from Rothschild & Co Redburn.

Maybe just the first one. You talked about a new foundry customer to whom you've shipped a hybrid bonding tool. Could you maybe just provide some color about the application and just kind of how meaningful that might be? And then I had a follow-up.

We are not -- let's say, we don't know the end customer in particular, but it looks like it's more in the mobile space. So that is as far as we know. Systems are ordered. They will be delivered in Q1. So then we may well know more, but customers are pretty careful in sharing end customer and end product details. Even for many, we are not allowed to see it. It's usually with code names. So our service engineers also are not able to track that, and one can understand also the reason why in IDMs, it's a bit more yes, let's say, easy because they typically have their end products, but in foundries that is a high level of -- yes, let's say, secrecy, confidential.

That's really helpful. And then just you referenced an order booking that slipped and looks like it's going to track into Q4 in terms of just the readiness of the customer. Could you just maybe -- is that an existing customer? Is it a chip maker? Or is it a packaging subcontractor?

Well, it's a chip-making foundry, and it's an existing customer. So that's as much as we can share.

Okay. Okay. That's helpful. Because I think maybe one of the -- my last question. If we look at where the strength of sort of hybrid bonding engagement has been, it's been at those customers who are front-end chip makers and you've referenced a couple like TSMC and Intel. What would be the indication that the market is extending into subcons who don't -- the packaging specialists who don't naturally have sort of chip-making front-end capabilities. Is that something that we can anticipate sort of being in the 2026 time frame? Or is that really sort of a much longer-term kind of target that maybe follows whatever happens in high bandwidth memory?

The largest subcon in the assembly space has taken ownership of hybrid bonding about a year ago and is in the process qualifying devices for end customers. It is very likely that you will see that trend, which has happened forever. And also, you can see it, for instance, 2.5D modules. 2.5D modules are now built at a whole range of subcontractors, the typical, the higher-end ones, and that is where the growth in our orders in the third quarter was very much coming from. So for hybrid bonding devices, you can expect a similar trend. It may take a few years, but it's all a matter of cost, and that is a normal trend.

And as I said, you see already preparation because for those subcons, the high-end devices also offer the highest margin potential. So there's a clear win situation on both ends in reducing cost and that's the trend in many of our products. It starts at IDM and it moves gradually into the subcontracting arena.

The next question comes from Martin Jungfleisch from BNP Paribas.

Yes. I have 2 follow-ups from some earlier questions, please. The first one is on the hybrid bonding order. I mean would you stick to your comments that you made during Q2 results where you anticipated H2 hybrid bonding order to increase significantly compared to H2 '24? Or is there -- do you see now some orders to slip even into Q1 '26?

No, no, no. That's a very good question. It's very much as we said a quarter ago. So there are more we expect in Q4 to come in. So it's not just the big order, which slipped from Q3, hopefully, to Q4. But there are several other customers where we expect orders in Q4.

Okay. That sounds great. And then just secondly, on the 2.5D orders, I mean, you flagged this for the big increase in Q3. Just wondering, how sustainable are these order levels? I mean, is this driven by a single customer? Or is it multiple customers? And also what do you expect kind of this trend to continue into 2026?

It's multiple customers. We mentioned several times that it is a group of 5, which we have been -- several we have been engaged in since over 10 years. So it started off with [indiscernible] already a decade ago, routers. And that has developed in our smaller geometries now into data center connectors. So that is a business which is growing and it's not a -- we don't expect that to be a onetime.

But in capital goods, there's always this cyclical behavior. So you have a growth period, and then you have capacity absorption. But as we guided, we expect some continuation of this trend on the short term. But with the adoption of AI, and if you look at this in a broader perspective, again, what the world is expecting in the next couple of years, to do with the AI in every different form, these data centers is expected to grow significantly. And in case we are able to maintain our market position that should lead to continued business, albeit not in a straight line, but typically in a growth pattern.

That makes sense. Can you just tell me the lead time for the 2.5D tools? Is it similar to the mainstream market? Or is it more closer aligned to the hybrid bonders?

Somewhere in between. So we have -- that's also a good question. We can turn around equipment for mainstream in -- yes, some even in 6 weeks, 8 weeks. But this is typically 12, 16 weeks. That's why we cannot turn around the orders received in Q3 in the quarter. That's why the guidance 15 to 25 and up. So a major part will be shipped in Q1. So that's how it works. So we have machines which are more than a year -- or more than 6 months, sorry, with new developments, it's more than a year, but then it varies between the purchase lead times is 6 weeks. And yes, usually to 12 to 16 weeks, that is what the pattern. Next question, please.

We have time for one last question, and the question will be from Adithya Metuku from HSBC.

Firstly, I just wondered if you could help us get some more clarity into 2026. When I look at your revenue run rate that you've guided to for the December quarter or the orders run rate that we've seen in the third quarter,and annualize that, I get to around 20%, 25% below consensus in terms of revenues for 2026. So I just wondered if you expect orders to pick up further in the December quarter, or will it kind of plateau the high levels you've seen in the third quarter? Any color you can give and also any color you can give on how any other drivers we should keep in mind when we think about 2026 growth and that would be helpful. And I've got a follow-up.

Excellent. Well, first of all, we try to share in the press release that the order momentum continues into Q4. So Q3 is not the highest level. We also indicated that we see renewed drivers for growth in '26, which are linked to mobile, for instance, but also in the careful mainstream recovery where we see the early signs. But on top of that, we have the hybrid bonding continuation based on further adoption, and that could lead to a much -- yes, let's say, stronger growth in '26 than what we have so far in '25. So those are the -- and don't forget the TC Next. So those drivers could result in, as I also answered to an earlier question, in a business model more focused towards the high-growth AI arena. And at the same time, recovery for those applications where we have had in previous cycles, significant growth in new model usually applications.

So that's in a broad brush what the market could develop in '26, albeit in an environment which we all know is under -- yes, let's say, also different. China what we see is many customers are building next-generation capacities outside China. With the current geopolitical situation, you can expect new capacities built in countries like Vietnam, but also India. India, there are 5 major customers setting up assembly capacities, starting with direct product moves from what is currently built in China then built in India. That also offers additional growth in the change of infrastructure. So there are many aspects which can have an influence on how '26 will look like compared to '25. But we don't guide further than a quarter out. But since you ask what could be different in '26, then those are the aspects you can take into consideration.

Understood. And then just as a follow-up. I know last quarter, you talked about price negotiations in light of the recent adverse FX moves. I wondered if you could give some clarity on how those negotiations are going and when you might be able to get back into your 64% to 68% target range that you've previously provided?

Well, interesting enough, if you look at the dollar decline versus the euro with about 12% and a margin impact of around 3% gross margin. So we have been able to offset that partly in new features, which always allow higher pricing, but also in carefully managing our supply chain. And in that sense, the -- those developments will continue in an environment where the market is, you could say, soft. So -- and if this is the low part of the cycle, then you have a significant upside potential. Also, if you look at revenue levels, EUR 134 million this quarter, what was it exactly, which is -- our peak was above EUR 200 million.

Capacity utilization is, of course, at a different level currently. And that all has an impact on the gross margin overall. So if you compare this gross margin to peak levels, yes, the delta is larger than the 3%. I think once we reached 66%, we haven't reached 68%, it also depends on the order mix. There are certain new developments, which always have a somewhat lower margin. And over time, that improves because of, yes, the full qualification of systems. So those are all impacts on those gross margins. But still gross margins well above 62% is a reasonable margin at this time. Any last question?

I think we do not have any more time for any last questions, but I will hand the word back over to you, Mr. Blickman for any closing remarks.

Well, thank you all for taking the time. And if you have any further questions, don't hesitate to contact us directly. Thank you for attending. Bye-bye.