Thank you. 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 for our third quarter and nine months ended September 30 of this year and update you on the market, our strategy, and outlook.

First, some overall thoughts on the third quarter. Besi reported strong third quarter and year-to-date 2026 results as favorable order momentum continued for both our traditional and wafer-level assembly systems.

For the quarter, revenue of €283.4 million increased by 13.4% and 113.6% versus the second quarter of this year and the third quarter of 2025, respectively. Year-over-year revenue growth was primarily due to increased demand for computing end markets, with strength in hybrid bonding, photonics, and data center applications.

Similarly, orders of €342.6 million increased by 17% versus the second quarter of this year and 96.1% versus the third quarter of last year, primarily due to increased orders for hybrid bonding, data center, and AI-related mainstream computing applications, including power management. Besi's orders for the last 12 months increased to €1.2 billion as the expansion of digital infrastructure capacity significantly increased demand for our advanced packaging portfolio.

For the third quarter of this year, net income of €107.2 million increased by 20.4% versus the second quarter of this year and 323.7% versus the third quarter of last year. Strong profit growth versus the third quarter of last year was principally due to higher revenue, improving gross margins, and disciplined expense management, which has significantly limited overhead development in the current revenue ramp.

In fact, baseline operating expenses as a percentage of revenue decreased from 32% in the third quarter of last year to 16.4% in the third quarter of this year, underscoring the operating leverage in our business model. As a result, net margins increased from 37.8% in the third quarter '26, double the level of Q3 '25.

Year-to-date results also confirmed the strength of Besi's advanced packaging portfolio in the present AI CapEx upcycle. Revenue of €718.1 million increased by 69% versus year-to-date '25, primarily due to broad-based growth in computing end markets and, to a lesser extent, an increase in mobile shipments.

Some modest recovery in more traditional automotive, industrial, and non-AI computing markets has also been experienced. Year-to-date '26 orders of €905.2 million more than doubled versus year-to-date '25, due to growth across all end-user markets with strength in hybrid bonding, photonics, and data center applications.

Further, net income of €247.8 million increased by 179.1% versus the year-to-date '25, primarily due to significantly higher revenue and gross margins primarily due to a more favorable product mix and the pass-through of price increases to help offset higher supply chain costs. Combined with ongoing cost control efforts, net margins increased to 34.5% versus the 20.9% a year ago.

Our results also indicate significant market share gains in our addressable markets, particularly for Besi's flip chip, multi-module die attach, and hybrid bonding systems. Increased profitability and careful working capital management have contributed to steady improvement in our liquidity position this year.

Net cash reached €246.6 million at September 30, up €254.4 million versus September 30 last year. As a result, we will increase our share buyback program to €100 million for the annual period starting October 27 of this year.

Next, I'd like to discuss the current market environment and an update on our strategy. TechInsights recently upgraded its 2026 assembly equipment forecast to growth of 37%, almost double their last forecast.

They now expect the market to reach $12 billion by 2028, an increase of 122% compared to their last estimate of 53% growth. We expect to grow more than double their underlying market growth rate.

We can report significant progress this quarter and this year on our wafer-level assembly agenda. Year-to-date, hybrid bonding orders have more than doubled versus fiscal '25 as orders increased for HBM applications, more logic, chiplets programs entered production, co-packaged optics use cases expanded faster than anticipated, and edge AI device production began.

To date in 2026, hybrid bonding customer adoption grew to 23 customers, and total cumulative unit demand exceeded 215 systems. In the third quarter of this year, orders were received from customers for logic, co-packaged optics, and memory end markets.

Of note, we received orders for both hybrid bonding and TCB Next systems from a leading memory customer as they focus on the development, current, and next generation HBM device structures. Our competitive position in wafer-level assembly is very strong, further underscored by the progress we've made this year and inclusion recently of an innovation partner at Applied Materials' EPIC Center showcasing both Besi's hybrid bonding, TCB Next solutions.

We do not view ASML as a competitor, contrary to recent analyst reports. To the best of our knowledge, ASML is interested in a roadmap for 3D transistor stacking inside the chip over the next decade and is not competing with Besi in the area of die-to-wafer bonding.

In fact, ASML and Besi are jointly working together on die stress programs to improve placement accuracy and, consequently, throughput, as well as other mutually beneficial areas of collaboration. Now a few more words about our guidance.

The favorable year-to-date order momentum should continue in the fourth quarter as the multi-year AI capacity buildout progresses. We see strong demand continuing for data center and photonics applications from Asian subcontractors and market share gains in our addressable markets, particularly for Besi's flip chip and multi-module attach systems.

Our results also indicate ongoing hybrid bonding capacity expansion planned by leading industry players for logic, CPO, and edge AI devices. Severe memory shortages this year have extended the use of lower stack heights in current HBM production beyond prior expectations.

However, the need for higher power efficiency, energy efficiency, performance, bandwidth, interconnect density, and speed has only increased interest in the use of hybrid bonding for HBM as the year has progressed. We anticipate that hybrid bonding will be adopted in a range of stack heights over the next two years and in new, innovative stacks combining HBM, custom advanced logic base dies, and the direct bonding of DRAM onto accelerator chips by decade end.

Based on our current backlog and feedback from customers, we anticipate that Besi's Q4 revenue will be flat, plus/minus 5%, as certain customers request delivery dates for third quarter '26 received orders in alignment with H1 production capacity plans. In addition, gross margins are anticipated to range between the 63% and 65%, and operating expenses are expected to increase between 5% and 10% versus the third quarter due primarily to increased strategic consulting costs.

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

Good afternoon, Richard. Thank you for taking my questions.

First, I wanted to ask your views on essentially the adoption of hybrid bonding in HBM, because I think clearly expectations were for that adoption to be delayed, but at the same time, you clearly mentioned that you've had orders on TCB and hybrid bonding from one customer. How should we think about the adoption curve now for that particular customer and for hybrid bonding in general in HBM for your business?

Is that more of a 2028, 2029 story, or should we have some meaningful revenues already next year?

What can we add here? The opinions are as they have been expressed by you.

Anyway, let me give a response to that. As I mentioned, the development activity to use hybrid bonding for HBM stacking has only continued.

Despite some noise about despecing, that only relates to satisfying current demand in current end application. The next generation future demand is being evaluated using the benefits of hybrid bonding, as very well known, less energy, higher performance, and that roadmap development is continuing at full speed with all three memory manufacturers.

The precise dates of when certain applications will hit mainstream volume is not known by us. It's determined by the end customer of those customers mentioned, and we are only further preparing the equipment, processes with windows which can be used in those end applications.

Again, the activity and the focus and the commitments, also illustrated by the orders received, are in full speed continuing.

Thank you very much. Maybe my follow-up question would be on the hybrid bonding ramp at your leading edge logic and foundry customers.

I think in the past you had mentioned that your leading customer had about maybe 60 tools installed at their facilities and ambitions were for that's AP7 factory to have at least 100 tools or something like 100 tools by '27. I just wondered if you could maybe give us an update on progress of that ramp and also related to that, there's been press reports that another customer is planning to open an additional three advanced packaging fabs.

Just wondered if you've got any updated thoughts on the magnitude of those particular opportunities. Thank you.

The precise volume of equipment orders is always unknown at the start, depends on end customers and their applications. The news is certainly a very good one that tells you that they expect a significant further increase in demand of devices, and that's a ever broader range in logic, but also co-package optics involved.

Your earlier question, we reported we received so far twice as many orders as the whole year '25 for hybrid bonders. That tells you that the expansion is continuing.

Very clear. Thank you so much, Richard.

Hi there. Thanks for taking my questions.

The first is just a follow-up on Didier question on memory adoption. There are multiple reasons why hybrid bonding is attractive in memory, which you outlined in your answer.

I guess you could say that that's been true for some time now in terms of the benefits of adopting hybrid bonding. When you look over the next kind of two years, is there a key bandwidth requirement or IO density milestone or anything else that you think will make adoption particularly compelling to the memory industry?

Let me answer it in this way. As I said earlier, the current demand is satisfied as much as possible with existing technologies.

The consequence of that is that adopting superior technologies has been somewhat delayed. That does not mean that that results in a showstopper for adopting hybrid bonding for those applications.

That's the message. If that would be the case, all three of them would have stopped the development of adoption of hybrid bonding at some point in time.

As I mentioned earlier, it is only continuing and increasing, and orders for systems are being placed.

Got it. That's very helpful, Richard.

Then maybe second question is on co-package optics. You mentioned that demand's increasing there.

Is that predominantly from the co-package optic switches that are ramping, or is it early capacity build for optical IO for GPUs and custom ASICs?

It can be for both. We're not fully, let's say, informed about for which applications.

As we mentioned earlier, we had a quite substantial order in the first quarter for those applications, and also we have received several in the last quarter.

Great. Thank you, Richard.

Hi. Thanks for taking my question, Richard.

Good afternoon. Maybe my first question, yesterday, Micron said that they are still going to hybrid bonding and other than well-understood the benefit of thermal, they see there is of course a bandwidth benefit because of the smaller bump pad pitch, which ultimately determines bandwidth.

Throughout my conversations with average investors, they still kind of associate hybrid bonding with HBM stack height thickness. Really people are really obsessed about what's the timing of 16-high, 20-high.

Maybe I understand the de-speccing, there is some temporary near-term cross-currents there, but what should we be looking at right now to really understand the timing of hybrid bonding adoption in HBM? Related to that, can you give a little bit of color, this particular customer, why are they buying more hybrid bonders over the last quarter because I would assume all the top three HBM suppliers already have hybrid bonders from Besi.

Why are they buying more? Thank you.

We have not disclosed the customer ordering the systems, but in general, let me respond in that way. It precisely explains that it's not a matter of stack height which determines the adoption of hybrid bonding.

One should simply imagine a hybrid bonded interconnect is superior for the reasons as was just repeated, energy, latency, bandwidth, in other words, and heat. Those are the issues which they face to create whatever new structures, and whether that is 16 or 20 or 12, and we've mentioned that earlier in the call end of July, I think also end of April.

There's a whole roadmap which has the simple decision to make, do we use a conventional reflow process, TC process, or is the performance improvement and some released data about that performance improvement, 20%, 30%, also about the heat, temperature reduction, and those are the developments those customers are working on. Driven by their customer, the big end customer we all know, is dictating at some point, I want devices to be hybrid bonded because of better performance.

Does that answer your question? Yeah.

Yes, Richard. Maybe a quick follow-up.

Richard, what do you hear so far from your customers' customers who are already testing hybrid bonded HBM samples from one of your leading HBM customers right now? What do you hear from them?

A very positive feedback. We hear simply, and as I said to an earlier question, why would more hybrid bonders be needed to test, to qualify pre-production if there is not a high-volume production somewhere in sight?

I'm sorry, we have to answer it in this way, because we are not involved, or we don't have any knowledge of end customer demand and decisions. What we do know is that the activity and the pressure on making this high-volume ready is only increasing.

Thanks, Richard.

Thanks, Charles.

Great. Thanks for squeezing me in.

I really just wanted to go back to the delay in deliverables for Q4. I think as I understood things, there's a push-out with capacity expansion plan at one or two of your customers from Q4 into the first half.

Just trying to understand, is this a multi-customer issue? Is this maybe something that relates to the AP7 that you talked about, or something more broadly?

Then I've got a follow-up. Thanks.

No, no, no, no. Then we have presented this message incorrectly.

It's not a customer push-out. It is simply customers order equipment, and we are in a whole line a certain range, and sometimes and only bonder in a production line.

The timing of delivery depends on the readiness of all equipment for that process, and at the same time for the infrastructure. Although we have received last quarter €340-something million in orders, that doesn't mean that they are all shipped in the next week.

Some are shipped in Q4, some are shipped in Q1, and also some are shipped in Q2. That's a normal pattern.

In order to explain that that results in a more or less flat revenue has nothing to do with any delays. That's just how the orders have been placed.

Great. Very clear.

Maybe just a quick follow-up. It does seem as though with the margin guide that may have a one-off feature in Q4.

I wonder if those one-offs would result in future savings, and if indeed you could actually give us a quantum what those savings are. Thank you.

Please repeat. Where are you referring to?

We're just looking at the margin guides for Q4, under the impression that that is some one-off in there. We just wondered if there was a result

of

savings to come from that.

No, there're no one-offs in that. To explain to you, and I understood you are new in covering our stock, our margin guide 63% to 65% is the same guide we had for Q3.

It depends ultimately on the mix of the orders totaling the revenue. Sometimes it is a bit more, and sometimes it's a bit less.

There's no one-off effect. There is in the OpEx, we have a one-off extra cost because of consultancy.

We highlighted that and why the OpEx is slightly higher, 5% to 10% compared to the third quarter. On margins, it is purely a guidance which was the same as for the last quarter.

The mix may result in a somewhat in that range margin. Is that clear?

All clear. Thank you.

Thank you.

Hi. Thank you for letting me on.

My question is regarding your TC Next tools. You've got this order in the current quarter as well in Q3 for TC Next tools as well as hybrid bonding.

Is the customer ordering for TC Next and hybrid bonding going to make a decision between one or the other, or is these separate kinds of products they're going to use these tools for? I have a quick follow-up.

Sandeep, that is for two different product roadmaps of that customer. They're all involved in memory.

That explains also there's not one end product. The fact that we received this order, this also, by the way, a repeat order for TC, that also shows that we have an excellent workhorse for the next generation in TC next to our hybrid bonding roadmap.

Understood. When you look at the TC Next roadmap, how do you see this?

Clearly, hybrid bonding has already become a major product line for Besi in terms of revenue generation. Is TC Next going to become this major product line next year or the following year, or is the timing still being worked out with the customer in terms of adoption?

We have now in total seven customers. This tool is developed for bond pad pitches below what is currently the market standard, and that is in the 20s.

This is below 20. At the same time, it's developed for fluxless processes.

The expectation is that for the next generation, somewhere between 10 and 20 micron bond pad pitch, that this will be, let's say, excellent for not only memory stacking, but also logic. Most customers ordered so far logic.

We shared data produced at imec last year on this machine where even below 10 micron bond pad pitch, and that comes very close to hybrid bonding, this machine gives the option to the customer to use either a reflow process or hybrid bonding if needed. That is where it's intended for, and that segment is opening up, and in the next couple of years, it could be a very sizable product range.

Thank you.

Hi, and thanks for taking my question. I guess just following up from the comments about ASML.

I know you can't talk too much about this, but can you just sort of talk a little bit about the IP you have around handling and the uniqueness of it? I think there's still this idea that even if they go into something like wafer-to-wafer, they'll ultimately transition to die-to-wafer, it seems like there's quite a lot of misunderstanding.

If you could just clarify that and how secure your position is in the die side. Thanks.

What we try to make clear, we have a very good relationship with ASML. What we understood is their strategy, as they explain to the market, is to focus on 3D integration using a hybrid bonding process once that becomes a mainstream market requirement.

They are not interested, as we stated, in the current market for two reasons. One is simply the size of the market, and also the current situation in this market.

Together, we are developing programs to further understand die stress. What is die stress?

Every die has stress for the reason that it is built in layers, and once it's cut, it has another stress introduced, and that stress determines the placement accuracy and also other technology you need for the placement process. Other than a quick conclusion, ASML, because of its size, will in the next couple of weeks take over Besi's market position, the situation is entirely different.

One could argue this stacking of transistors into dies, the 3D development continuation of Moore's Law, that will require accuracy placements below 10 nanometers, where today shipping this quarter the first 15 nano, which is by far ahead of the current curve, 100 is more or less the standard, that's all the 215 machines sold so far is on that generation, generation 1 and generation 1 Plus. 50 is generation 2, but then next to 50, which will be sufficient for the next four or five years, is 25, and 25 is currently under development, which we are definitely able to reach, and then the next step beyond that is going down to 10.

That's all still chip-to-wafer bonding, and that is not chip-in-wafer bonding. That's a step beyond that.

There's still many years ahead of us on further development of this technology in line with further shrink in chip design.

Could I just quick ask a follow-up just on automotive and industrial? In the past, you had quite a strong from that back in '21, '22.

It seems like the mature side is getting very tight now. Have you seen the beginnings of a sort of a cycle on the mature side?

Do you think that's sustainable? Thanks a lot.

Whether that's sustainable, that will probably follow the same cyclical pattern as the industry has done forever. There are certain signs of pick up.

We see that in orders from subcontractors. We see that in orders more broadly.

The key is of course, and we try to explain that in the press release, our business is driven by the high end of all of the applications, whether that is submicron or whether that is in the industrial market. You always have new developments.

Power today is hot, because that power is also used for data centers, and that's older, you could say, technology in brackets. It's also across the board.

It's picking up. That's reflected in the updated TechInsights forecast for '26, and that should continue in '27 if the cyclical pattern is similar to that which we experienced in the past.

That then explains that they expect a continuation of this growth cycle in the next year.

Thank you.

Hi, Richard. Thanks for taking the question.

Just as a follow-up on your comments regarding ASML. I was just wondering if you could provide some more color on your nature of the partnership with ASML.

Sounds like you're working together to solve some of the problems that you're seeing in die-to-wafer bonding. Where is the R&D taking place?

How does the relationship function in practice? Thanks.

Oh, this is an, let's say, ongoing dialogue for many, many years on different programs which have to do with and in this case understanding die stress. The whole critical nature of manipulating die, to speak, you have to pick up a die, you have to place it, and you have to place it ever more accurately, is to understand what are the reasons for the behavior of that process.

It takes place inside our development, inside their developments, also at other locations here in Europe. You should see that on a program basis and an alignment regularly on roadmaps.

That's very clear. Then just a follow-up.

I think you mentioned that targeting different applications, more intra-die 3D transistor formation. Could you talk about what's fundamentally different between bonding or hybrid bonding with intra-die 3D transistor formation and the die stacking that Besi currently looks at?

Are there fundamental challenges that are different that you and ASML are looking to solve here? Thanks.

We connect dies. Our job forever is to connect dies, chips which are coming out of front-ends and are ready to be connected to something.

We are not engaged yet in the chip manufacturing process. The angle we understand ASML is looking at is how to continue the roadmap in chip design and manufacturing.

We are at the end. We connect the dies so far.

You can connect dies with a wire, you can do that with a mass reflow flip chip, you can do that with a TC process, a bit helping with a bit of pressure, you can do it with hybrid bonding, and that's all manipulating individual dies. Is that more clear?

Very clear. Thank you.

I should also add we have this angle with Applied Materials. Simply imagine Applied process, front-end, how to make those dies are running into the same challenge in ever smaller design geometries than finally connecting those dies into end products.

The foundation behind our relationship with Applied, already for seven years, and now in the EPIC Center, where the first program we will set up is both for hybrid bonding and TC for a major US customer. That is the reason why we come ever closer.

Hi. Good afternoon.

Just have two quick ones. First one is on the order trends in the press release this morning.

You mentioned that you would expect the order momentum to continue into Q4. Is this primarily driven by the OSATs demand for 2.5D and photonics?

Or is this also driven by hybrid bonding demand of the leading foundry? Then is there any reason to think that there is some digestion period now for hybrid bonding at the leading foundry after the levels in the first nine months, or would you not agree with that?

Thank you.

No, no. That's an excellent question.

The answer is that we don't expect the momentum to decline. The momentum should continue.

We are not in control of exact order timing placement, but as we said in our press release and also in the comments now, the momentum is still very strong.

Okay. That's in both, right, classical

OSAT

and then hybrid? Interesting.

Thanks. Then my follow-up is really on the new tool release in photonics.

I think you had initial target of the submicron photonics tool for Q4. Is that still expected in Q4?

Then maybe also if you have some initial customer interest or orders for that tool. Thank you.

As we said in general terms, there is increased activity all the time for hybrid bonding and certainly for those applications. As we highlighted, so far twice as many orders as what we had in the whole year '25.

A part of that is also the co-package optics part.

Yes. I was also referring not to the hybrid side of it.

I was referring to this thing that you call the Evo

1 that you originally

supposed to be Q4. Yeah.

Exactly, exactly.

Yeah. For the pluggables, the same, continued momentum.

That is directly related to data centers, huh? It's also a very good question because if you would start to see a digestion or slowdown, you should ask the question, is this a first sign of an overall overcapacity or is this a sign of a timing issue?

Those are the critical signals to understand. Far, until October 22, the order momentum continues.

Okay. That's good to hear.

Thank you very much.

Hi, Richard. Thanks for taking the question.

It's just on the strategic review. The last did one, I think, was in '25, and then we got a nice upgrade to the hybrid bonding targets.

Could you just give us a bit more color on where this one is focused? Is it on some of your other tools that are maybe not getting as much attention as they perhaps could get in this market?

Thanks.

Sorry, the line was very difficult. Can you repeat it one more time?

Sure. Can you hear me better now?

Yes. Try again.

Please.

Just on the strategic review. The last time you gave us a strategic review, we saw some upgrades to hybrid bonding TAM numbers.

Could you just give us some color on where strategic review is focused? Is it around some other areas of the business that could be getting some more attention?

Perfect. Why do we do this review?

Your question is also very important. We did the last review beginning of last year, which we then shared with the market.

Much is happening in this industry in this last 1.5 years that we are engaged again in a 16-weeks deep dive with our key customers on understanding even better their roadmaps and being prepared to serve them not only on the short term, but on the longer term. One of the questions is, of course, hybrid bonding, TC, and mass reflow.

How, which roadmaps, and timing is going to be determining the demand in the next 3 to 5 years. That focus has always helped us.

It helped us for several reasons. One is, of course, to understand better what is the market requirements, but at the same time, the customer engagement always increases when you do this kind of process.

Also involved is of course Applied Materials and simply to understand on this whole advanced packaging, what can we do more? What is missing?

What can we do in partnership and better development? Those are the key deliverables to be expected in the next let's say 16 weeks.

We're already four weeks down the road, early January, we should come to a conclusion.

Great. Thank you.

Yes. Thanks for the question.

First question is on a lot of your peers are talking about increasing longer visibility, longer lead times with the clients. Is that something that you're also seeing?

For example, the element that you have very good order intake is in this quarter, but some of it will be for the first half next year or is that just not directly linked to that longer lead times?

Yeah. It's inevitable with this enormous demand.

We are able to grow nearly 70% year-on-year, which is a lot more machines. Also, customers, because of longer lead times of other tools in the process, and that's the reason we explained also earlier that you can't ship immediately what you ordered in Q3 and Q4 because also of that reason.

That's a natural development. If you look at the front-end companies, supposedly their lead times have gone much further down the road, even two years.

We are able with our operating model to increase the capacity rapidly, as mentioned, the 70% is simply because of a supply chain model and operating model which allows that. The question could also be, could you do more if you can deliver more?

For some cases, it may be. Overall, we're more because the industry needs to increase the infrastructure.

Also, as mentioned earlier, the supply of a complete line. There's a bit of longer visibility.

The visibility in lesser years behind us was about a quarter, 1.5 quarters. Now the visibility is longer.

It's a quarter, two quarters, six months, and for some programs, even longer.

Okay. Very clear.

My second question is on the power management business opportunity. You highlighted for the second time, I think you also did it at the previous quarter.

This seems like a quite sizable opportunity given that it's quite a big change towards the higher voltage, yeah? Can you maybe explain a bit more where exactly does Besi come in and how big can that addressable market be from the perspective of Besi?

I can explain it in this way. Power has always been part of our business and especially for automotive.

That was forever, and still today, is a reasonable segment. We always said between 15% and 20% of our revenue is related to automotive.

That has declined because of the enormous increase in AI data centers, somewhere around 10%. If you look at power increase, part of that AI data center modules, that certainly brings that back to the levels which overall automotive was, including automotive then.

The equipment used for that is typically copper frames and power devices soldered to those frames, packaged, and in all shapes and sizes. Next time you visit Duiven, we will show you some examples.

Great. That's clear.

Thanks.

Hi, and thanks for taking my question. My first question is on hybrid bonding adoption in logic and specifically on your US logic customer, because he didn't order at least at a significant level for some time now.

I think he's facing CPU shortage. I was wondering how do you see this customer potentially increasing its capacity for hybrid bonding in the following quarters on that CPU demand?

I have a quick follow-up.

Excellent, Martin. Thank you.

You're absolutely right. That customer ordered a large volume order in '24, it was even, and that is shipped in '25 and installed and full preparation now for mass volume production.

The question is, when is that filled, and is extra capacity needed? With the success of that company regained, one can expect further expansion in next year.

What will be interesting to see, their original plan was to set up also operations in Asia, which is very close to our manufacturing facility. There're certainly plans to further expand.

Okay. Very clear.

The second one is more on the GPU opportunity. I think you previously described GPUs as potentially an inflection point for hybrid bonding demand in logic with the leading GPU player adopting it.

I think qualification was supposed to be this year, possibly orders next year, and volume the following year in 2028. Is that timeline still valid, and how big do you think that opportunity could be?

The timeline is still more or less as you have remembered. The volume is hard to predict.

If we would mention an indication, it would be a guess.

Understood. Thank you, Richard.

Thank you.

Yes, good afternoon. Thanks.

I was just curious around your Applied partnership, right, with the announcement a few weeks ago talking about a bit more TCB scaling to the die-on-panel and some new hybrid platforms. Should we expect that to turn into new, let's say, jointly offered systems, something like the Nexx type integrated lines for TCB specifically or die to panel?

If so, could you roughly talk about a timeline when you think a commercial product could be offered here?

Let me explain it as follows. The Applied Nexx you need for achieving the maximum cleanliness, particle-free processes guaranteed, and that is hybrid bonding.

Of course, you need a certain cleanliness with a reflow process, a TC process, but that's far less critical than with hybrid bonding. It's not aimed at developing Nexx type of solutions.

It's aimed at developing bonding processes. How can we create, and that's a question of material and also the preparation for that, the most superior TC bonded device?

There's TC in several ways. There's also copper-to-copper bonding using a TC process.

Those type of developments is expected to take place in the EPIC Center in Sunnyvale for US customers directly. On top of the development, we have the TC development in imec in Belgium.

We have TC development in Radfeld. We have TC development in the Center of Excellence in Singapore.

Supporting the development of next generation, and that's all smaller bond pad, is all related to fluxless, developing superior processes to the processes used today. Applied has immense material knowledge, huh, Applied Materials, also enormous experience in preparation, in cleaning, and those type of processes are vital not only to hybrid bonding, but also to other interconnect like TC processes.

I hope that answers your question.

Very helpful. Thank you.

Just a final one on China. It went to 40% of orders, and it drove the majority of the sequential increase, I think.

What are Chinese customers mainly buying now? How confident are you that it's sort of tied to end demand rather than maybe a capacity build ahead of demand coming in?

Thank you.

There's a whole broad spectrum of Chinese subcontractors. Don't forget a lot of the high-end smartphone infrastructure is built in China.

That business is very much the Foxconns and many others. It's not just one application.

There's also developments in TC, development in hybrid bonding. To your question, is that something which we expect a limited future?

No, the answer is it will have a long-term future. There are many devices built in China and ever more.

Okay. Thank you.

Also, by the way, for the pluggables.