Cohu Inc

NASDAQ:COHU

Good morning, good afternoon, ladies and gentlemen, and welcome to Besi's Quarterly Conference Call and Audio Webcast to discuss the company's 2025 first 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 whole or in part without written permission from the company.

I will now turn the call over to Mr. Richard Blickman. Please go ahead, sir.

Thank you. Thank you for joining us today. We will begin by making a few comments in connection with the press release issued earlier today and then take your questions.

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 view 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, 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. For today's call, we'd like to review the key highlights of our first quarter ended March 31 of this year and update you on the market, our strategy and the outlook.

First, some overall thoughts on the first quarter. Besi reported solid first quarter results important new advanced packaging orders in a challenging market environment. Revenue of EUR 144.1 million was down 1.5% versus the first quarter last year due to the ongoing weakness in mobile and automotive end user markets, partially offset by strong revenue growth from hybrid bonding and other AI-related computing applications.

In contrast, orders increased 3.3% versus the first quarter last year and 8.2% versus the fourth quarter last year, due primarily to increased bookings by Asian subcontractors and AI-related data center applications. Besi's profitability in the first quarter of this year remained at attractive levels despite ongoing weakness in mobile, automotive and Chinese end user markets and expanded R&D investment in next-generation assembly solutions.

Net income of EUR 31.5 million decreased 7.4% versus the first quarter of last year, primarily due to lower revenue and gross margins realized, partially offset by an 8.9% decrease in operating expenses. Besi's gross margin has trended towards the lower end of our target range over the past 3 quarters due to primarily a less favorable product mix, particularly with respect to high-end smartphones and net ForEx and headwinds beginning in the second half of 2024 from adverse movements in some of our principal transaction currencies versus the euro.

On a sequential basis, Q1 '25 operating expense growth of 10.3% versus the fourth quarter of last year was due to higher consulting costs. It was at the lower end of guidance as we continue to control overhead development despite increased R&D investments. In addition, cash flow generation remains positive with net cash increasing by 10.8% versus the fourth quarter last year to reach EUR 159.4 million, which we presented 26.3% of our latest 12-month revenue.

As such, we continue our highly attractive capital allocation policy. In Q1 '25, we repurchased EUR 22.1 million of our shares, bringing total cumulative purchase to EUR 51.4 million under the current EUR 100 million program. Further at our AGM today, shareholders approved the proposed dividend for 2024, which will distribute EUR 172.5 million in cash to shareholders, which equals EUR 2.18 per share.

Next, I'd like to speak a little bit about the current market environment and our strategy. TechInsights currently forecast a 13% assembly market upturn in '25, followed by a 26% increase in '26.

Most analysts continue to expect significant growth in advanced packaging for AI applications in 2025 and the second half upturn in mainstream assembly markets. However, the slope of the current trajectory anticipated for mainstream markets this year is subject to many variables, including the potential impact of tariffs on a global trade.

Despite near-term uncertainty, we believe that the long-term fundamentals of our market remain strong due to the ongoing AI build-out, onshoring of advanced packaging fabs in the U.S. and Europe and increased adoption of hybrid bonding and other next-generation assembly systems for new AI, logic, memory and consumer use cases.

Of note, significant progress was made on Besi's wafer-level assembly agenda this quarter as we received hybrid bonding orders for 2 leading -- from 2 leading memory producers for HBM4 applications as well as follow-on orders from a leading Asian foundry for logic applications.

Further important announcements were made by 2 leading semiconductor producers with respect to future hybrid bonding applications such as ASICs and co-packaged optics. In addition, a leading U.S. logic manufacturer successfully began production of AI-related logic devices, utilizing Besi's hybrid bonders in integrated production lines.

Finally, Applied Materials announced on April 14, a 9% ownership position in Besi. Besi and Applied Materials have been successfully collaborating since 2020 to develop the industry's first fully integrated equipment solution for die-based hybrid bonding. The collaboration brings together Applied's expertise in front-end wafer and chip processing with Besi's leadership position and bonding, accuracy and speed.

We welcome their shareholding as a strategic long-term investment and view it as further validation of our wafer-level assembly technology and strategy.

Now a few words about the guidance. Besi's business development this year reflects the contrasting growth trends seen in the assembly equipment market between AI and mainstream applications. The timing and trajectory of a mainstream assembly upturn is more difficult to predict now given new tariff uncertainties. However, demand for advanced packaging for AI applications remains strong, given upcoming new device introductions and use cases planned in '26 to '28 time period. We continue to assess the potential impact of tariffs on Besi's customers, supply chain and end user markets.

For the second quarter '25, we forecast that revenue will be flat plus or minus 10% versus the first quarter of '25 with gross margins in the range between 62% and 64%. In addition, aggregate operating expenses are forecast to decrease 0 to 10% versus the first quarter this year, primarily due to a reduction in consulting costs.

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

[Operator Instructions] We will take our first question from Nigel Van Putten, Morgan Stanley.

I got a question on the hybrid bonding orders you've received for HBM. If my understanding is correct, both customers already received tools a couple of years ago and were also working with you in Singapore and Austria. So now they put in new orders. Can you provide some context what you think that means, what the stage of development they are in? And if you have any clarity on their intent? And I have a follow-up.

Well, as we all know, the preferred technology to stack memory is a reflow process. DC, thermal compression used by all 3 major memory manufacturers today because the cost is simply more advantageous than using a hybrid bonding process.

Hybrid is better because it has less metal in between the contacts because the reflow process is reflowing a soft metal and driving current through a circuit, you need simply more current, and that means that the performance of a hybrid bonded device is better theoretically than a reflow bonded device.

The cost is also a factor 4, some people say 6, higher using a hybrid process compared to a reflow process. So nothing new in the past year, exactly a year ago. JEDEC released their standards to allow 16 stacking of devices using this reflow process. It resulted in a height above the standard. So the industry has been developing, which is very, very understandable, this technology, which is proven in the next generation.

And at the same time, they are developing the hybrid bonding process as an alternative, number one, for a better performance, less energy consumption, less heat dissipation required. And so all of them are testing this hybrid bonding as a next preferred solution. And you can see in this quarter, continued testing and also taking the latest machine.

So we're talking about single machines, not a whole production facility, but these are testing machines, which should provide the answer in the second part of this year is for HBM4, a certain part using a hybrid bonding and another part using the conventional process, TC process. So it's important to understand what this means. It means simply continuing development, testing the benefits, understanding better how we can impact the cost and in preparation of a next step is stacking 20 dies, where it is likely that a reflow process has hit certain limits, and then hybrid bonding is the only way to go. So that's in a nutshell where we are.

Understood. That was actually very helpful. I do have a quick clarification, I guess. You just repeated, customers are still looking at sort of a dual parallel stack for 4 or 4e, 16 high. I think you're involved with at least one customer on both sides, also TCB.

So maybe can you provide any color on your TCB developments within memory? And also where you think sort of both tracks stand? I think you sort of implied that there may be some testing going on already at 16, but the big implementation is beyond that. Did I understand that correctly? And then also, how do you look at your Besi's TCB venture?

Well, our TCB is a continuation of focusing on this reflow for the past 25 years. And you have seen in this first quarter, a significant increase in demand for mass reflow flip-chip machines. TC is used for those processes where mass reflow is too critical to accomplish with smaller bump at pitches.

So we entered the TC world 12 years ago. And we have developed 6 years ago, a next-generation TC, which can produce at smaller bump at pitches than the current market suppliers. And that process is coming into the window of those customers gradually as we reported in the last quarters.

So we have said all along that the first half of this year, '25, should also make clear whether this next-generation TC is being adopted for HBM4 to a certain extent. So that will be exciting to understand in the next few months, what will happen. Besi has in its offering, so all 3. So we have mass reflow flip-chip for many, many years and today, very successful. We have TC, which is also very successful in focusing on the next generation.

We shipped another system this quarter, which already has produced data for bumped pitches well below the current next generation, so below 20-micron bumped pitch, data has been produced below 10-micron bumped pitch. So that system comes very close to hybrid bonding. So to offer the customers both solutions in a reflow process and the most advanced hybrid bonding process. But this year, again, will be a very important year to understand how much of HBM4 will be a refloat and which are the machines used in the industry by which competitor? And then which part may be hybrid bonded, is that HBM4e or is that for the next generation? So we will update you quarter by quarter. Also in the next months, we have the Capital Markets Day. So an exciting development in the memory front.

We will take our next question from Charles, Needham & Company.

Richard, congrats on the 2 additional hybrid bonding orders from 2 memory customers. It sounds like you are expecting maybe they will be qualified for HBM4e generation, the earliest. And the -- but if I look at NVIDIA's latest road map update, the Rubin, Rubin Ultra HBM4 will come to the market second half '26. HBM4e is like second half '27. Yes. And memory makers probably need to start production a little bit before those NVIDIA product availability time line.

So I wonder what's your thought on getting your hybrid bonding tools in right now is that there's enough time for you to get qualified for the HBM4e generation? How much confidence you have in that timeline?

Well, that's, of course, a very important question. So there are 2 important factors. Number one is what is the window of introduction that is often up for revision because it is not only the bonder, but there are many other processes. And number two is the competitive landscape. What is available by what time. So as I said, also in response to the earlier question, it will be very exciting in the next couple of months to see how the industry will fill in this demand and which processes will be used. That's as much as I understand today.

Maybe a second question, Richard. I think in your earnings press release last quarter, you talked about maybe a second half recovery for the mainstream business, but obviously, the language in the press release in the current quarter turned a little bit more cautious. So you're no longer saying you're expecting maybe a second half recovery. Wonder what you think about the potential revenue trajectory for the mainstream business can be?

Do you think at least the second half can be flattish over the first half or any additional color on the second half would be great. Maybe let me add one thing. The reason why I asked this is when I look at the consensus revenue numbers, it looks like a lot of my peers are modeling a much bigger second half revenue estimates. And what's your thought? And any color you can provide would be great.

Well, it's I think fair to say that since February '20, and April '23 in the world, a lot has happened. And there are major, major impacts to be, let's say, expected or considered, however you want to phrase that, of these different tariffs. If you look in a historical perspective, any disruption to an economical situation is for our industry, 9 out of 10, I would say, 10 out of 10 has a negative impact. So more caution.

And that is why we are, let's say, probably more careful than simply following the statistical trends in this industry that often results in when things have crystallized that you may have an even steeper recovery.

So that could also very well happen. Take the 2008, '09, the financial crisis as a reference or take for that matter, 2019, '20 and then the impact of COVID. So these impacts are first causing some caution and then after that, again, a stronger increase.

So our question is because of this phenomenon. We also see that at customers. We mentioned in the call, February '20, that already before the Trump tariffs, there was a big -- yes, let's say, slowdown in automotive, also disappointing in high-end smartphones, not a big, big year to be expected.

So why with all this tariff uncertainty, would all of a sudden the market look better in the second half. We do see and you see that in our order intake, some improvement year-over-year, quarter-over-quarter. But as long as it is single digit, it is not a very strong upturn. And when you would see second half of this year, a major uptick, you should see orders increase dramatically in the first and the second quarter because there's a lead time.

So in a longer answer, Charles, I hope that you are right that we do see a strong recovery in the second half. Besi has demonstrated we can ramp very quickly. We're also prepared for that, of course. But in the current state of tariffs and economies and a lower dollar, that usually has a negative effect.

We will take our next question from Didier Scemama, Bank of America.

Yes. I just had a question on the first quarter order. Did it play out the way you expected? Or did you see any pushouts from anyone, whether it's the IDMs or the subcontractors? The reason why I'm asking is because it's really in response to the previous question. It's really because if the customers are worried about tariffs in some respect, they might pull in as they see -- as they've done probably for semis.

But when it comes to back-end capacity, they probably pushed up, right, not knowing exactly how it's going to play out. So if we were to have a bit of more clarity on those tariffs and who knows that might happen, do you think that they come back quickly in Q2 to actually catch up for the seasonal uptake for smartphones and PCs that you normally see, especially smartphones for Besi? I've got a follow-up as well.

Yes, that is what I tried to explain just now. If you look back in whatever downturn scenario we've had in the past many, many years, you have a hesitation because our customers don't, let's say, have the confidence in the end market demand. You sense that it takes longer. Also take the Intel specific situation, Intel restructuring, whether it's announced or not, but people expect a major, let's say, restructuring impact that usually slows down orders for some time, but then it catches up. But it always -- we have this nice slide in the deck going back to 2006. I invite you to look at that quarter-by-quarter. And you see those impacts in every cycle along the way. And always the recovery is very steep. The question is, however, will that be in the second half or the first half of next year, but it will recover.

No, no. I get that. Sorry, trying to be a bit more precise on that. Did the quarter play out the way you expected e.g., you miss by EUR 10 million, EUR 20 million? Or just help us understand how much you think your orders miss, so we can try to understand how much it can sort of recover if and when Tariffs are a bit more clear.

Well, revenue was fine. It was okay. It was also more or less in the mid of the guidance. So that was not disappointing. Orders, I just phrased it in other words, but we feel some delay. How much that delay is, whether it's EUR 10 million or EUR 20 million, it could very well be.

On the other hand, look at the orders and the guidance, so we received in orders much less than what we guided. So we expect some orders still to come. And that will -- if all goes well, may well have an upside that we guided plus/minus 10% on EUR 144 million. The order intake was much less. So that more or less gives you the answer, Didier, but the bottom line is that we feel there is some caution. Let's put it that way.

Yes, I think it makes complete sense. And sorry, just as a quick follow-up. On the hybrid bonding side from foundry customers, you said you saw some follow-up orders. Can you elaborate a little bit? Is it coming from AI applications, computing applications, co-packaged optics? Just give us a sense of what sort of the magnitude of that contribution to revenues in the second half, please?

Well, we've said all along that, that customer in Taiwan is not very transparent in terms of the end application, the end application we hear from their customers. So whether that's AMD or whether that's NVIDIA. NVIDIA, as we all know, is developing very much on this co-packaged optics. We also know from high-end smartphone applications where certain modules are being assembled using the hybrid technology.

So there is an ever-growing application field using hybrid bonding. Currently, a fleet of 50, a few more systems will be added August, September. If all goes according to plan, there's a big program for '26. There's a site announced advanced packaging, which potentially is twice the size of the site, which currently is operating the hybrid bonding systems. So they are big plans and a lot of development, and that should kick in somewhere '26. So then you should receive orders in the later part of '25. But anyway, that is what we understand today.

We will take our next question from Robert Sanders, Deutsche Bank.

I just had 2 questions. Firstly, if you could just talk a bit about Hynix. They're in a quite public dispute with Hanmi at the moment. I was just wondering if that could potentially create an opportunity for you either in TC or hybrid bonding. And then the second question would just be if you could just confirm that you had mid-single-digit orders in the quarter, just to help our modeling for hybrid modeling, that would be great.

Yes. The last is even more precise. Yes, in the mid-single digit. On the Hynix situation, and that is public knowledge, they are like the other 2, developing both technologies continuing using the MUF process and reflow. And they are also continuing to test hybrid bonding solutions. They are also preparing local supply and a second source. So that fits into the total picture that '25 may well show some direction in those 3 continuing to use reflow, plus starting to more seriously develop hybrid booming stacking.

So partly relying on our -- the AMAT Central of Excellence in Singapore, also development in Radfeld, and that's where it stands. On the dispute you are mentioning, yes, there's -- and we've mentioned that in every call, the current landscape of TC bonders is able to produce with a certain bumped pitch and a certain process, but that's not good enough for the next generation. It needs to be a tighter bumper pitch. It needs to be fluxless, then which fluxless solution? Is it a plasma solution? Or is it some type of chemical solution? That is currently a full swing. And some are more advanced in their solutions offering than others.

We will take our next question from Adithya Metuku, HSBC.

So firstly, Richard, maybe when I look at the road maps for AI accelerators, Rubin Ultra, which is meant to ship in the second half of 2027 is expected to be the first accelerator that uses 16 high HBM. Until then, the market seems to be dominated by 12 high HBM. So given this backdrop, even if hybrid bonding gets qualified for 16 high HBM despite all the short time lines, would it be fair to assume that hybrid bonding will not be used in volume production until 2027? Or do you think it could still be earlier maybe with some versions of -- some test versions of 12 high HBM? And I've got a follow-up.

Yes. Well, the way you summarize it is pretty much how it is presented to us. And again, we come back to the basic reason why using hybrid bonding and stacking is simply the performance. So is there a differentiator using a hybrid bonding process, which is more expensive compared to a reflow. And that is up for the debate at each of the 3 producers. And that's where we are continuously doing tests in the hybrid, but also in the TC.

So in another way of putting it, it's on everyone's mind, but how it will sort itself out is not yet clear.

Understood. And as a follow-up, you mentioned earlier that smartphone applications are looking at hybrid bonding. TSM on the last call, not the one this one, but in the previous quarter, they were specifically asked if they see smartphones going to hybrid bonding anytime soon, and they basically said no.

So I don't know, is this -- are these use cases being developed by somebody else, maybe not your main customer in Taiwan. And also, I just wondered if you might be able to comment on the competition from China. There seems to be a lot of companies coming up in China addressing hybrid bonding. And apologies for the third question.

No, they are good questions. Number one, there is no final answer in whether it will be used or not at this moment. So they're all testing, still understanding the benefits of one or the other. So the hybrid bonding process over reflow. They're all aware of the -- yes, the performance increase. Cost is a major issue right now because it has to take place in clean room and there's a factor of 4 to 6 more cost.

So to ask the question, will it be included in the iPhone 2025? The answer is no. We have always said going back Capital Markets Day, I think, 3 years, maybe even 4 that we expect 2027 to be the first in using that. So you have to be also aware that this testing is ongoing. And that doesn't mean that this is -- well, maybe not used this year, but it can very well be next year or the year after. Please help me with your second question.

Hello. Sorry. Can you repeat your second question?

Yes, apologies, I was on mute. I was just saying recently, there's been a lot of news flow around hybrid bonding competition coming out of China. I'm probably going to mispronounce the names, but the names of [indiscernible], Shanghai, YiNGUAN and a couple of others. So I just wondered if you could give us your thoughts on how you see hybrid bonding competition arising out of China.

Well, there's, of course, a lot of activity, not only in China, but also outside of China. Everyone now understands that hybrid bonding at some point for the applications we discussed earlier is a fact of life. So anyone building a precise bonder is developing also a hybrid version. So that's not new. How advanced they are is still a big question. But you can expect more competition coming from every region.

We are currently with the 100-nanometer well advanced in comparison to everyone else. And certainly, when we will ship the first 50-nanometer, which should be by the end of Q3, early Q4, that is not available at anyone else in the world. Also 100-nanometer is not a standard used by anyone else. So we have -- we continuously build on our lead. We follow closely the road map of TSMC, and we also fulfill that road map.

In China, the status today is with many of the devices like also the CoWoS is a reflow process. So a TC process, mass reflow process not yet in production, an HB process, a hybrid bonding process. We are very successful in the Chinese market today, delivering a lot of mass reflow flip-chip systems. We have 2 types. We have the 2100, and we also have the quantum, which are used for similar 2.5D modules in China.

So the process of reference having a market share of close to 100%. But anyway, your question is absolutely right. But with bonders forever, we have seen Chinese competitors trying to capture a certain segment in this market.

We will take our next question from Ruben Devos Kepler Cheuvreux.

I just had one regarding the co-packaged optics. I think in your prepared comments, you talked about 2 leading semiconductor producers that have made announcements. I think you've alluded to NVIDIA on this call, and I think Intel has also been quite explicit on its road map with hybrid bonding applications into ASICs and co-packaged optics. Yes, could you maybe talk about your engagement with those 2, but also maybe with some others that might also be candidates just thinking Marvel or AMD and these types of names?

And then just a follow-on on that. I think last year at the Investor Day, you talked about sort of the market scenarios for hybrid bonding, but you always excluded co-packaged optics. Very curious to hear your thoughts on how you think that side of that market could be maybe relative to, for instance, the HBM market or the high-end logic market with server processors.

Well, we did not -- I cannot remember that we left co-packaged optics out of the potential market applications. It's more -- well, there are 2 areas in this application field. There is these connectors where we are forever with photonics applications, which is done with 2 micron, 1 micron precision. We have major market position in those applications with a broad range of customers, also the top ones. Those are those connectors, which are included in the data center processors directly and then we have co-packaged optics, where we are also engaged with the customer you just mentioned and also the other one since early days.

So that is a very -- yes, let's say, interesting market, which is also in early days. We have the right technology for that, also the customers for many, many years. So at the next Capital Markets Day, we will make very clear to you what the potential of that market may well be and what are the different processes, but it's a key focus of our product strategy.

All right. Great. Looking forward to that. Then a second question just on Applied Materials. I believe you've been working with them since 2020. Could you walk us through how that collaboration has evolved in the past 5 years? And now that they've taken a stake, do you see that more as a strategic hedge? Or could it be the first step towards maybe deeper integration? Yes, how are you thinking about your independence long term as a packaging and maybe process control starts to converge?

That's a very big question. But let me start with the beginning. We started 9 years ago with TSMC to develop this hybrid bonding for, yes, CPUs for logic devices because the smaller bumped pitches required a different process. As soon as that became, let's say, into an end-market device in 2019, I remember vividly the first discussion we had with Applied and how can we work more closely to -- or let's say, work together in a closer way for the industry to support the adoption of this technology with offering, number one, process development applied as a Center of Excellence in Singapore and at the same time, support the development of automated lines because that improves the process window and the process quality.

So from then on 2020, we formalized that corporation just before COVID broke out. And ever since, we have worked very closely on a weekly basis, a review in bringing this technology to ever next and higher levels. It also has created in the market a clear model whereby the front-end company having the experience with front-end processes operating in front-end fabs and a back-end company, for which this next generation has to be developed.

So that has worked step by step, customer by customer in the logic arena, in the memory arena, also in other areas. And we are very positively surprised by Applied announcing an ownership stake of 9% in Besi that will only intensify our cooperation and leads to leadership position ever more into this market. So that's what it is.

We will take our next question from Martin Marandon-Carlhian, ODDO BHF.

My first question is on memory. What do you think are the key advantages in memories that allow you to win orders versus competitors for hybrid bonding? Because in logic, it's maybe a bit more obvious as the precision requirements are higher, and it seems that there is a technological gap between busy and competition in that aspect for hybrid bonding. But in memory, I'm curious to know about what makes the difference to win over competition.

Well, for memory, number one, the dies are thinner. So the complexity is different. So on logic, you have more contacts. Memory, you have less contacts, but the device is thinner. So the stacking is very much influenced by how you are able to control this variation in ultra thin devices. And one of the advantages of a reflow process, and that means a soft metal in between the 2 guys accommodates for some of the delta in thickness.

With hybrid bonding, it has to be completely flat. You have less room to accommodate the variation in thickness. But the process itself is better because you need less current, produce less heat. So the performance of a hybrid bonded device is better than that of a reflow bonded device. And then you have the cost. The cost is a factor of 4, 6 more expensive using hybrid.

So the whole memory world is developing on an ongoing basis. The advantages of both processes and understanding when our crossroads in using one or the other. In logic, you concluded already yourself. That decision is more close and has been made by many customers as we explained in detail, Intel finally also moved second half of last year. And TSMC, already 3 years earlier. And the advantages are ever more clear because you have closer contacts, the bumped pitches are smaller, and you have simply less issues with heat.

So that is where we are today. Besi, again, follows both road maps. So the reflow process from what I mentioned also earlier, mass reflow. It starts flip-chip, then TC individual devices, then hybrid bonding, and that is the complete offering.

Okay. And I have a follow-up on your U.S. customer. What do you think needs to happen with your U.S. customer before receiving a follow-up orders for hybrid bonding? Do you see they would wait to see what's the commercial success of the new CPU using hybrid bonding, you need the -- do you think they need to figure out what restructuring they need to do first? Or do you think they will necessarily have new capacity soon as their new products now have moved more and more away from TCB and transition more and more to hybrid bonding?

It's about, as you have just summarized. They are sampling the market with hybrid bonding devices, Clearwater Forest. We will understand that's what we are told towards the early part of the summer, how successful that is and what we can expect in terms of capacity increase. We are currently running every single day, and they're all test devices tested with customers.

And then there's a next device already, let's say, presented, which will be more for high-end computer applications, not so much for data centers. It's hard to forecast how successful that will be. But there's an enormous amount of pressure commitment to make this all work.

We will take our next question from Madeleine Jenkins, UBS.

I just have one. We've been hearing that the Koreans are potentially working on a sort of combined HBM solution. So the hybrid bonding and TCB are in the same kind of HBM stack. I just kind of wanted to get your view on like the technical side, what the viability of this is and also if it's something you've been involved with.

We have been involved in development stage. There are many different ways to skin a cat. So yes, this fits into the category, what's all possible and what makes sense. The excellent thing of our Center of Excellence in Singapore from Applied Materials is that, that is a center of all kinds of those developments for the industry.

We've added into our part also the TC Next product, which then allows customers to develop both and exactly addressing those 2 -- combination of those 2 processes. How far that will go, where you will also see that is at Imec. We shipped, as I mentioned earlier, in the first quarter, the latest TC Next. They're already producing already when the system was in [indiscernible], all kinds of devices, but we will share more data about that at the Capital Markets Day.

We will take our next question from Martin Jungfleisch, BNB Paribas.

I just have one quick one on the Q2 guidance. If you could just on us to the moving parts here quarter-on-quarter. So would you expect hybrid bonding shipments to be relatively flat quarter-on-quarter given the remaining shipments to the Taiwanese and also the U.S. IDM? So basically, does the midpoint of the guidance imply flat mainstream and flat hybrid bonding quarter-on-quarter? Or is it not the right way to think about it? And maybe if you can also tell us the backlog of the hybrid bonders that you should now have? Is it in the teens?

Well, we are not providing specific information about, yes, what we ship or what we scheduled to ship. But your assessment is also quite interesting. And it ties to an earlier comment. If there is a recovery partly, which are quick turnaround machines, then you can see a positive effect on which then may result in the higher end of the guidance.

How many hybrids are still in backlog, we don't disclose. But the key of your question, if I understand it correctly, the wide guidance range is typically or is because of this certainty will the second half of this year show a recovery in the mainstream business. If that happens, we have quick turnaround of epoxy bonders, soft solder bonders, flip-chip bonders not to forget. Quoting activity, yes, continues. So that's the reason why we guide this range plus/minus 10%.

Okay. Got it. But just to follow up, the order to the U.S. IDM, has that been fulfilled of the -- as of Q1? Can you comment on that?

No. There's still a few systems in Q2, but the -- I could say, 80% has been shipped. And exactly according to plan, what we have shared in the last updates.

We will take our final question from Timm Schulze-Melander, Redburn Atlantic.

I had 2 quick questions, please, Richard, if I could. First one, if we just leave hybrid bonding to one side and look at the core sort of much well-established business, could you just maybe share with us any color you have in terms of the utilization rates that are on that?

I think if we look at foundries, they're running at about 80%. I'm just curious if you could share some kind of flavor of what the specialist test assembly packaging industry is tracking at? And then the second, maybe more forward-looking question I had was on the launch of the Gen 2 or the introduction of your new hybrid bonding platform. And I just wondered if you'd be able to share with us what the sort of expected time line would be? When is the first tool coming? How long should we think about sort of eval qualification and just kind of how that plays out through 2025 and into 2026?

Excellent. Well, the first, why are we a bit cautious? Utilization rates, yes, are somewhere in the 80%. But if you look at the data provided by the industry, on inventory and pricing, and pricing has gone down again recently in the latest data from March. January, February looked up, March went down. That is, yes, let's say, an overall picture, which doesn't tell you there's an immediate shortage. But that's hard to tell whether customers are reacting more cautiously because of the overall situation, we don't know.

But in any case, for certain areas, there are shortages. One of them is the AI devices. As I mentioned earlier, there are Chinese versions, which are definitely in high demand. And -- but then that's not the general market. Automotive, we've also not seen yet, we've seen technology buys, new devices, power modules, but the big volume is still to come.

Then we intend to ship the 50-nanometer generation 2 by the end of September, early October and what the plan is that throughout '26, that tool will be made fully production ready in Taiwan and used for mainstream production '27 for those devices under 2-nanometer design geometry.

So that is well on track. And we will continue to update you on that progress.

Thanks, everyone. Excellent questions. If you have more questions, don't hesitate to contact us directly.

Thank you for joining today's call. You may now disconnect.