Good morning, good afternoon, ladies and gentlemen, and welcome to Besi's quarterly conference call and audio webcast to discuss the company's 2023 Q2 results. You can log into the audio webcast via Besi's website, www.besi.com. Joining us today are Mr. Richard Blickman, Chief Executive Officer, and Mr. Leon Verweijen, Senior Vice President, Finance. At this time, all participants are in listen-only mode. Later, we will conduct a question and answer session, and instructions will follow at that time. 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. Please go ahead, sir.
Thank you. Thank you all for joining us today. We will begin by making a few comments in connection with the press release we issued earlier today and then take your questions. I would like to remind you that some of the comments made during this call and some of the answers in response to your questions by management, may contain forward-looking statements. Such statements may involve uncertainties and risks as described in the earnings release and other reports filed with the AFM. For today's call, we'd like to review the key highlights of our Q2 and six months ended June thirty, and also update you on the market, our strategy, and the outlook. First, some overall thoughts on the Q2 . Besi reported solid Q2 results, with revenue and operating profit above the midpoint of prior guidance in a challenging industry environment.
For the quarter, revenue of EUR 162.5 million and a net income of EUR 52.6 million increased by 21.8% and 52.5%, respectively, versus the Q1 of this year. Sequential revenue growth benefited from increased smartphone demand this year versus 2022, partially offset by weakness broadly in computing end user markets. Net margins also grew to 32.4% versus the 25.9% in the Q1 of this year, reflecting revenue growth, gross margin improvement to 65.6%, and strict cost control efforts of production and operating overheads.
Revenue and profit development in the H1 year, 2023, also reflected the impact of current adverse market conditions on Besi's business this year, with revenue and orders each declining by 28.9% versus the H1 year last year, and net income decreasing by 39.2%. Current year revenue and order trends have been adversely affected by a broad-based downturn in computing applications, in particular, versus the H1 of 2022, partially offset by a slight uptick in demand for high-end smartphones versus last year's levels. Automotive order trends remained favorable in this H1 year, although slightly below the strong contribution reported in the H1 of last year.
Of note, revenue from China increased by EUR 10.5 million, or 11.4%, versus the H1 of 2022, reflecting modest improvement in demand for automotive, power, and smartphone applications, although no meaningful uptrend has been established yet. We are pleased with our profit performance in the H1 of this year, despite industry challenges with peer-leading growth and net margins of approximately 65% and 30%, respectively. As seen in this chart, Besi's performance this cycle is also significantly ahead of the last downturn versus the comparable period of the prior cycle. In addition, we completed a four-month strategic review of Besi's business in the Q2 of this year with a leading consulting firm to help advance our ambitions for expanding revenue and profit potential in this next upcycle.
Besi is committed to enhancing shareholder value via long-term financial performance, sustainability efforts, and capital allocation. To date, 2023, we have distributed EUR 367 million to shareholders, of which EUR 289 million was distributed in the Q2 of this year in the form of dividends and share repurchases. This brings total capital allocation to EUR 1.7 billion over the past 13 years, equal to approximately 30% of total revenue. In addition, we're on track to complete our 300 million share repurchase program by October this year. We ended the quarter with a strong liquidity position, including cash and deposits of EUR 378.3 million, post the large Q2 this year capital allocation.
The decrease in our cash position at the end of the Q2 is consistent with historical trends, following the payout of the annual dividend and increased share repurchase activity this year. Next, I'd like to speak a little bit about the current market environment and our strategy. It appears that the assembly equipment market formed a bottom for this down cycle in this past Q2 , post a steep decline beginning last summer. In addition, customer utilization rates have increased recently, although it is too early to say whether such increase represents a seasonal or a structural trend, and when a meaningful upturn may begin. If traditional assembly equipment cycles hold, one would expect an upturn either at the end of the third or in the Q4 of this year.
Most analysts anticipate an industry upturn in the H2 of 2023, with significant growth returning in 2024 and 2025, as per TechInsights, based on the strong long-term secular drivers in Besi's customer and user markets. At present, we are primarily focused on profitability, navigating the current downturn, and initiatives to capitalize on market opportunities in the next sector. We've reduced overhead and headcount in alignment with current market conditions, and continue to increase our R&D investment for the next generation systems. We've updated our strategic planning to help achieve business model growth objectives over the next 5 years. Progress also continues on our hybrid bonding and wafer-level assembly roadmap. Activity associated with hybrid bonding adoption has increased significantly over the past 6 months, with the primary focus on customer qualification and testing of processes for next generation architectures and new market applications.
We believe that the prospects for wafer-level assembly growth have increased successfully each quarter. This belief is based on the high level of interest expressed by significant resources committed to, and sampling work done by, leading front-end customers, also OSAT, and the development community, particularly in the areas of data center, AI, mobile, and high bandwidth memory applications. The favorable outlook also reflects Besi's first mover advantage, successful move to volume production, improved yields, and ongoing progress in developing integrated hybrid bonding production lines together with Applied Materials. We are also encouraged by the shipment of Besi's next generation TCB system for qualification in high volume production. Charter, the Singapore clean room facility, was completed recently to support process development for hybrid bonding adoption. From an operational perspective, Besi is targeting key expansion projects by year-end, including a Vietnam tooling support facility.
In addition, we formed a new company in India as the result of increasing mobile assembly demand in that country. There were also some organizational changes at the end of Q2 with the retirement of Geert Boonsma, CTO. Geert will continue his involvement with Besi as chairman of a new Besi technology board, to be organized in the H2 of this year. His responsibilities have been assumed by the CTO office and members of senior management. I personally want to thank Geert, on behalf of everyone, for his important contributions to the growth over many years. Now, a few words about the Q3 guidance. Near-term market outlook remains uncertain, despite initial signs of improvement, and varies per end user market.
Accordingly, we anticipate that revenue will decline by between 20% and 30% versus the Q2 , due to typical seasonal patterns and current industry conditions. We expect gross margins to range between 62% and 64%, and for operating expenses to decline by 10%-15% versus the Q2 . We also expect that the Q4 revenue will significantly exceed the Q3 revenue levels, based on scheduled shipments from backlog, particularly for wafer-level systems. That ends my prepared remarks. I would like to open the call now for questions. Operator?
Thank you, sir. Ladies and gentlemen, if you would like to ask a question or make a contribution on today's call, please press star one on your telephone keypad. We'll take our first question from Ruben Devos from Kepler Cheuvreux. Please go ahead. Your line is open.
Yes, good afternoon. Thank you. I have two questions. The first one relates to the production capacity for the hybrid bonders. I think in the past, you talked about 50 per month or about 180 on an annualized basis. I was just curious, at what point could we sort of reasonably assume that this capacity would be filled? You've been quite specific historically in terms of how many hybrid bonding orders you realized. I think at one point it was about 7-8 per quarter, sort of the run rate. We missed a bit of that detail in this release. Also, could you maybe talk about the phasing of orders heading into 2024 and beyond? That's the first question.
Excellent. Well, as we have explained, the theoretical production capacity of 180 systems per annum has been put in place in the past 2 years, simply to convince customers that when adoption becomes mainstream, they can, let's say, expect delivery of systems to those levels from us, which should satisfy their demand in the next 3 years. We've also said that 2023 and 2024 are typical qualification years. As we mentioned in this call, the qualification is continuing very rapidly. Those of you who have witnessed the SEMICON West event last week, were in a panel discussion with our key customers, AMD, Intel, Qualcomm. Confirmation was provided that this ongoing qualification will lead to adoption in the mainstream in the time frame, as we have explained also in the Capital Markets Day.
2023, 2024 qualification, major volume, CPUs, 2025, 2026, memory kicking in. What was interesting, we mentioned that following memory, high-end smartphones would kick in. Message last week that that could also be earlier. That tells you there's an enormous drive in the industry to design next generation devices for interconnect using hybrid bonding processes. That roadmap stands. Orders typically come in, in batches. There's no average in... Well, in hindsight, you can calculate an average, but it depends on programs. The question would be, where are we? We have received earlier this year, orders for our next expansion in Taiwan. Those orders will be delivered this year. That's also why the comment for the Q4 was made.
We expect certain orders for setting up initial production capacity here in the U.S., on top of already systems installed. Those should come in, in the next quarter, next two quarters. Production sites are being prepared. Don't forget that we're talking in hybrid bonding about front-end environment that takes longer than the typical assembly. The key message I'd like to share with everyone is that it's perfectly on track.
All right. Very helpful. Thank you. M y second question, I think just around high bandwidth memory. I, I think we've seen quite a big update by the Koreans. They've also been quite aggressive to grow strongly here. You've talked a bit about the memory market roadmap in the CMD, but is there a fair possibility that these roadmaps could be accelerated considering the very strong demand out there? Yeah, and how should we think about the relative use of both TCB and hybrid bonding within memory or within high bandwidth memory?
Also very good question. As said to your earlier question, the roadmaps memory are also as they are, they have not been pulled forward or delayed. There's an enormous qualification development activity with the major memory manufacturers, using hybrid bonding technology. Simply, that roadmap stands. The other question on TCB, that's another application area. The question, of course, is, will that be a TCB or hybrid bonding? The advantage of hybrid bonding is also confirmed in many of the public documents, conferences, over TCB. They will be dual, let's say, technologies for the coming years. Ultimately, hybrid bonding should be the major ones in the future horizon, because TCB is limited for further shrink of interconnect design geometry. That difficulty with TCB remains. That's as much as we can say about this.
All right. Thank you very much.
Thank you. We'll move on to our next participant, Charles Shi from Needham & Company. Please go ahead. Your line is open.
Hey, good afternoon, Richard. Thank you for allowing me to ask a couple questions. If I look at your guidance for Q3, you provided some preliminary color on Q4. Q3, the numbers will be slightly below seasonal, but I'm actually a little bit surprised to hear you're seeing significant higher revenue in Q4, which is definitely above seasonal, if I understand correctly, because your typical seasonality, Q4, it should be down Q on Q. Can you provide a little bit of a breakdown to why the pattern looks like this? Was there some delays in terms of backlog shipment from Q3 to Q4 that caused this pattern?
How confident you are in terms of shipping out that backlog in Q4, 'cause, well, just because it's in the backlog doesn't necessarily mean you will ship. Really just want to have a little bit more understanding about your confidence level at this point. Thank you.
Excellent, Charles. Well, it is like you, like you summarized, typically, H2 is less than the H1 , and Q3, with some exceptions, is also typically lower than Q2. You can see that statistics also in our presentation. This time, it is a mix between the Engine 1 business, which is still impacted by the industry downturn, huh? Orders are on the lower levels, huh, with some positive, maybe because we see utilization rate increase. I say that carefully because it doesn't tell you whether that's structural yet, huh, that needs a bit more time. Why this increase? To say that again, is because of shipments of systems in the backlog for the submicron technology.
They are tied to customers being able to accept and to install these systems in the clean room. Also the ASP, times the number of machines, helps to simply understand the higher revenue level. As always, there are risks, but most of it is, of course, in the backlog, huh? Certainly the longer submicron systems already, that's all scheduled, all being prepared. That's the picture supported by a backlog in place.
Yeah. Thank you. Is it fair to say that incremental revenue from Q3 to Q4, the primary driver is the wafer-level assembly systems here?
Yes.
That's the first part of the second. Okay.
Yes.
Are you, expecting, some additional orders coming into Q3 to really support that, the incremental, revenue from wafer-level assembly? Sounds like, the U.S. customer may be the top driver for that, or is that correct?
It is the other side of the Pacific, the orders to be expected. We still have a lead time of nine months for these systems, especially with some additional features. Those will be delivered in the H1 of next year. These are orders placed in the H1 of this year.
Thank you, Richard. Okay, I just want to clarify the U.S. customer, what's our current outlook right now in terms of wafer-level assembly? When do you expect that the orders for the initial volume production capacity, and when do you expect those will be shipped? Thank you. That's my last question.
Well, if all goes well, we should see that in the coming months.
Thank you!
Thank you. We'll move on with our next participant, Madeleine Jenkins from UBS. Please go ahead. Your line is open.
Hi, thanks for taking my question. My first is just on your new TCB tool. I was just wondering if you have seen any interest from other customers, and also whether it would be a candidate for kind of next generation CoWoS?
Your first question is on the TCB tool. The answer is yes. There's interest from two other major customers. If all goes well, the machine, as you all know, has been shipped, is being qualified as we speak, in a traject 3 of about six months. We could expect October, November timeframe, whether this is already to be used in the next generation or would that take a bit longer? That's always with certain uncertainty, but the pressure on the system performance is very high. That's good news. The other two customers will use it for slightly different applications, but with the same process. One is related to memory, the other for another CPU type of device. That looks very good. Your second question on CoWoS.
Well, there's a lot happening, and there's a lot of, let's say, information shared publicly about CoWoS expansion. We are certainly involved in that, and not only with one customer, but also second customer in Korea. Still, we can't say that often enough, the volumes in that CoWoS are relatively limited in a sense that this is not huge volume. Anyway, it's good business.
Okay, great. Thank you. Then I just had a very quick follow-up on hybrid bonding. I was just wondering, is your major competitor's tool the same as yours in terms of is it targeting similar applications, or are there any differences between the two?
Well, there's. That's, again, an excellent question. At this time, why use hybrid bonding? I said that a little bit in answer to the earlier question. The reason to do that is to have a more, or let's say, a higher yield interconnect solution compared to a TCB solution. That drives this solution. At the same time, when this is really feasible for mainstream, it allows you also to create these chiplet architectures, and some have already reached the market, take AMD. There's a lot of, let's say, enthusiasm about this next step, because that also has the potential to extend Moore's Law, which is very important for the overall cost of semiconductor manufacturing, apart from the performance of the devices.
Since this hybrid bonding, July, September 21, officially, the first devices were on the market, the industry started to really focus broadly on this. Before that time, it was of course known, but not that expected to be coming at that time frame. Also competitors, after that first product successful in mainstream applications, started to focus on developing hybrid bonding applications. Like in any market which has a growth potential and which is at the forefront of technology, it invites everyone and customers also like to have different choices. So far, we have been very fortunate that we were early adopters in a sense. 7, 8 years ago, we started with this whole development. We have mid-20 machines in full production per year and more others qualified.
In this chiplet architecture, you can imagine not every device is the most complicated device. You may end up with a selection of devices, some needing accuracies below 150 nanometers. We now delivered the first 100 nanometer machine, 50 should be available in 2, 3 years from now, and that's really for the most complicated devices. There are other devices which need less accuracy and placement, and you may see markets developments in that direction. This is still early days. Our systems now have this 150 to 100 nanometers and 100 nanometer cover the entire spectrum and are running every day with higher yields, faster speeds. That's the way to maintain a certain head start in this technology.
Does that answer your question?
Yeah. Thank you very much.
Thank you.
Thank you. We'll move on to Nigel van Putten from Morgan Stanley. Please go ahead. Your line is open.
Hi. Yeah, thanks. I have a follow-up on the hybrid bonding roadmap for smartphones. Thanks for posting the video of the AMAT symposium on the website. To summarize Qualcomm's comments correctly, I think they basically said, "We're thinking about it a lot. We want to do it, but cost limits adoption right now." My question is 2-part. I guess first one is: How large is the cost effort differential versus, I guess, Flip-chip? Where do you think companies like Qualcomm would be happy with? Then second, have you formulated more of an explicit roadmap, maybe together with AMAT, to bring costs down? If so, could you provide some color on what the upcoming signposts are and the time frame? That's my first question. Thanks.
Well, the best signpost is throughput. As you may remember, we indicated our machines today run with the devices which are being produced at about 1,500 UPH. The machines as such, are able to run twice that speed, 3,000 UPH. Any close competitor is at a third of that. It's 500 UPH. Of course, there are models, cost of ownership models. We should expect in the next year that we bring this throughput speed up. It all has to do with software, with also preparation. You have to recognize certain device structures and translate that into accuracy placement. That, in the end, will determine whether it's cost competitive using whatever flip chip today, partly TCB, but flip chip for simply the majority still.
Those machines run at 6,000, 7,000 UPH. It's all about the design of the device. It's the combination of design, geometry, cost, and production cost. Your question, are we focused on that? Of course, not because we, these customers really are demanding those solutions. It will take some time, every day, significant progress is made. That's why we also expand the capability of process development. Our Singapore lab is currently ready, machines are being installed, as of August, early September, we will be able to do side by side with AMAT on the other side of the road, tests on individual bombers and on integrated lines. That all has to do with adoption in those categories, especially in the high-end smartphones.
Got it. Thanks. If I summarize, I mean, you with AMAT, well, your part of the equation is still increasing throughput. I can imagine there's plenty of other developments needed as well, but, there doesn't seem to be any structural hold hurdle. It's more improving the current process. I won't try to pin you on-
Yeah.
What now the new timeframe is for smartphone adoption, but that does seem to be, or you, you flagged maybe toward 2027 at the Capital Markets Day. Let's just say that that is more likely now, it seems. Is that sort of fair to say?
I would say it differently. I was a bit, or let's say, positively encouraged by the fact that Qualcomm, it's not only Qualcomm, that they are more actively 2027 is too far out. It all depends whether we can realize that. There's a very positive pressure on us to be able to achieve that.
Yep, understood. Maybe switching gears a little bit, different question. In your prepared remarks, you said you formulated a new business objective for the next five years. I'd be very interested to hear what the revenue target is for that model. If you're able to share any color or a number, that would be helpful.
Well, first of all, we have to always reach the targets we have set so far. Our model, EUR 1 billion plus, plus, plus, which equates to the EUR 1.3 billion model, is the first hurdle, and with a very, again, concentrated effort on what does it take to get there. With our Engine 1 business, the Engine 2, the submicron world, customers engaged our whole organization in a period of 16 weeks. We need, of course, an upcycle for that. Anticipating that an upcycle comes. Yeah, and in the past, we have always been able to reach those goals and even higher than that. Remember the EUR 800 million model, the EUR 600 million model, the EUR 400 million model. That's how it works.
No, that's exactly the reason for asking. I think the 2025 model is now the $1 billion plus, plus, plus. This seems to be more of a sort of cycle beyond that model. Anyway, I was just curious to hear if I could tease out a number, but I'll leave it there. Thank you.
Next question.
All right. Thank you. We'll move on to Marc Hesselink from ING. Please go ahead. Your line is open.
Yes, thank you. Maybe, first question, on the strategic review that you're doing. Can you maybe share what's your intention and what kind of things are you looking at? Is that increasing the revenue opportunity? Is it further increasing the efficiency, which is already at a very high level? Just your thoughts about doing this review?
Well, like we shared each time, there's an enormous development at this moment, ongoing. If you simply imagine this wafer-level universe, hybrid bonding, also chip-to-wafer in clean room environment, that requires from a company to shift from its assembly market environment forever in, into a front-end universe. Different sides of customers, different process requirements, and you have to organize that very carefully, because the risk you run, if you don't do that right, you may lose both. The fact that we organize these reviews with the entire management, is to simply do as much as we can with our customers and partners to be able to get there. Don't forget, we're in the initial phase, 150, 200 nanometer accuracy, now gradually going to 100 nanometers, then down to 50.
That is something. Compare that to, to lithography, for instance, in the past decade. These are major inflection points. The reason to do these exercises is to make sure, as much as we can, that we realize those objectives. Organizations change. There are new people on board. The organization structure has to be adjusted. Business developments, don't forget the support in the U.S., in Taiwan. A couple of years ago, we only heard of the name TSMC, and then Korea, for that matter, huh. A lot has to be organized, apart from the change in the geopolitical or changes due to the geopolitical landscape. You mentioned Vietnam, balance China, and then India, not to forget. There are many aspects which provide enormous growth opportunities.
To maintain that growth, and then with margins, which we have demonstrated, simply requires that you spend more on your strategy than the usual process in an annual sequence. That's a longer answer to your question.
Okay, thanks. Clear. Second question is actually on the gross margin, again, exceptionally high, and I know you just updated the guidance at the Capital Markets Day. Given that you're so many times ahead of the year of that guidance, is it appears to be upside risk to that numbers on a quarterly basis?
Gross margins, there are many factors which influence, but number one is product position. On an ongoing basis, our products are in a position where pricing allows these margins. There are other factors like ongoing cost reductions. One of the, yeah, major things you have to extend the basis, we adjust very rapidly to lower demand or higher demand. That is also unique, and then you have currencies, and we're supported by hard dollar. The Asian currencies are under pressure. All those factors, they. We manage this business, and some of you know, on a practically daily business, every Monday, Wednesday, Friday. That all adds to controlling your cost, but also at the same time, improving your market position. That combination is what determines your margin.
Yeah, we can easily see that over the years, our product position has improved cycle by cycle. It's the result of a very, very clear focus. Not making too many mistakes, although we make many mistakes.
The reason that you're still ahead of your own guidance, is that then, what's driving that then, in like the very short period that you sometimes have even higher than the normal range?
Well, that can be in a moment, I mean, don't forget, we have 18 or now with TCB, 19 different platforms, and different supply chains, cost structures. As I said, there are sometimes pluses and there are sometimes disappointments. It's not that if we would have one product, it would be different. That's the best answer.
Okay, thanks. The final question that I have is, I was just wondering, like, on which level do you have the strategic discussions? You clearly have it with your direct clients, but do you also have that on hybrid bonding and the other submicron, do you have that strategic discussions with the fabless companies?
Of course. Of course. It's all the customers in those advanced development of devices for the winner end products. In the three categories, communication, so high-end smartphones, but also the other devices, wearables, et cetera. You have the whole computer space, and you have automotive. In all three, there is this forefront technology development ongoing. That you have to constantly analyze, are we focused on the right programs? Stop with those programs which are not mainstream feasible. There's no roadmap cast in stone, you have to align those roadmaps constantly. What's key at this time, when there's a major change in this whole interconnect technology, that you align with the right customers and the right partners, and one of them is Applied Materials.
The other is, of course, to understand what's happening in the lithography space. It's not just one customer.
Okay. Very clear. Thanks.
Thank you.
Thanks.
We'll move on to Michael Roeg from Degroof Petercam. Please go ahead. Your line is open.
Yes, good afternoon. I have a couple of follow-up questions on the Q4 sales guidance. Is the expected uptick in Q4, is that only because of hybrid bonding, or is TCB also making a contribution in that quarter?
It's only a hybrid bonding.
Only hybrid bonding. Will that big order for your Taiwanese customer, is it entirely delivered in the quarter, or is there a tail slipping into Q1 as well?
The tail is slipping in Q1.
Okay, that's good.
Anyway, that is answer to earlier question, we're end of July. Those are the schedules as they are right now. As always, these, there are other factors which impact those shipments as well. Anyway, that's the best knowledge we have.
Okay. Then maybe a trick question for Q4. The significant increase quarter-on-quarter, is significant meant to represent the 12%-20% according to the midpoint scale?
Whatever it is, it's more than single digit.
Okay. Good to know. I have another question on AI. Lam Research said something interesting about AI. If 1% of all the data centers would convert from enterprise servers to AI servers, then it would represent $1 billion-$1.5 billion wafer fab equipment spending. Have you seen something similar, or have you made calculations what the AI potential could mean for your business, for your end market?
Yeah, certainly. These models are underlying the model for adoption of hybrid bonding. That's why we have this lowercase, mid case, higher case, apart from whether it's CPU or CPU and memory or, let's say, high-end communication device applications. There are different scenarios, but there are many models made today, and I, I would simply be a bit careful. Those models are always the best, the best ever.
Okay, they indicated indeed, AI servers have a lot more leading-edge chips and much more DRAM and NAND storage. AI for you is especially in the hybrid bonding segment, or is TCB also involved in there?
Yeah.
for you?
Well, a combination.
Combination.
Big bonanza is bonding.
Okay. Engine 2, opportunity for you. Good. That's it from my side. Thank you.
Thank you for your question.
Thank you. We'll move on to Timm Schulze-Melander, from Redburn. Please go ahead. Your line is open.
Hi, good afternoon. Thanks for taking my questions. Just have a couple. Maybe, could you just maybe take a step back just in your hybrid bonding product development? You talk about, you're moving to a sort of a gen two platform. Could you just give us a quick update of how that's going, Baked into that, you are managing your operating expenses. Clearly good news for profitability. Just curious, you know, is that falling equally between R&D and SG&A? Any color you could give on that, and then I have a follow-up. Thank you.
Well, if I understand your questions correctly, first on the cost, huh? On, yeah, it's clear that the R&D cost for, yeah, next generation hybrid bonding tools is higher than in the Engine 1 space. Especially at this time, the investment is significantly higher in a market which is in early stages of development. As long as major customers are driving this adoption, that justifies, and also that they're buying our machines, that we're on the right track. We don't look at it from whether that is equally shared. Every product, in our view, is always a company. It's an individual entity, and it has to simply perform financial metrics which justify long-term success and return on capital for everyone investing.
I can't answer that question in a different way. Is this what you, you asked, or
Yeah.
Did I not understand your question?
No, no, no. You did. I guess what I'm specifically thinking about is, your spending hybrid bonding portfolio. You know, is it protected, and is it an area that's still gonna grow within the backdrop of cost control, or is it also subject to cost control and less investment in Q3, Q4? as part of that cost control effort?
No.
Okay, great.
No, certainly not. Let's say that question is very important. If you look at the financial metrics we just explained, that you can develop this hybrid bonding technology based on the financial metrics of your Engine 1 business, also in a downturn where business is down by more than 25%, tells you that all of our business engagements are in a very high margin environment, and cost is under control in operations, because we simply build less machines. You need less people. You need less people to install machines when you sell less machines.
Right.
That's where we cut costs, but on R&D, we only increase.
Okay. That's reassuring. Maybe, you know, how is that next tool platform development going and when will you be able to showcase something to your leading customers?
The second or the 1-plus, huh, so from 150 to 200, which is generation 1, we shipped the first generation 1-plus, that's the 100 nanometers, which has demonstrated to be able to do that, I wouldn't say easily, but certainly also below the 100 nanometer. That should cover from next year onwards, once it's qualified. The most advanced devices with hybrid bonding, we're engaged heavily on developing the 50 nanometer, which should be ready in 3 years down the road.
Got it.
programs are fully supported by our financial means, our balance sheet, our backbone. There's no slowdown in that development.
Okay. Then the last question I had was just on shipment cadence. If do we understand it correctly that in 2Q, 3Q, likely you had low or no hybrid bonding shipments, and they are now clustered in Q4? If you could maybe just provide some color on that, that would be really helpful. Thank you.
We did ship some hybrid bonder tools in Q2. As I explained earlier, it is lumpy, sometimes they call that. It's tied to orders from customers. Your assessment, yes, it did. This was always planned for shipment towards the end of this year, H2 , we mentioned. It is in Q4. It could have been in Q3, but the place where the machines are installed was not yet ready. Anyway, that is also the background.
That's really helpful. Thanks so much, Richard.
Thank you.
Thanks, Tim.
All right, we'll move on to our next participant.
Any further?
Yes. Didier Scemama from BofA, please go ahead. Your line is open.
Good afternoon, Richard. Thanks for squeezing me in.
Thank you.
Dumb questions, I'm afraid. Yes. First of all, can you tell us in Q2, how many hybrid bonding systems you booked and how many did you ship? Can you also, reiterate, or change, whatever your number of hybrid bonding system shipments for this? I think you used to say flat this year. Is that still the case given your guidance for Q4? Thank you. Got a follow-up as well.
Yeah, it, as we have said, in many instances, 2023 and 2024 are initial qualification and a little bit of expansion of these first generation devices, huh, which we have talked about many times, the AMD products, the initial qualification for the others. High volume production orders we have not received. It's of course exciting to expect, and hopefully we will see that by the H2 of this year, for then preparation, more high volume production to be installed in 2024, to be ready for 2025, huh. That's the situation. No programs have been delayed.
Programs are on track. Important, again, to repeat that, is that customers, AMD, Intel, Qualcomm, openly, let's say, support that this is the technology they are developing for the next generation. That should result at some time in more orders.
Got it. That's helpful. The second question is going back to hybrid bonding and the longer term picture. I think on the last earnings call, you, you mentioned that high bandwidth memory could be a bigger opportunity in AI for you guys than perhaps GPUs. I mean, is that still the view you've got, given the number of layers in high bandwidth memory? Or have you started conversations with GPU players for AI servers for adoption of hybrid bonding as well?
Let's say, the development is across the board, so for all these devices, there's a lot of development in the Center of Excellence in Singapore. Shortly also in our own facilities, simply because we can't accommodate everything in the Center of Excellence. Center of Excellence is more focused on integrated lines, simply with the advantage, reducing particle risk further, and then on individual bonding processes in our own lab, but that's combined. What will be sooner or later translated into mainstream applications is not fully crystallized. We mentioned earlier to an earlier question, we thought three months ago that smartphone adoption would be later. If you hear the current comments in publicly made, it could also be earlier. That tells us as long as our machines can do those applications, once it becomes mainstream, we're ready for that.
Makes sense. Maybe final question on bookings. Obviously, Q2 bookings were pretty low. Not gonna ask you about Q3 bookings unless you want to comment, but the normal seasonality typically is for bookings to be down, I think Q3, Q4 versus Q2. I guess you're looking at things slightly differently, but I just wondered if you could give us any color on the trajectory of bookings for Q3, Q4.
Some statistics. If you look at the last down cycle, 2019, 2018, it started in the Q2 , and then in a similar pattern, the trough was in the middle of 2019, and orders started to pick up in Q4. We had the abnormality of COVID hitting the world in February, March 2020. Yeah, it should become more clear in the next 2, 3 months, whether this increase in utilization rates currently, whether that's only seasonal or whether that is leading into the next up cycle. There are definitely different views, and TechInsights is very, yeah, let's say convinced 2024, 2025 will be a major up cycle again. There are some others which are a bit more careful, that it more, yeah, should start in the H2 of 2024.
Nobody is right in following cycles. We are prepared whatever way it goes. One thing is for sure, current levels are significantly higher than the levels we had in the past down cycle in 2019, which is very encouraging. Anyway, that's as best as I can answer your question.
No, that's brilliant. Thank you so much, Richard. Hope to seeing you soon in London.
Look forward to.
All right. We are at the end of this schedule end time, and I'd like to turn the conference back to Mr. Blickman for any additional closing remarks. Thank you.
Thank you all very much for your interest. If you have any further questions, you know where to find us. Please do not hesitate. We'll be happy to answer. All the best. Have a good summer. Bye-bye.
Thank you for joining today's call. You may now disconnect. Have a nice day, everyone.