Good morning, good afternoon, ladies and gentlemen, and welcome to Besi's quarterly conference call and audio webcast to discuss the company's 2023 third quarter 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 Wubben, 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. Thank you.
Thank you. Thank you all for joining us today. We will begin, excuse me, by making a few comments in connection with the press release issued earlier today, and then take your questions. I would like to remind you that some of the comments made in 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 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 third quarter and nine months ended September 30, 2023, and also update you on the market, our strategy, and outlook. First, some overall thoughts on the third quarter. Besi reported solid third quarter results, with revenue and operating profit above the midpoint of prior guidance.
For the quarter, revenue of EUR 123.3 million and a net income of EUR 35 million decreased by 24.1% and 33.5%, respectively, versus the second quarter of this year. But orders grew by 13.1%, reflecting increased demand for next-generation AI, high-performance computing, Hybrid Bonding, and photonics applications. The sequential revenue decrease was principally due to decreased shipments for high-end smartphone applications, post the first half 2023 seasonal capacity builds and general market weakness. Profit levels remained elevated, driven primarily by gross margins of 64.6% and a 15.6% decrease in operating expenses, both of which were better than prior guidance.
For the nine months ended September 30 this year, revenue and net income declined by 28.4% and 39.1%, respectively, reflecting the impact of adverse market conditions on Besi's business this year. This year's revenue and order trends primarily reflect a broad-based downturn in demand for computing applications by both IDMs and Asian subcontractors, and to a lesser extent, reduced demand for automotive applications following strong growth over the past two years. Our profit performance in this market environment remains strong, with gross margins up 3.7 points and a net margin realized of 29.1% as a result of the timely adjustment of Besi's operating model to current market realities.
Besi continues to significantly outperform its results as compared to the last industry downturn, with latest twelve-month revenue and operating profit tracking 56.5% and 115.4% versus the comparable period of 2019, with operating margins up 9.5 points. Our performance this down cycle highlights our revenue and profit potential for the next upturn. To date, in 2023, capital allocation to shareholders has grown by 17.4% versus the comparable period last year. This brings Besi's capital allocation over the past 13 years to a total of EUR 1.8 billion. In addition, we will complete our EUR 300 million repurchase program on October 27, 2023, under which we will have purchased approximately 4.1 million shares at an average price of approximately EUR 70.
Effective November 1, we will initiate a new EUR 60 million program designed to further reduce share dilution from the conversion of our convertible notes outstanding. Besi ended the quarter with a strong liquidity position, including cash and deposits of EUR 391.2 million, up 3.4% versus the second quarter of this year. Of note, our cash position reflects the capital allocation of EUR 412.4 million to shareholders this year. Next, I'd like to speak a little bit about the current market environment and our strategy. We believe we are in the early phase of a new assembly market upturn based on independent research data and customer utilization rates, post the trough reached in the second quarter of this year....
As seen in this next chart, courtesy of TechInsights, it appears that the industry conditions improved further in the third quarter. We caution, however, that there are many variables which could affect the upward slope of any new recovery, including global growth rates, geopolitical conflict, and the development of each of our principal mobile computing and automotive end-user markets. Analysts expect the assembly equipment market to fall more greatly in 2023, and to rebound more strongly in 2024, than either the front-end or the test markets. TechInsights expects the assembly equipment market to rebound from trough levels in 2023, to reach next cycle peak levels in either 2025 or 2026, in the range of $6.5 billion-$7.1 billion.
At present, we are primarily focused on maintaining solid margins in the current downturn, and executing development and operating initiatives to help capitalize on market opportunities in this next upturn. We are also pleased to report significant progress on our advanced packaging roadmap this quarter, as we saw orders increase for generative AI, high-performance computing, photonics, and CoWoS applications. New hybrid bonding orders were received in the third quarter from two customers, including the first order from a leading subcontractor, as well as significant orders for photonics applications from industry-leading customers. Subsequent to end quarter, we received further hybrid bonding orders from a logic and a memory customer, and anticipate additional orders in the fourth quarter of this year. In addition, we received die bonding orders from multiple Asian subcontractors for chip on wafer on substrate applications, both in the third and to date in the fourth quarter.
Now, a few words about our fourth quarter and full-year 2023 guidance. Given order trends in the third quarter and customer feedback, we remain cautiously optimistic, encouraged by the ongoing progress of our wafer level assembly portfolio and market leadership position in key advanced packaging assembly systems. For the fourth quarter, we anticipate that revenue will increase by 15%-25% versus the third quarter, due to planned shipments of, from backlog of hybrid bonding and other advanced packaging systems. In addition, we expect gross margins to range between 62%-64%, and for operating expenses to increase by approximately 5% versus the third quarter of this year. That ends our prepared remarks. I would like to open the call for questions. Operator?
If you would like to ask a question, please signal by pressing star one on your telephone keypad. We will take the first question from line, Charles Shi from Needham & Company. The line is open now. Please go ahead.
Well, thanks for taking my question. So, Richard, I think the first question really is about the Q4 uptick in the revenue outlook. And if we back out some really unique unique factors like hybrid bonding, how would the revenue trend from Q3 - Q4? And, what's your read on where we are in terms of the cycle? I think you could provide some comments and prepared remarks, but, what exactly are the leading indicators you're looking at, how bullish you are about 2024? Thank you.
Well, Charles, thank you. First of all, if we back out the, advanced packaging new technology, still the overall market is, I would say, flat. As we, explained, utilization rates are up. We shared three months ago, they were in the range of about 60% on average, I would say. They've climbed further to 70%-80%. And, the key question will be, is TechInsights, right about, their view that the industry will recover strongly next year? Or will, as we said, the slope of recovery be a bit more, careful? It's always hard to tell.
End of October, if we draw this with a similarity in 2019, when we presented the third quarter at the same, I think, date or a day before or after, we saw a similar situation, and the market picked up before the end of Q4. So summarizing, I would say, it typically follows a similar pattern, everyone being cautious after 6-7 quarters of downward trends. We also saw numbers of competitors, peers in the sector in the recent week, indicating a similar picture. So again, Besi is very strong in the new applications. We're enjoying, as explained in the AI environment, photonics was also very strong.
Continued orders for hybrid bonders for more customers, further proliferation, adoption, however you want to call it, in a broader sense. So that moves the needle at this moment. Does that answer your question, Charles?
Yes, thanks a lot, Richard. So maybe a follow-up question on the orders. I think I've heard a couple new things here. One is, you seem to say that you have got orders, Hybrid Bonding orders for memory customers. Can you elaborate a little bit here? 'Cause, well, we kind of were expecting this order, right, for a while, and-
Yeah
Glad to hear it. It seems to like it finally arrived. But the other thing, I think it's more interesting from my perspective, is the orders for photonics application. Can you kind of tell us what do you mean by photonics? Is it the optical communication kind of photonics, or is it the mobile sensing kind of photonics, so which would be more of your traditional business? Any color on that would be great. Thanks.
There are two types of photonics applications which, today, are resulting in significant orders for us. One is related to the plugs, connecting the optical fiber to servers, basically. The other one is directly connecting it into an electronic circuit, and that needs to be far more precise, as we can imagine. It's, as I mentioned, a broader range of customers. We've been active, as you know, Charles, for many, many years in this photonics world, in still, let's say, lower volumes. Clearly right now, there is a decent increase in demand, in a broader sense. For both, the lesser critical, but also the more critical applications.
Sorry, I cut you off, but you had a comment on the memory or order for Hybrid Bonding.
Ah, yeah, the memory. As you, nicely summarized, we've been expecting this for, quite some time. We've been doing tests for multiple memory, customer applications using hybrid bonding as an interconnect technology. The first, after qualification and testing, the first real order, has been received. So that also follows the, roadmap we have indicated for the past nearly three years. So first, focus on logic, memory next, and also continued development in, chiplet applications for, high-end smartphone applications.
Thanks, Richard.
Thank you.
Next question.
If you would like... Thank you. If you would like to ask a question, please signal by pressing star one on your telephone keypad. We will take the next question from line Marc Hesselink from ING. The line is open now. Please go ahead.
Yes, thank you. The first question is actually following up on the last bit on the hybrid bonding for mobile applications. I think in the previous call, you said that it seemed that maybe it becomes a little bit earlier than first year initial expectations at the capital markets day. Can you update on that part?
Well, it has not changed from our comments three months ago. So continued development, tests, and you could say according to plan. And if we translate that in or project this in mainstream applications, you are still talking about earliest 2025. And as we indicate on the capital markets day, maybe 2027, but who knows, whether it's 2025 or 2026. But the key is that the hybrid bonding process becomes ever more established and on the levels of 150, 200 nanometers with by year and over 40 machines in the field. Every day gaining significant progress, and progress means reliability and speed. We also shipped successfully the first 100-nanometer tool, as we have informed the market.
Also a lot of further development on this next generation, we call it the generation one plus, is continuing according to plan. And that should lead to ever more applications. And for the high-end smartphones, is typical in high-end computing, is the chiplets capability. So to combine multiple devices, same like in CoWoS, where you have a 2.5D type of design structure, in using hybrid bonding to achieve a higher density, shorter distance, a direct interconnect, less power consumption. So all the benefits should add up and ultimately resulting in lower cost. So integrating electronic components, the big driver is not only performance, but is especially cost.
Okay, clear. Thanks. Second question is a follow-up on the photonics. To understand even a bit better, did something change on the technology side of your client that because of that, they now need more equipment for you? Or is this more of a coincidence that now it was a step up in demand for this business versus previous quarters?
No, it is more tied to the CoWoS environment, the CoWoS generation 2.5D. And one of the technology steps is this photonics, which has been on the roadmap for, as I explained, many years. And currently being accepted by the market and designed in at much higher volumes.
Okay. So this is something we can continue to be quite significant for the coming quarters and years.
That's what we of course hope for, and if all goes well, that's certainly a possibility.
Great, thanks. And then the final question on, on TCB. I believe, if it's correct, last quarter, you had the first products for pilot production. Any feedback, any visibility on the follow-up orders from that?
Yeah, that's progressing very well. We are currently in the qualification phase, as we indicated, which should be completed by quarter end Q4. That then should lead to orders in the first half, first quarter of next year, to be delivered in the second half of next year. All according to previous plans.
Okay, thank you.
Thank you. We will take the next question from line, Rolf Mahalingam from New Street Research. The line is open now, please go ahead.
Yes, thank you for taking my question. I wanted to ask a question on high-bandwidth memory as well. One of your Korean customers, or one of the Korean memory manufacturers this morning mentioned that they expected HBM to grow at a 60%-80% CAGR over the next five years. And I would love to get your insight into how those type of growth rates compare to the scenario that you laid out at your CMD in June, where you guided for 800-1,900 hybrid bonding tools. So did you take into account those type of growth rates when you set those targets? And how should we think about potential upside from those levels? Thank you.
Excellent. Well, we always interpret these comments as a confirmation of a technology direction, simply promising market growth. How much it will be also depends. No customer has any definite knowledge about the next five years. But key is that it is a mainstream significant growth driver. Those applications are certainly in our forecasts. But as we always, yeah, indicate, every year we update, or if there's a special event, of course, we update those models. But it's too early based on these indications to say, well, that means that the outlook for our systems is significantly higher. Let's first understand these comments as a confirmation that we're on the right track.
Okay, perfect. Thank you. And as a follow-up, switching to logic, your other major customer last week mentioned that SOIC, which is their version of hybrid bonding, will be one of the faster-growing segments of the advanced packaging market next year. Do you see that optimism reflected in orders or at least in discussions that you are having with that customer?
This fits precisely in what we have indicated three months ago, but also at the Capital Markets Day. That the trajectory for adoption and expansion is based on a model 2023 and 2024, are initial volumes and mainstream markets, initial devices, and we've seen that very positively, the AMD devices, for instance, manufactured in Taiwan. And then we indicated the U.S. to follow. And we have also mentioned today, orders received in Q3, more orders to be received in Q4, and also to date in Q4 received, both for logic and for memory. So that all fits with the broader market information.
Great. Thank you, Richard.
Thank you.
Thank you very much. We'll take the next question from line, Madeleine Jenkins from UBS. The line is open now, please go ahead.
Hi, thanks for taking my question. I was just wondering if you could maybe kind of quantify more what you're expecting for hybrid bond orders this year and also next year. And also in the release, you flagged the first order for a kind of leading subcontractor. Could you tell us what end market that relates to? Thank you.
Well, number one is logic devices, and there are many logic devices. Also a combination of logic and memory. Subcontractor is very important, as a next step. And to explain it in the most simple way, that means that higher volumes are expected in the future. Otherwise, pooling subcontractors is not, let's say, successful, for the future. So that means that the customer who is driving this, or customers, but we think it's one customer, but we don't know the name because that is not disclosed, is expecting significant volumes in the future.
So again, broader proliferation for different types in logics, also the first memories, and then the combination, logic and memory, and preparing, also, beyond that for applications in high-end smartphones, portable devices. That trajectory is certainly confirmed in the last three months.
Okay, thank you. Then I guess a quick follow-up on that. Are you seeing kind of any new applications emerge for hybrid bonding outside the sort of high-bandwidth memory or kind of logic plus DRAM use case? I've heard somewhere that maybe millimeter wave will be, will be connected to hybrid bonding. So yeah, I just wanted to see if you've heard anything on that.
No. What I can confirm is, like we shared on the Capital Markets Day, many customers are testing this technology for many applications in, you could say, initial pilot laboratory environment, even for certain camera modules. So the interconnect technology, which allows smaller dimensions, which has the advantage of less heat cycle. So there are many other applications tested in the Center of Excellence , the Applied Materials Center of Excellence in Singapore. As of this month, we also open our own development center in Singapore, next to Applied. Our focus is more on standalone hybrid bonding development. In the Center of Excellence , we also have the integrated line, the cluster tool solution. And the reason for expansion is simply the constant more applications being tested.
So there are more drivers in a broader sense than only logic and memory, and certain chiplets for high-end smartphones.
Okay, thank you.
If you would like to ask a question, please signal by pressing star one on your telephone keypad. We will take the next question from line, Simon Coles from Barclays. The line is open now, please go ahead.
Hi, Richard. Thanks for taking the question. It's just to follow up on the hybrid bonding numbers. I think you said a minute ago that 40 units in the field by the end of the year, but if I look back on what you said in the past, I think at the end of last year, there was 35 in the field. So is that you've only shipped 5 units this year? Just trying to understand how that hybrid bonding shipments have been progressing in 2023, so that we know what we're working with as we go into 2024.
... Well, as we said, it's over 40. It's not yet at 50, but the, but the 30 number was not reached. Let's say a year ago, we were, let's say, low 20s. So it's, it's following, it's following the trajectory which has been shared all along. Not the precise numbers. In the beginning, Taiwan indicated 50 donors. Well, they haven't reached that yet. In a similar way, the U.S. indicated a similar level. But what is more important is, of course, the applications and then having this technology in mainstream next generation devices. And we see that increasing as we go along. And as I said in the beginning, what is very positive is that we see an increase in volume in Taiwan gradually.
We see an establishment of an initial mainstream volume application market in the U.S., and this all follows the 2023, 2024 plans, which should lead to significant volume increase in 2025 and 2026, and that roadmap still stands.
That's very helpful. Thank you. Then, I mean, this has been a pretty severe downturn. China went into the downturn early, and you always talked about six to eight quarters being a typical downturn, and now you're past the bottom. Is that increase in utilization rates as you would have expected sort of for a normal cycle upturn to begin, or is it slower or faster than what you would have thought?
No, it's a typical pattern. And that's why we believe that the TechInsights, but also others, this industry has always in all these cycles, and supposedly we are at cycle number 10 in the past 40 years, and it follows the same pattern. So you see utilization rates at subcontractors go up, and those are the first to expand capacity. And on an ongoing basis, you see new devices, and new devices in low volumes for market qualification. And in a full cycle, you have yeah, major expansions in every direction, but this time it's no different. The only thing which we also try to indicate in the press release, but also in the comments earlier, one never knows the slope or how do you say that?
The extent of an upturn. It all depends on the economic situation. The geopolitical situation is not easy. For some high-end products you see, models connected to China as well. How will that continue? So every cycle in that sense is different.
Great. Thanks very much.
Thank you. We will take the next question from line, Didier from Bank of America. The line is open now. Please go ahead.
Hey, good afternoon, Richard. Thanks for taking my question. I have to say I'm a bit lost. I find it very difficult to understand the nuances in all the things you've said and written in press release, et cetera. So I'm gonna ask a few questions, and it would be great if you could help us. How many units for Hybrid Bonding do you think you will have shipped by the end of this year? Is it still the sort of high twenties you were talking about?
And then second, if you look at the orders you have at hand now for CoWoS, your backlog for hybrid bonding and expected orders for next year, as well as the TCB orders that you expect in the first half, I mean, are you looking at, you know, significant growth in calendar year 2024, or at least directionally, that business goes up, and the core business may be rallied as well? Or, I mean, just to help us understand a little bit, because you're giving us too many hints, and it's just difficult for me at least, a simple analyst, to understand all these things.
Well, thank you, Didier. What is quite, let's say, positive in many ways is the continued adoption and also orders for these hybrid bonders. That means that the customers are ever more designing the interconnect using this hybrid bonding process into next generation devices. If this continues, this could lead next year, and we've indicated that also in the capital markets day, to a significant further increase, simply to prepare for real mainstream adoption in 2025 and 2026, because you first need those capacities established before you can run significant further. In the year 2023, 2024, it's all about preparation, initial volumes, the numbers shipping the mid-2020s, as you indicated, and maybe next year it's more, who knows? That is always with new technology.
No customer will give us a precise number of machines they will need in this phase. So that remains as it is. But the positive message, to repeat that again, is that yes, we receive multiple orders from multiple customers, and that is confirming that this technology is seeing a broader adoption. First logic, now, and more to come. So that bodes well.
Yep. And I just wanted to-
I can't give you detailed numbers in this phase.
Okay. Now, so if I had to take a step back, so you've got one leading foundry that has got a huge install base already and is buying more. Is that, is that right? Number two-
Yes.
You've got two memory customers in Korea that have placed orders for HBM4, and apparently, they are placing more orders, as you said. Then you have, now a Korean subcontractor-
Sorry.
Sorry.
Let me correct you there.
Yeah, sorry.
The Korean memory has not yet jumped on the bandwagon. The US memory has jumped on the bandwagon.
Oh, okay.
Korea, Korea is very close, let's say, the development could lead to, to adoption of Hybrid Bonding in, in next year and maybe also early next year. But that also has not changed from our previous speed of how this Hybrid Bonding adoption will take place. Sorry to interrupt you. Then-
No, no, that's fine. Yeah, sorry, go ahead.
Your next statement, where I interrupted you, was, if you can help me?
No. So then on the subcontractors, it's just the choice of words is a bit unclear. So you announced today, effectively, two subcontractors, as far as I understand.
Yes.
So two OSATs, one in Korea and one somewhere else, and they are, they are buying hybrid bonding systems to test the technology, anticipating major volume production in, let's say, 2025 or something like 2026. Is that, is that the right way to think about it?
Let me be a bit more clear. One subcontractor based in Taiwan for hybrid bonding. There's subcontractor in Korea for 2.5D, CoWoS related. So that's different.
Got it. Okay, but these two guys are OSATs, they are not foundries?
They are OSATs. They're the largest OSATs.
Perfect. Perfectly clear. That's great. And then CoWoS, I think when we talked, when you came to our conference in June, you said that there was a mid-single digit million EUR contribution, so it was fairly de minimis. Now that CoWoS is, you know, growing very strongly and you clearly seem to benefit from it, would you venture in telling us, you know, how much CoWoS could be, either in Q4 or next year?
Well, we mentioned that the projections publicly stated were a threefold increase in the capacity compared to this year. For that, you need more equipment, and that's now happening. It's not only in Taiwan, but also in Korea and in China.
Perfect. And then my final question I wanted to understand: Is the photonics business incremental? I presume it's within the U.S. IDM, but is that incremental to the hybrid bonding business related to their chiplet architecture for servers?
Yes.
Okay.
Yes.
So you've got a broadening of adoption at that customer, and then you've got TCB coming up next year as well. Is that... So you've got, like, a fairly large engagement with that major U.S. customer?
Yes, yes, but it's more than one customer. It's five customers in photonics so far.
Five customers. Brilliant. Thank you so much, Richard.
We've been working on this for more than a decade, so it's not overnight, but that is linked to this 2.5D CoWoS wrap.
Brilliant. Very, very good. Thank you.
Thanks, Didier.
Thank you. It appears no further question at this time. I'll hand it back over to your host, Mr. Richard Blickman, for closing remarks.
Well, thank you all very much, and if you have any further questions, don't hesitate to contact us. All the best. Bye-bye.
Thank you for joining today's call. You may now disconnect.