Start Time: 10:00 January 1, 0000 11:12 AM ET
BE Semiconductor Industries N.V. (OTC:BESIY)
Q3 2022 Earnings Conference Call
October 20, 2022, 10:00 AM ET
Company Participants
Richard Blickman – CEO
Conference Call Participants
Didier Scemama – Bank of America
Francois-Xavier Bouvignie – UBS
Robert Sanders – Deutsche Bank
Ruben Devos – Kepler Cheuvreux
Marc Hesselink – ING
Martin Marandon – ODDO BHF
Charles Shi – Needham & Company
Timm Schulze-Melander – Redburn
Peter Testa – One Investment
Operator
Good morning and good afternoon, ladies and gentlemen. Welcome to Besi’s Quarterly Conference Call and Audio Webcast to discuss the company’s 2022 Third Quarter Results. You can log into the audio webcast via Besi’s Web site, www.besi.com.
Joining us today are Mr. Richard Blickman, Chief Executive Officer; and Mr. Leon Verweijen, Senior Vice President of Finance. At this time, all participants are in a listen-only mode. Later, we will conduct a question-and-answer session. The instruction 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.
Richard Blickman
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 for our third quarter and nine months ended September 30 and also update you on the market, our strategy and the outlook. First, some overall thoughts on the third quarter.
Besi reported Q3 ’22 results which were at the favorable end of guidance but reflected the impact of a new industry downturn. For the quarter, revenue, orders and net income of €168.8 million, €125.3 million and €57.3 million decreased by 21.1%, 18.2% and 24.2%, respectively, versus the second quarter of this year.
At first, revenue and order developments this quarter reflected typical seasonal weakness for mobile applications, but also more general weakness in high end server, data center and general computing applications. Such weakness was partially offset by continued strength in automotive and industrial end user markets and ongoing shipments of hybrid bonding equipment to customers. Similarly, Besi’s backlog of €240.6 million at the end of Q3 declined by 12.6% versus Q2 ’22 but remained at higher than typical levels.
Despite a challenging market environment, we maintained profit efficiency at high levels with gross margins of 62.3% exceeding guidance and a net margin of 34%. We realigned Besi’s production model rapidly in response to changing market conditions. As a result, total headcount has declined by 12.7% and temporary Asian production headcount by 65.2% since the end of the first quarter of this year. We will continue to adjust overhead levels as necessary in accordance with market developments.
For the first nine months, Besi reported revenue of €585.1 million which increased by 1.3% and net income which decreased by 6.9% versus the comparable period of last year 2021. Growth was favorably influenced by increased demand for Besi’s computing, automotive and hybrid bonding end-user markets. Such strength was partially offset by reduced demand for high-end smartphones following a large capacity build in 2021. It also reflected a 37.6% revenue decrease from Chinese customers primarily associated with overcapacity, slower economic growth and COVID-19 related lockdowns.
Orders of €483.3 million decreased by 34.4% as industry conditions materially weakened post the significant assembly capacity build over the past two years. The €14.8 million decrease in Besi’s net income between the comparable periods principally resulted from a 49% increase in development spending as we increased investment in future areas of growth for the next market upcycle.
Our liquidity position continues to build with strong cash flow generation of €185.2 million during the first nine months of this year which supports Besi’s capital allocation policy. We ended the quarter with cash and net deposits of €661.8 million and net cash of €342.5 million that represented increases of 12.1% and 19%, respectively, versus September 30 last year.
Liquidity has improved this year despite the distribution of €351.3 million to shareholders in the form of dividends and share repurchases. We completed our prior €185 million share repurchase program in July and began purchases under our new €300 million program in August. During the quarter, we repurchased a total of 0.9 million shares for an aggregate amount of €45.5 million.
Next, I’d like to speak a little bit about the current market environment and our strategy. As seen in the next chart, an industry downturn began in the second quarter after a long capacity upcycle. Industry conditions deteriorated significantly during the third quarter, highlighted by slowing memory mainstream computing and data center markets, continued weakness in Chinese markets and CapEx reductions in arms [ph] by many of the largest semiconductor producers. The only market which has enjoyed positive growth this year is automotive.
The outlook for the assembly equipment market also turned more negative this quarter as industry conditions weakened, global GDP growth rates decelerated and customer caution increased. At present, it appears to be a traditional industry downturn marked by overcapacity and order push-outs by customers.
Down cycles are typically the periods in which Besi looks to improve its business model and plans investment in those products and technologies which will drive revenue growth in the next upcycle. The announcement of new restrictions recently by the U.S. on sales of frontend and assembly equipment to China has added more uncertainty to the industry outlook. We are currently evaluating the proposed restrictions to better understand whether any of Besi’s below 10 micron accuracy systems could be subject to such provisions.
Strategically, we are accelerating investment in Besi’s future despite near-term headwinds, particularly for our hybrid bonding and wafer level assembly portfolio, as the long-term drivers of our business remain intact and sub 10 nanometer device innovation continues apace. In fact, R&D spending has increased almost 50% this year, highlighting our enthusiasm for Besi’s future growth opportunities.
We see continued interest in hybrid bonding applications as the natural extension of Moore’s law to drive technology gains in new heterogeneous 3D architectures for next generation logic, memory, mobile, automotive and data center applications. Of note, AMD announced last year in the third quarter its first hybrid bonding chiplet which was the starting point for the further adoption of hybrid bonding in the past 12 months. Besi received orders subsequent to quarter end for incremental hybrid bonding capacity and for systems incorporated in hybrid bonding integrated lights. Additional orders are anticipated in the fourth quarter.
Now a few words about Q4 guidance. For the fourth quarter, we estimate that revenue will decrease between 15% and 25% versus the third quarter, reflecting current market conditions and seasonal trends. However, Besi’s gross margin is expected to remain in the 60% to 62% range due to the flexibility of our production model and anticipated product mix.
Further, operating expenses are anticipated to increase by approximately 5% versus the third quarter principally due to higher R&D spending. The midpoint of Q4 guidance implies full year revenue of approximately €720 million, down approximately 4% versus 2021 and a gross margin of approximately 61%, up 1.4%.
That ends my prepared remarks. I would like to open the call for some questions. Operator?
Question-and-Answer Session
Operator
[Operator Instructions]. The first question comes from the line of Didier Scemama of Bank of America. Please go ahead.
Didier Scemama
Thank you very much. Good afternoon, everyone. Richard, I’ve got first a question on hybrid bonding. Could you give us a sense of how many orders you booked in Q3 and how many orders could come in Q4? And in general, just give us a sense of where we are in terms of TSMC making progress on yields, on the alignment and the cluster too, just the general enthusiasm from your customers on hybrid bonding? And I’ve got a follow up. Thank you.
Richard Blickman
Excellent. Quarter-after-quarter, we continue to see increased adoption in the industry more broadly. So, of course, the cart is drawn by Taiwan in that matter. And also in the U.S. it’s gaining traction. We have in a year’s time shipped over 20 machines, of which the major part is in full production. So the adoption rate of architecture using hybrid bonding and also the first chiplets are developing according to a roadmap, which we understood about three years ago is simply in tax. And so Taiwan being the first mover, the U.S. following thereafter in the first stage standalone equipment, and then also the cluster to the integrated time simply following a concept whereby ultra-clean processing will be ever more critical when we move further down the design geometry roadmaps. We are now in the 150-200 nanometer accuracy range, which will move down early next year. As a start with 100 nanometers accuracy requirement, this supposedly is following the design rules for frontend from 7 nanometers today down to 5 and then three to five years down the road, we have to move down to 50 nanometers. This whole roadmap technology, but also adoption, as I said earlier, follows a roadmap timeline which is simply to expectation. Some can also claim that it is accelerating, but if you follow closely, our guidance over time, also the Analyst Day presentations June last year, June this year, the best way to summarize it is that it’s on track. How many machines in total will be required in certain timeframes? We have simply interpreted the demand by the end customers indicated. It should come close to 50 machines in a first round. So more orders are to be expected as we mentioned in Q4, which have been indicated. Then we have next year the start in the initial phase in the U.S. preparing for production volume in ’24 that may also let’s say account for about 50 machines simply based on numbers of devices produced. The major ramp is more and more, let’s say, guidance for ’24, ’25. There are some numbers indicated in Taiwan, for instance, but also in conferences where the full adoption of hybrid bonding and especially the chiplet architecture by that time will require significant more machines. Besi is preparing for that in several ways. Number one, our R&D which we also mentioned in this call, which has increased simply by the ever increasing applications of using hybrid bonding technology. At the same time, our infrastructure, our supply chain to build those machines, we are gaining everyday experience in supporting machines operating in full production environment. So step-by-step, I’d say day-by-day, we are gaining experience and the adoption is continuing as expected.
Didier Scemama
Very interesting. Very useful actually, Richard. So on the order intake — if we look at the shorter term environment, on the order intake for Q4 normal seasonality obviously is down. Would you say that the normal seasonality for Q1, which is no major order intake, would be reasonable, given what we missed effectively last year that big customer that did not place orders, given that you reached perhaps sort of bottom of orders in Q4? And also taking into account what you just mentioned on, let’s say, the orders that may come in for hybrid bonding, is that the right way to think about it?
Richard Blickman
Well, we all know that in current times, there are more uncertainties than just mentioning China, for instance. We are faced globally with significant inflation that could dampen the demand for semiconductors in many ways. But if that uncertainty is sort of caveat for — statistically, there’s, of course, always the first half of the year is stronger than the second half. That has been, let’s say, a rule of thumb for many, many years. So you should then see orders come in, in Q4, Q1 simply organizing the demands and then the delivery into Q2, Q3. Those are the typical patterns. This year and the beginning of next year will evolve. The only thing I can say so far in October are things continued positively. But we also said the downturn started, let’s say, Q1, Q2. So we are three, four quarters into the downturn. Typically, it last six quarters to eight quarters depending upon outside factors. So the summary is a bit — the message to convey is a bit more uncertain than it was, for instance, at the end of ’19 and then early ’20, but then all of a sudden COVID hit the world. Anyway, for us, what’s important is also order levels. If you simply imagine, 200 million Q4 last year, Q1, then dropping down to 154 million, now to 125 million. Revenue, we guided down 15% to 25% versus what we achieved in Q3. So that would mean somewhere 140 or 135 or whatever. But those are still very much higher levels than what we had in the previous cycle. And then with the strong margins supported of course a bit by a strong U.S. dollar, but that bodes well in the semiconductor, yes, typical downturn and then with the positive traction of new technology. So that’s a bit longer answer, Didier, how we look upon what may unfold in the next two quarters. Needless to say that we are ready to ramp rapidly in case that demand is required. But also as we mentioned, if the world turns further south, we have many ways to adjust our model to more difficult times.
Didier Scemama
Very useful. Maybe one final question for me. You did mention on your press release this morning that some of your tools might be subject to the export controls. Apparently, the rules apply to tools that have 10 micron accuracy and below, which is a large portion of your shipments. And China obviously is an important geography for you. So can you perhaps give us a sense of how much of your revenues would be at risk, at least as per your understanding of those rules? Thank you.
Richard Blickman
Well, as we currently understand that, that’s less than 5%.
Didier Scemama
Thank you.
Richard Blickman
A very simple answer. So the traction for the below 14 nanometers is for us certainly not out of China. So that risk is limited. The only thing which from our point of view we’d like to share, it definitely will have an impact on the entire state of the industry. One could argue there will be production moving out of China to surrounding countries, maybe also onshoring in certain cases. That also is a positive thing, because more equipment will be needed. But overall, it may have an impact of slower growth for the industry. The verdict is out. It’s always exciting. We’ve gone through many of these situations in the world. If we go back to Japan in the late ’80s, the start of Korea early ’90s, all the other crisis, the Internet bubble, the banking crisis. But the key is to understand these are always unique opportunities to build a better company.
Operator
The next question comes from the line of Francois-Xavier Bouvignie of UBS. Please go ahead.
Francois-Xavier Bouvignie
Hi. Thank you very much. I have two quick questions, mainly follow up to Didier’s questions. On the hybrid bonding, I’m not sure I followed your comment, Richard. I’m sorry. How many shipments do you expect for 2023? I heard like 50, but you were talking about a U.S. customer only. So I’m just trying to better understand your answer, like how many shipments do you expect for 2023 hybrid bonding? And how many orders do you have today total in your backlog? Just to clarify my first question if it’s possible?
Richard Blickman
Well, let’s say the 50, as I mentioned, is an initial capacity indicated to offer this technology and mainstream production for high end computing. So that’s the start. High end logic devices and connected with that certain chip. But then a wildcard is of course the application in high end smartphones. And that could lead to significant more capacity required. So the 50 as a number is what is required both in Taiwan and also in the U.S. What the number precisely will be in the end is difficult to forecast, because they’re not precise numbers. It depends, first of all, on the continued adoption of hybrid bonding as the next mainstream technology. But number two, as I mentioned, on which applications it will be used. And ’23 is still a year of early adoption. The big volumes are expected in ’24, ’25 and ’26. Taiwan was very explicit about the demand they would expect in ’26, which then is an enormous tenfold increase over the initial phase. So that’s many machines. But that’s also still a few years away. How much we have precisely today in backup? We don’t disclose. We don’t do that for other machines. But as I mentioned, so far we’ve installed not only in Taiwan but also in other places close to 30 machines already this year. So it’s moving in that direction.
Francois-Xavier Bouvignie
That’s very clear. Thank you. So the second question is again a follow up on Didier. It’s on the China restrictions. So direct impact from what I understand is less than 5%. What about the indirect impact? For example, let’s say, why TSMC is not [indiscernible]. It’s not like maybe a lot of below 10 microns or — if some Chinese can’t operate because of these restrictions, did you try to quantify how much it can impact your revenues in total, so direct and indirect as such? I know it’s very challenging, but I thought I would try.
Richard Blickman
Well, it’s a very good question. And that’s why I made the remark that it’s too early to tell precisely because only tomorrow or the day after, I don’t know which day precisely, these new restrictions will be coming into effect. But as I also mentioned, the effects of those restrictions on the industry, it’s hard to tell. But if we look back in historical perspective, and I mentioned in the past 30 years, are there moments in time that always has had a negative impact on the industry as a whole. So a direct impact less than 5%. So that’s very small. But what would happen to the entire industry, we hope, of course, that it is similar to a COVID impact, where first the initial thought was, it would be very negative for the industry, but immediately it turned into many opportunities in growth in other areas. And I also mentioned that you see, for instance, the slowdown in orders from Chinese subcontractors is not only because of Chinese economy and COVID but that also has to do with capacities which are being organized elsewhere. We have mentioned this in previous calls. But this is, of course, accelerating with the measures which are becoming more and more restrictive in terms of growth in China, and that will have a positive impact outside of China.
Francois-Xavier Bouvignie
That’s very clear. Thank you very much.
Operator
The next question comes from Robert Sanders of Deutsche Bank. Please go ahead.
Robert Sanders
Yes. Hi. Good afternoon. I have a question on the smartphone application processor opportunity for hybrid bonding. So my question is basically TSMC is obviously developing a lot of capability in this area, and I suspect they’re working with the lead customer, like example Apple. But is the kind of expertise and intellectual property mainly with TSMC as opposed to with Apple, because obviously from a Besi point of view, you want as many smartphone application processes to use this technology. So is there any kind of barrier from MediaTek or other companies using hybrid bonding quickly, or is there some kind of learning that they need to do before they ramp in 2024 or later? Thanks.
Richard Blickman
That’s a very interesting question we are also discussing among ourselves. It’s hard to tell whether there will be any restrictions in terms of IP ownership. So far, we have not come across that yet. But it could well be. You don’t know who owns them. Is that the TSMC or is that the high end smartphone manufacturer, the designer? Well, there’s some literature in the past month about that which tends to point that towards the high end smartphone, let’s say, leader in the world. But how that will in the end restrict that adoption to others, probably it will be licensed that could also be a solution, which has been done in the past and is still done in many instances.
Robert Sanders
Okay. And then as things stand, the TSMC 10x the initial phase, it sounds like the huge majority of that capacity expansion is for smartphone application processor. Is that right? Or in a broad brush more than three quarters?
Richard Blickman
Yes.
Robert Sanders
Okay. Thank you. That’s it for my questions. Thanks.
Richard Blickman
Thanks.
Operator
The next question comes from Ruben Devos of Kepler Cheuvreux. Please go ahead.
Ruben Devos
Yes. Good afternoon. I just had a question on the frontend. Basically, we’ve been seeing some conflicting updates lately in the industry. In particular, it looks so far that the frontend equipment suppliers are somewhat less impacted by a deteriorating macro environment, whereas the backend, such as yourself, is more feeling the pinch, let’s say. I was curious whether you could help us understand what could potentially explain this difference and why this time it may or may not be different from a previous down cycle?
Richard Blickman
Well, we’ve said in the previous call, and thanks for this question, this is a very positive, let’s say, phenomena. If you would have a significant frontend downturn, you would probably have also a longer backend, let’s say, capacity adjustment. So what typically happens is you can see that statistically over many, many years, backend is simply more shorter time adjustable in terms of capacity than frontend. Frontend fabs take much longer time to build, to qualify, et cetera. So the roadmap for the next three years for frontend remains significantly strong simply because of underlying technology changes in our society and in the whole world. Backend clearly follows more closely capacity and also new technologies. And as we try to explain, there is in certain areas, simply because of enormous growth in 2021, an overcapacity which will be absorbed and you have new technologies which are continuing adoption and simply following the roadmaps for next generation technology and all the derivatives from that. So we have been trying to share since February already that we have seen a peak after eight quarters of growth end of ’21. And we’ve seen gradually Q1 was still very strong, but Q2 clearly evidence by orders and Q3 even more that we are following a typical downturn pattern for demand of assembly equipment.
Ruben Devos
All right, it’s very helpful. And then maybe to follow up on the R&D spend, I think in the press release we’ve also seen that these were somewhat higher this quarter and also for next quarter will be higher. Obviously, a lot of innovation and progress in chip performance will now come from advanced packaging. I think if you look at consensus overall for the next few years, it looks like they expect it to trend down R&D spent relative to sales towards 6%. I believe peers are averaging about 10% of sales. So my question is, is that a valid assumption that the market is taking or how do you think about your R&D spend now that advanced packaging is getting more attention in the semiconductor space?
Richard Blickman
Well, two comments. Number one and this has been, let’s say, a characteristic for many years, our R&D spend is very focused on customers. It’s all driven by clear customer programs. So very focused, which is also the key to our business strategy. Over time, that translates into relatively lower percentage of sales compared to comparable competitors. On the other hand, it has doubled, like our revenue in the past five years. And we have simply told the world also in the Analyst Day that we’ll continue to increase, simply moving the accuracy down from 150-200 nanometers to 100 and then down to 50, but then also the development roadmaps for all of our other 18 different platforms. It becomes ever more complicated. Accuracy is more complicated, devices, thinner packages and that’s an ongoing development effort, which is wonderful. And if you do that focused, on average, it should remain below 10%. Maybe in downturns, it can be sometimes above 10%. But that is simply sharing our philosophy. So it’s not driven by percentage of revenue. But it’s very clearly driven by customer programs. And the engagements with Taiwan, with the U.S., with Korea, those customers demand an enormous R&D engagement. And the consequence of that is to be continued to be selective in what we do and what we do not do. What we do clearly mainstream focused on to be a bit more precise on that and not to be, let’s say, diverted into unique, special solutions. It’s all mainstream; mainstream and product application driven.
Ruben Devos
All right. Thank you very much, Richard.
Operator
The next question comes from Marc Hesselink of ING. Please go ahead.
Marc Hesselink
Yes. Thank you. My first question is again on hybrid bonding. And looking at what you shared for the Capital Markets Day saying that they’re now at the field — the machines are in the field but you expected a quite rapid improvement in the performance from 1,000 units per hour to the 2,000 before the year end. How do you see that? Is that a trend? Is that technology progression going as planned? And also related to that, where do you currently see the competition?
Richard Blickman
That’s also good. Well, as I tried to explain earlier, we’re making significant progress every single day. That’s also why we receive repeat orders. The exact number depends also on the device size. There are certain devices where we do reach the 50 number and also where we are also already at 2,000. But that’s not the key criteria at this moment. The key criteria is yield. Yield at those throughputs and reaching levels above 80%. And to get there is a long, long journey. We started with this around seven years ago. And we’re getting better at it continuously. About competition, we mentioned also earlier, competition clearly Japanese was at the very beginning, a company which has also a wonderful system, but not with the throughput we’ve been told what we can accomplish. And with the adoption increase, also others are telling the world they want to enter into the hybrid bonding technology space. So some have announced that next year they will introduce hybrid bonding in a similar accuracy level which we can achieve today as prototypes, but that only confirms that hybrid bonding is becoming a major mainstream technology for the years to come. So there is definitely activity on the competition, right. I should also mention Korea, which typically wants to have also some independence. So compared to a year ago when there was still doubt on whether hybrid bonding would establish as a mainstream connect technology, that doubt is gone one year later. And that also then convinces others to focus on the opportunities in this emerging market.
Marc Hesselink
Very clear. And I believe already that they’re not going to stop at the current [indiscernible] continue. What does that eventually will do for your HB? Is it logically that you [indiscernible] with the improved performance?
Richard Blickman
Well, it turns up with higher complexity and with more throughput, cost of ownership. You always in this industry forever, the measure is cost of ownership because in a simple way, the customer needs to produce an X number of devices and the key is how many CapEx [ph] does he need to achieve that? So in other words, how many boomers does he need to get that for you? And that is how you sell the value of any product. So if we look at this hybrid universe, clearly 150 is already a major achievement compared to the 1 micron, well, even 3 micron, which is reflected in the current mainstream. That’s a major, major step and much more expensive machine. And going down that curve to 100 will increase only the cost already, and even more or so going down to 50 nanometers. So the trends that will be higher is b, in any case, because of complexity. And when we manage to continue to increase the output of these platforms over time, offering a better cost of ownership that also can increase the price of these machines.
Marc Hesselink
Yes. And then the final question I had was on the [indiscernible] standalone machine or a clustered integrated machine in integrated cluster. What’s the tradeoff that they have to make? Why did they pick one and why did you pick the other? And also — is there differences in what kind of applications you’re going to use it for?
Richard Blickman
Well, in the most simple way to look at this, the biggest enemy of a hybrid bond, because it’s copper to copper, it has to be ultra clean, no particles. Any particle will simply cause that bond not to be established. So ultra clean is the name of the game. If you move any part from one machine to the next, you run the risk that you in whatever way create particles. So to integrate these processes in one tool, that’s the whole concept. Then by definition, you can reach the highest level of cleanliness. So zero particles. But it’s not as easy as it sounds, because connecting these machines into a tool is already a major step. And we’ve come since the early beginning, early last year, the first machines one is being shipped to Taiwan, the other in the U.S. and that will be a development program for at least another 12 months. But ultimately, the vision is that the industry will use this technology in a clustered format. Standalone is the way to get there because then you can optimize every individual process, optimize the technology moving into smaller geometries. So we expect side by side for the next many years that you will see this development you see on the one that’s standalone tools, and then for more mainstream adopted closer to solutions.
Marc Hesselink
Okay. Thanks.
Operator
The next question comes from Martin Marandon of ODDO BHF. Please go ahead.
Martin Marandon
Hi. Thank you for taking the question. My first question is on hybrid bonding. Continuing the profit warnings that we saw recently from CPU players and the slowdown of the server and the PC market, does that change in any way on top of hybrid bonding in the short term, meaning maybe a bit more backend loaded than we expected? And I have a follow up.
Richard Blickman
Well, how this typically, let’s say, is established in this world, today’s volume mainstream is based on technology of the last three to five years, because that is how it develops. And the hybrid bonding is for the next technology and today in a very early and in some customers still in infancy phase. So it does not impact the technology progress. This has more an impact on today’s tools used in those applications. And we see that. As we mentioned the slowdown in the press release but also in the comments, there is a certain centralization where you can see an overcapacity and that has to be absorbed and new technology will be leading the way to a next generation.
Martin Marandon
Okay, very clear. Thank you. And maybe a question on overcapacity, as you mentioned it. On the risk of overcapacity lasting longer than expected, you are already experiencing some inventory digestion at the moment as the market is going down. But how do you assess the risk of overcapacity in the industry with all the fabs which will come alive in 2024, 2025 and the CapEx plans program ongoing [indiscernible] be very swanky issues?
Richard Blickman
Well, to give you a very direct answer, this is a recipe for overcapacity by definition. And I hope that in investor calls in ’24, ’25 this will be a moderate of overcapacity. But we have seen this in the past. In the late ’80s when Japan became too strong, there was a similar slowdown forced onto Japan. And then Korea started with massive support of the whole world, and U.S. in particular, which led to an overcapacity in the second half of the ’90s with the Korea crisis. So it by definition, any factory is a capacity and is based on a model which is always an overcapacity. But then how much that will be also depends on what’s happening in further digitalization in artificial intelligence. If you look at all the business models, which should change society for the better, which requires an enormous growing amount of semiconductors in the next generation technology, there are many who forecast that that will be another period of enormous expansion.
Martin Marandon
Thank you very much.
Operator
The next question comes from Charles Shi of Needham & Company. Please go ahead.
Charles Shi
Hi. Good evening, Richard. Thank you for giving me a chance to ask a couple of questions. So, Richard, I think how your business has been trending this year in 2022 almost exactly follows the same script for 2018. Meaning like in Q3, you got a big decline and in Q4, your decline kind of narrows. And if the same trend follows looks like a Q1, may down a little bit in terms of your quarterly revenue, but it seems like that’s when you could reach a bottom in terms of a quarterly revenue, then there’s probably going to recover from there. But I want to ask you about how you think about ’23 in terms of year-on-year decline. It looks like if the same — you’re following the same script into ’23, your revenue may decline by double digit next year. In 2019, you were down like 30% or something on the top line. So my question is, as you look at the industry cycle, this new down cycle into next year, how to really think about, are you going to fare better or fare worse than 2019? Or what would be the reasons for you to doing better or for doing worse next year compared to 2019 relatively speaking in terms of a year-on-year growth? I hope my question is clear.
Richard Blickman
Crystal clear. Let’s say the better ’23 compared to ’19 depends on the further adoption of hybrid bonding what we discussed earlier in this call. We had zero hybrid bonding in 2019. Number two is the next smartphone cycle with, let’s say, major, major new element features. Those two will impact in a positive sense in ’23, because in ’19 we did not have a smartphone cycle. We had the iPhone X in ’17, which then was over excited in ’18 which caused the enormous correction for us then. And that is different this time, because 2022 is not an iPhone or, let’s say, a high end smartphone year in that sense. So that’s different. The key is, of course, but all of us don’t know, if you look at the economic environment, we had no interest rate at that time. We had no inflation. We had no Korean war. We had some debates with China, but not to the extent we have right now. So as I tried to explain earlier, we have also major risks. So that tells how we see the world.
Charles Shi
Got it.
Richard Blickman
I’ll also add to that because that’s also important. We have a much broader customer base compared to ’19. We already see that in the order levels. We have better margins. Our cost structure has improved significantly. So there are many positive things for ’23.
Charles Shi
Got it. Thanks. Maybe the next question I think, Richard, you mentioned about the downturn, what you want to do. Well, one is developing new products, technology. I mean, that’s a given. The other thing you mentioned, but I don’t think you elaborated much, is this business model improvement? What exactly do you mean with that? Can you tell us how — what are the specific actions you’re trying to do or you maybe have been doing in the down cycle? And how should we think about that that going to impact the next upcycle to your business? Thank you.
Richard Blickman
Well, that’s also an excellent point. COVID has forced us to be much more directly engaged in our supply chain, simply because there were major issues every single day but also because of that, we had to further expand our supplier base, qualified different suppliers, different components. So the focus on our supply chain has increased significantly. And that we already see some impacts of that in a positive sense in this quarter, but also in ’22. So our operating model, which is very much using multiple suppliers strategy that has further expanded, so we have a tighter grip on the supply chain. That creates cost improvements on an ongoing basis, but also flexibility. So those are the key targets. And we simply repeat each time we’re far from perfect. The way we build machines is, let’s say, compared to other industries which are far more tight, like the automotive industry. We also engage with people who are experts in those industries to teach us how we can better manage our total operating model. So we use downturns typically to improve our operating model in every sense. And that’s a never ending challenge in a similar way to product development.
Charles Shi
Got it. Maybe allow me to squeeze in my final question on Q4. So, Richard, I think last year in Q4 2021, you did get some initial orders from high end smartphone applications and do you see something similar this time? That’s one. The other is the OpEx increase in Q4, I know you have started amortizing your capitalized R&D for hybrid bonding as you are recognizing revenue, shaping new tools. How much of that 5% OpEx increase is really coming from increased amortization of hybrid bonding R&D that has been capitalized in the past? So how much of that is actually new incremental organic R&D development activities? Thank you.
Richard Blickman
About 50/50.
Charles Shi
Okay. What about the other question about high end smartphone?
Richard Blickman
Okay. Well, if we would know, we would be able to guide that. But that will be very, very critical to understand. And by the end of February, we will all know much more how this first half ’23 will develop and whether there will be a real cycle in that sense supporting growth in ’23. So we did mention already, we had some long lead item orders in Q2 already for next year to be safe on the ramp potentially. But, as always, there are no guarantees.
Charles Shi
All right. Thank you.
Operator
The next — sorry, carry on.
Richard Blickman
Any further questions? Sorry to interrupt.
Operator
We have two more questions. The next question comes from Timm Schulze-Melander of Redburn. Please go ahead.
Timm Schulze-Melander
Hi, Richard. Thank you for taking the question. I just had three very quick ones, if I may. First, just on hybrid bonding, just wanted to check some of my math. Year-to-date, a high double digit euro contribution to revenues. Is that right kind of ballpark?
Richard Blickman
Yes.
Timm Schulze-Melander
Perfect. In the non-hybrid bonding part of the business, could you just maybe talk a little bit about pricing discussions and pricing dynamics, given the changing backdrop? And then I had a follow up. Thank you.
Richard Blickman
Well, in response to an earlier question, the key in selling this type of equipment is a constant improvement of course of ownership to customers. And that simply means that there’s pricing pressure is always in any business engagement. There’s always a discussion about price. But in offering improved cost of ownership versus previous generations versus competitive choices, results, as you can see, in margins, which are at current levels. It also makes no sense in softer periods to try to sell more machines at a lower price, because that’s not the differentiator. And certainly at the end where we sell our equipment, that’s definitely not the place where we want to be. So in general, because of inflation, we have a hard challenge which is a very good one, by the way, to convince our customers that costs have gone up. We’ve done that this year twice already. Last year, because of all kinds of logistics cost and COVID related costs. So on the one hand, we are able to increase our prices once we can explain the reasoning. But that’s the situation where I think capital goods industry is always in.
Timm Schulze-Melander
Right. That’s helpful. Thank you. And then just on the roadmap, you’ve talked about accuracy. I think we’ve sort of touched on just very briefly throughput. Could you just share what commitments you’ve made to your customers about throughput, about accuracy improvement? You’ve mentioned sort of 50 nanometer and 100 nanometers. So just maybe whether any of that is contractual, or any damages or penalties other than obviously missed sales for missing that roadmap? That will be my last question. Thank you.
Richard Blickman
It’s missing roadmaps. It’s fair to say that with all these programs over many years, there are always setbacks, setbacks because of ourselves not being able to get that right immediately. There are also other impacts. We are in this whole production process. We are — certain elements, so there is an impact pre our involvement and their impacts in materials. So this is complicated environment, which is dependent on many factors. It’s not what you mentioned. As the first, it’s not contractual with dates which are cut off and penalties. These are development roadmap programs where everyone is aware that there are definitely areas still to be proven. It’s design of concept. And then many, let’s say, unforeseen in a way unexpected. But that’s our world and that’s what we like, and that’s what our customers are used to. And then many customers who are excellent in setting the pressure to accomplish what we need to accomplish in the shortest period of time with every help, because it’s very interesting.
Timm Schulze-Melander
Very clear. Thank you.
Operator
The last question comes from the line of Ricardo Romeato [ph] of One Investment. Please go ahead.
Peter Testa
Hi. Actually it’s Peter Testa calling. Two questions, please. One was just, if you could help us on the ecosystem around you in hybrid bonding and just talk a bit about maybe the progress being made, particular test and inspect, metrology to keep up with your roadmap as you’re bringing down the geometry, where they are versus you? How you would expect that based upon what you understand and develop, and immediate next period? And I have one other question.
Richard Blickman
Well, that’s a very important question. Average smaller geometries and then also 3Ds and so stacked devices, where you can’t look inside whether every bond is correct. Metrology is one of the critical factors in making this next step in this industry. On the other hand, we are talking about 150-200 nanometers and then going down. Don’t forget in the frontend, they’re talking about 7, 5, 3 nanometers. So the inspection methods are clearly — we’re also under development in many years progress testing the same way. So how do you test these devices? Functional test, but that’s all in — and I’d say the development of an end product.
Peter Testa
Okay. Would you say that the test and inspect and the metrology is keeping up with you at this stage, or will they shortly be keeping up with you?
Richard Blickman
Well, I think it’s fair to say that we’re all making that progress. Otherwise, the end customers couldn’t make the commitment using this technology.
Peter Testa
Okay.
Richard Blickman
We always try to look at the end user and the end application. Where is this device used? And that tells you how — and the question earlier about high end smartphones, you can imagine there’s a lot of development. But is that already completely, let’s say, established for mainstream application?
Peter Testa
Okay. Thank you. And the other question please was just if you could give a sense on the Chinese subcontractor utilization, what understanding you have as to where they reached and the extent to which you may expect them to reengage in a normal pace in next coming quarters?
Richard Blickman
Well, there are two answers to that question. Number one is we see utilization rates carefully improving from what we mentioned in earlier calls, around 250 and sometimes even somewhat lower and that could lead to around early next year. On the other hand, as we mentioned, there’s a clear trend of non-Chinese customers using Chinese subcontractors to move out of China. And the question will be at what pace? And what is very important is there has to become more clarity on what the restrictions, the latest restrictions of the U.S. imply, because that will determine, of course, also the infrastructure capacity expansion or not in China. And that is today not really known. That’s why we also in our press release simply made that comment that that is additional uncertainty. But we expect that to be clarified pretty soon because all of us are simply asking what does it mean?
Peter Testa
Very good. Thank you for the answers.
Richard Blickman
Thanks, Peter.
Operator
There are no further questions. I will hand back to you, Mr. Blickman, for the closing remarks. Thank you.
Richard Blickman
Well, thank you all for your questions. And if there are any further, then please don’t hesitate to contact us directly. Thank you. Bye-bye.
Operator
Thank you for joining today’s call. You may now disconnect.
Be the first to comment