Good morning and good evening to all attendees. My name is Raman, and from the investor relations side, I will be the moderator for today's call. On behalf of ASMPT Limited, let me welcome all of you to the group's investor conference call for the second quarter and first half of 2024. We would like to sincerely thank you all for your continued support and the interest in the company. Please note that all participants will be on listen-only mode when the management is presenting. We will start the Q&A session only after the management has gone through the entire presentation. During the Q&A session, priority will be given to the covering analysts. As part of our standard disclaimer, please do note that during this conference call, there may be forward-looking statements with respect to the company's business and financial conditions.
Such forward-looking statements could involve known and unknown uncertainties and risks that could cause actual results, performance, and events to differ materially from those expressed or implied during this conference call. For your reference, the investor relations presentation for our recent results is available on our website. On today's call, we have our Group Chief Executive Officer, Robin, and the Group Chief Financial Officer, Katie. Robin will cover the Group's key highlights, outlook, and the next quarter's guidance, while Katie will provide details on the financial performance. With that, let me now hand the time over to Robin.
Thank you, Ram. Good morning and good evening to everyone. It's a pleasure to have you all for our 2024 interim earnings conference call. Today, we will cover the Group's highlights for the second quarter and the first half of 2024. Before we delve into the details of our business performance, let me take this opportunity to give some highlights on the semiconductor industry. In the first half of 2024, the semiconductor industry presented a varied picture. At one end, there was surging demand in the logic and the memory segments, primarily driven by the rapid growth of generative AI. However, the general semiconductor or the mainstream side experienced slower than anticipated pace of recovery. This was mainly due to the tepid consumer spending and electronics demand, further exacerbated by softening in the industrial and the automotive markets.
Amidst this dynamic external environment, there continued to be strong demand for the Group's advanced packaging, or AP solutions. Our first half bookings for AP solution contributed a higher proportion of overall Group bookings on both a year-on-year and half-on-half basis. Mainstream bookings for the Semi segment were up half-on-half in the first half of 2024. There has been an increasing inquiry levels and customer engagements that are typical green shoot recovery signals. However, Semi mainstream order flow has been rather sporadic and lacking the bulk orders or volumes that would more properly indicate a broader-based recovery. For SMT mainstream business, this state boiled over two years and only started to soften in the second half of last year. Even as the SMT consumer market continues to remain soft, our SMT business has maintained its leading market share position.
These differing trends in our business segments and solutions clearly demonstrate the advantage of the Group's unique and broad-based portfolio. Semi and SMT follow different business cycles, so a slowdown in one can be compensated by momentum in another. Moreover, our AP and mainstream solutions have exposure to different facets of the industry, which also helps the Group to navigate through different industry cycles in a resilient way. Let me now move to advanced packaging solutions. As highlighted before, we firmly believe that the Group has the industry's most comprehensive suite of AP solutions that serve a diverse range of applications. Among these applications, the demand for generative AI and high-performance computing continues to be insatiable, and we are deeply embedded in the supply chain that comprises the biggest names in AI. We believe we are in a commanding position to capitalize on this rapidly proliferating demand trend.
Propelled by this trend, our AP solutions grew the percentage share of group revenue. Last year, AP contributed about 22% of our group revenue. For the first half of 2024, AP's contribution increased to around 25%, or approximately $210 million. Within AP, the highest revenue contribution was from thermo-compression bonding, or TCB, followed by system-in-package, or SIP, under SMT and our photonics solution. For bookings, it was a similar scenario, with our AP solutions having the strong order momentum for the first half of 2024, and AP bookings showing significant year-on-year and half-on-half growth, with TCB, SIP and photonic solutions dominating the order flow for AP in the first half. I will highlight some recent updates on key AP solutions and provide details on order wins for this.
First off, let's talk about the high bandwidth memory, or HBM. Here, I'm pleased to update that we won orders in HBM for both our hybrid bonding and TCB tools. For hybrid bonding, the group had yet another major breakthrough. We won maiden orders for two next-generation hybrid bonding tools in the second quarter for HBM applications. This win demonstrate a strong recognition of our technology and competitiveness for this emerging solution. Next, TCB. Our ongoing engagements with key HBM players are progressing well for 12-high and above stacking, and our tools are showing promising results. In addition, we won orders for two tools in July 2024 for our next-generation fluxless TCB solution. We strongly believe that our fluxless TCB is gaining more traction in HBM.
For logic applications, TCB continued its order winning momentum in the second quarter of 2024, as we won orders for chip-to-wafer applications from our leading IDM and OSAT customers. In addition, joint development with our leading foundry customer for our next-generation fluxless TCB solutions remains on track. We are therefore confident that our next-generation fluxless TCB solution will become the preferred choice for ultra-fine-pitch logic applications. Our TCB order winning momentum was also seen for chip-to-substrate applications. Here, there was continuous and meaningful order flow from our leading foundry customer and its OSAT partner in Q2. Based on the robust expansion plan for our foundry and OSAT customers, we remain confident of winning more TCB orders in the rest of the year for chip-to-substrate applications. Let's now look at our AP solutions for photonics and SMT. For photonics, I will cover in a later slide.
Now, for SMT, its SIP solution had strong order wins in the first half, despite overall weakness in the SMT market. SIP demand came mostly for RF modules, for high-end smartphones and wearables from leading global players, and also from AI and server-related applications. For this slide, let me give more color on TCB's potential. We are convinced that TCB has a unique positioning across the market, and I will highlight three interesting market developments here. First, TCB is a key enabling technology to power generative AI's computing architecture as logic chips evolve into multi-chip configurations of CPU, GPU, and NPU to drive AI applications from the cloud to the edge. These complex architectures have multiple and larger chip interconnects that require flexible chiplet handling, and large die bonding capabilities.
Second, a fast-growing HBM demand, primarily driven by generative AI requirements, is accelerating memory stacking from 8 high to 12 high and above, and these have far more stringent bonding requirements. Coupled with the recent relaxation of HBM package thickness requirements, these HBM trends place TCB technology in a sweet spot to intercept increased demand in the near future. This is also in line with our view that mass reflow beyond 8 high will face technical challenges, while hybrid bonding for HBM is still at an early stage. This means a longer runway for TCB to serve HBM. Third, TCB is also well-positioned to capitalize on the proliferation of AP in edge servers and edge devices. This trend is still in the early stages, but will rapidly grow as AI applications move from cloud to edge servers and devices.
These market developments signal an accelerated adoption of TCB and an obvious expansion of the TCB addressable market. You may recall that in our Q2 2023 earnings call, we highlighted a graph in the middle with the dotted trend line. Since then, we estimate the TCB potential getting even stronger, and thus, we are presenting this updated graph now, and we will provide an updated AP addressable market in the coming quarters. Against such a strong market potential, the group TCB solutions are enabling the most demanding industry application to its best-in-class capabilities. These capabilities and their evolution include: improvement in placement accuracy from 3 to 5 microns to less than 1 micron, bump pitch reduced from 30 to 50 microns down to ultra fine pitch of 10 microns for fluxless TCB, and large die handling of up to 70 by 70 mm.
And for HBM applications, handling of thin die with thickness from over 50 microns to less than 30 microns, and chip gap of below 10 microns. These capabilities are reflective of a technology advancement, and our R&D teams are tirelessly working on further improving these capabilities. Another exciting area in our AP portfolio is the photonic solution, which shows promising growth potential that will continue over the next few years, driven by an increased demand for optical transceivers at data centers to meet strong growth in generative AI and expansion of 5G networks. Thus, there's an increased traction for 800 G and higher optical transceivers, which translate into a compounded annual growth rate of about 31% for our addressable market from 2024- 2028.
This high growth potential and momentum is mainly driven by major AI players, who require faster transmission speed, higher bandwidth, and lower latency. Again, this demand backdrop, our market-leading photonic solution, saw meaningful order flow in the first half, as our advanced solutions are capable of handling the complex requirement for higher bandwidth optical transceivers of 800 G and above. And let me add that the estimates of this addressable market mentioned only includes application where higher accuracy is needed for bonding of laser and photodiodes, which is an area of our expertise. The automotive market has contributed strongly to the group in the past few years, and particularly for SMT. While this market has softened of late, it continued to contribute to the highest proportion of group revenue from end market application perspective.
Automotive applications contributed about 24% of the group revenue in the first half of 2024, or approximately $200 million. Both our business segments had similar revenue contributions for automotive. For Semi, the solutions catering to certain niche areas of the automotive supply chain contributed the most to revenue, including solutions for power and silicon carbide modules, and for smart LED heat sinks used in high-end vehicles. SMT also contributed strongly by converting its backlog, while deftly navigating ongoing softness in the automotive market. With those highlights, let me now pass the time over to Katie, who will talk about our group financial and segment performance in Q2 and first half of 2024. Katie?
Thank you, Robin. Good morning and good evening, everyone. Let me take you through the group financials. This slide covers the group's key financial performance for the first half of 2024. Revenue for first half was down half-on-half and year-on-year, due to declines in both our segments. For group bookings, there was solid growth of 11% sequentially, albeit a slight decline year-on-year. Semi registered growth in its bookings for both half-on-half and year-on-year, while SMT bookings were down in line with its market softness. The group ended the first half with a backlog of about $820 million, and the book-to-bill ratio was marginally below one, at 0.98. Group gross margin improved to both half-on-half and year-on-year, mainly driven by Semi. Group's operating margin was at 5.8%.
It was down by 512 basis points year-on-year, mainly due to lower sales, but increased by 212 basis points half-on-half from a low base. Adjusted net profit of HKD 314.6 million Hong Kong dollars followed a similar trend. It was down 49.5% year-on-year, but increased 158.1% half-on-half. The group continued to have a healthy balance sheet at the end of first half, with cash and bank deposits at HKD 5.44 billion Hong Kong dollars, while bank borrowings were at HKD 2.53 billion Hong Kong dollars. For first half, the group revenue of $828.7 million US dollars declined 17.1% year-on-year, and 5.8% half-on-half, due to declines in both Semi and SMT.
From end markets perspective, automotive continued to have the highest contribution to group revenue. Communication was next, mainly due to high-end smartphones and photonics applications. Industrial declined in line with market softness and was mostly in SMT. Group bookings of $808.6 million declined 3.6% year-on-year, but was up 11.0% half-on-half. Semi bookings recovered, and the segment's book-to-bill ratio was above one for the first half. AP contributed strongly to the group bookings. Group gross margin improved to 40.9%, mainly due to Semi's favorable product mix. The margin improved by 67 basis points year-on-year, and by 276 basis points half-on-half. In the second quarter, the group delivered revenue above the midpoint of guidance previously issued.
Revenue of $427.3 million was an increase of 6.5% quarter-on-quarter, mainly due to growth in Semi, and partially OSAT by decline in SMT. I would like to highlight that our AP solutions registered strong quarter-on-quarter growth. Group bookings of $399.3 million were down slightly by 2.4% quarter-on-quarter. It was mainly due to decline in SMT, while Semi registered bookings growth. Group bookings were up 3.5% year-on-year, with strong growth from Semi. Group gross margin of 40.0% was down by 184 basis points quarter-on-quarter, mainly due to declining SMT, while Semi's margin remained stable.
Operating margin of 4.0% was down 360 basis points quarter-over-quarter, in line with lower gross margin and higher operating expenses. That was mainly due to timing of provision for incentive shares. Semiconductor delivered a revenue of $212.5 million in the second quarter, an increase of 20.9% quarter-over-quarter. The IC discrete business unit had quarter-over-quarter revenue increase, mainly driven by TCB. Optoelectronics business unit's revenue increased quarter-over-quarter, mainly due to photonics and high-end automotive headlamps. CIS business units also had revenue growth quarter-over-quarter, but from a low base, and it was mainly driven by high-end smartphone applications.
Semi bookings increased 11.6% quarter-on-quarter to $221.9 million, and it was driven by a strong growth in AP. The book-to-bill ratio continued to remain above 1 for 2 consecutive quarters. I'd like to highlight that since Q4 2023, Semi quarterly bookings have been increasing year-on-year, with Q2 2024's growth at 37%. Semi continued to have a healthy gross margin due to higher volume and favorable product mix. It was at 44.5%, down just 14 basis points quarter-on-quarter. SMT delivered a revenue of $214.8 million in second quarter of 2024. It was a decline of 4.7% quarter-on-quarter, mainly due to softness in the automotive and industrial end markets, mostly from Europe and Americas.
However, revenue from AP grew quarter-over-quarter for SMT. SMT bookings declined 15.6% quarter-over-quarter to $177.4 million, in line with its market softness, and it was mostly due to automotive applications. However, SMT continued to maintain its leading position in market share. SMT gross margin was at 35.6%, a decline of 409 basis points quarter-over-quarter. Its gross margin moderated in the second quarter from a higher margin in the previous quarters due to product mix and volume. This slide highlights our best estimates of revenue breakdown by end market applications for the first half of 2024. These end markets portray the extent of our broad-based portfolio and our wide exposure to diverse end market applications. As highlighted earlier, automotive remained the top revenue contributor for the group.
Automotive has remained in this position since 2022, owing to our comprehensive range of automotive solutions, strong backlog, engagements with a growing base of customers, and our presence in certain niche areas of the automotive supply chain. However, this end market is witnessing softness that may continue. The communication market was the second highest revenue contributor to the group at about 17%, as its revenue grew on both year-on-year and half-on-half basis. Revenue growth was mainly due to high-end smartphone and photonics applications. The industrial market declined as the market softened and contributed about 14% to group revenue. This decline came mostly from SMT. The consumer end market had similar contribution at about 14% of group revenue. Please take note that others include revenue from spares, services, and other applications that cannot be meaningfully identified, and this revenue has remained stable.
Our diverse customer base includes IDMs, OSATs, fabless, foundries, high-density substrate manufacturers, high-bandwidth memory players, EMS companies, and others. This customer base is also spread across the globe, and it gives us the advantage of maintaining a low-level customer concentration risk. For the first half of 2024, our top 5 customers accounted for approximately 16% of group revenue. In this slide, we can observe the first half revenue contributions by different geographies and compare them on a year-on-year and half-on-half basis. Note that China had a stable revenue year-on-year, and its contribution to group revenue increased from 30% to 36%, while Europe and Americas declined in revenue year-on-year.
Europe's share of group revenue was down year-on-year from 30% to 23%, and Americas from 19% to 17%, due to softness in automotive and industrial end markets. We have an existing dividend policy to maintain dividend payouts at about 50% of group's profit on an annual basis, and it will remain fully committed to enhancing shareholder value and returning to shareholders. For our 2024 interim results, the board has declared an interim dividend per share of HK$0.35. This is a decline of 42.6% year-on-year, and is in line with decline of 49.6% year-on-year in net profit. Let me now pass time back to Robin for revenue guidance for the third quarter.
Thank you, Katie. Let me capture our views about the near term. First, we remain very positive about the prospects of our AP business. However, recovery of our semi-mainstream business is taking longer than anticipated due to the tepid consumer spending. Lastly, the SMT business continues to experience a softening market. In view of this continuing dynamic situation, we expect revenue for the third quarter to be between $370 million-$430 million. At midpoint of $400 million, this will be a decline of 9%, 9.9% year-on-year and 6.4% quarter-on-quarter. This quarter-on-quarter decline is mainly due to the lower revenue from SMT. Looking at the longer term, we remain optimistic about prospects and potential for growth owing to our unique and broad-based portfolio.
This confidence is further supported by long-term structural trends of automotive electrification, smart factories, green infrastructure, 5G, 6G, IoT, and AI across cloud, data center, and edge devices. On a broader level, these structural trends are also moving in tandem with a sustained increase in two key areas of increased CapEx spend from nations securing the supply chains via more onshoring and organization preparing themselves to deal with a more dynamic global supply chain. This concludes our presentation for the second quarter of 2024. Thank you, and we are now ready for Q&A. Let me pass the time to Raman to facilitate.
Thank you, Robin and Katie, for your presentations. For asking questions, please either use the Raise Hand function or type your questions in the chat to ASMPT-Q&A. Please ask your questions one by one and limit them to two questions at each turn. With that, can I request Gokul to unmute yourself and ask your questions? Hi, Gokul, can you hear me? And maybe we will ask Tony to unmute yourself and ask your question first. Gokul, can you try to unmute yourself and ask your questions?
Yeah. Hi, can you hear me now, Raman?
Yes, we can. Oh, yes. Thank you.
Thank you.
Okay. All right. Thank you. Thanks, and good morning, Robin and Katie. So, firstly, I wanted to ask about the TCB segment. You seem to have raised your addressable market estimates. I think from my calculation, looks like around 50% higher than your previous target from last year. So could you also talk a little bit about what are your expectations for ASMPT shipments? I think last time we talked about, I think, 350 tools cumulative until 2023 end. That was back in, I think, February of 2024. So could you talk a little bit about what you're expecting for on a forward-looking basis? How, like, when—let's say, when do we hit 500 tools cumulative shipment?
Is that something that will happen by middle of next year? Also, since you talked about fluxless TCB, when do these fluxless TCB tools start shipping? And, do we think, ASMPT can keep, the similar market share in fluxless TCB compared to the flux-based TCB that you have dominant market share in, given there is more noise from some of your competitors now coming into the market for fluxless TCBs? That's my first question. Thank you.
Hey, Gokul, I will break down your questions into two parts. Let me request Robin to address on, you know, the growth we are sort of showing in our slide for the addressable market of TCB. Maybe Robin's expectations and how should we read this? And yes, you said correctly that by end of 2023, we had an installed base of about 350 tools. So how is that progressing? Maybe Robin can update that as well.
Yeah. Let me answer the question first. Yeah, I think we said sometime back that, you know, for the first 10 years of our TCB business and, and the from 12 to 2020-
One.
21, right? From 2012 to 2021, we said that during this period, 10 years, and also for the next three years, from 2022 to 2024, we will ship, you know, incremental amount of TCB tools. So I think we are on track in this from this perspective, right? So that's, I think, that's the first question Gokul asked, right?
Yeah.
All right. What's the second question?
So next is on the Fluxless and what is the progress on that? And you know any indication when the tools will start shipping? And can we give more indication on you know whether we'll maintain a high level of market share when it comes to Fluxless as well?
Yeah. So, Gokul, I think you know that Fluxless is a ultra fine pitch, next generation of TCB tools. We are in the joint development with the leading foundry for chip-to-wafer. I think the results within our expectations is, we pass some reliability tests, and I think we are confident that, you know, our Fluxless TCB tool for chip-to-wafer will be the preferred choice going forward for logic applications. Now, in terms of HBM, I mean, I'm pleased to announce that, you know, we have won two tools for Fluxless HBM applications. So, I think this is also a testimony of our capability.
You know, we are recognized for, you know, TCB technology, not just for logic, but also for HBM. So in terms of, you also asked about whether we will see Fluxless TCB shipment this year. We believe so. Well, in the second half, we should see some of these shipments flowing through, you know, in our, for our business for TCB. Yeah.
Okay, Robin, just to clarify, so this 350 tools by end of 2023, is there any forward-looking guidance that you can give us, in terms of what, what kind of run rate are we going to see, given that you now see an expanded TAM, by 2027?
Yeah, I think I think going forward, I mean, looking at... By the way, the chart is a trend chart. It's not, you know, it's, it's not really the- it's a trend chart for the addressable market, right? So don't try to use a ruler to try to measure what is the incremental between the last, you know, last update and this one. It's just showing you a trend. Now, yes, I think we are confident that going forward, TCB will be a key contributor, you know, for our AP business. We believe HBM 12 high and above will have to use a TCB because of the limitations on mass reflow, which we mentioned many times.
For HBM, for 12 high and above, for logic application, chip-to-wafer will be, you know, also, will need to also use the TCB, you know, because there are more, more chips to be packaged at the chip-to-wafer level. So I think, going forward, TCB will be a key driver, you know, for AP business.
Okay. So just, just to clarify, I think, I can put that as my second question is, one, let's say your 2027 addressable market expectation, how will you rank order HBM, logic, foundry, and OSAT, and IDM, if you take those three categories in terms of most important within the TCB shipment? Is HBM up on top, followed by logic and OSAT and then IDM, or is there a different order? And also for HBM, are you already ship-qualified for 12-high production at any of the HBM manufacturers? Or are we still talking about, mostly sampling tools and not yet qualified for production? Because the production will obviously start coming through with a much bigger, much bigger, volume of tools.
Hey, Gokul, I'll break into two as well. So, your first question of this part is, in 2027, as per our expectation of addressable market, and if everything goes well, how will the ranking look like between, say, our major, you know, customer groups, from HBM to OSAT, IDMs, in terms of the pecking order?
Yeah. Now, looking at the, Gokul, looking at the number of interconnect for HBM versus the, logic side, I think naturally the potential for, TCB application in the HBM, in the HBM arena will be definitely higher. Followed by, you know, the emerging, you know, AI trend for data center. So logic, you know, 2.5D packaging logic will follow, will follow, suit after HBM, and then perhaps, you know, the HPC, market arena.
Okay. The other part of the question is whether we are qualified already at any of the major HBM players for 12 high, and our tools which are planned, and Robin talked about, are they still in the sampling phase, or evaluation, or this going more towards the production?
Yeah. So Gokul, I think you are aware that we've been saying we already have a couple of tools in the key HBM player doing a 12-high and above. So we are qualified for 12-high and above, for sure. And with the recent win in the 2 tools, you know, for Fluxless HBM, I think we, you know, we with that win, I think we will establish ourselves as a serious player in the HBM application market for TCB.
So, is that for multiple customers, or is it just for one customer on the HBM 12 high?
The first couple of tools which has been around for a while is for one customer, and the recent win is for two customers.
Okay. And do you think you will be the tool of record, the main tool for that 12 high process, or do you still not know that?
Yeah. I, I think let's put it that way, Gokul, we are confident of our capabilities... Right. So I think with this, these two wins in terms of Fluxless TCB's HBM application, we hope to establish ourselves, and we believe we can in the HBM market.
Okay, thank you very much. I'll go back to the queue.
Tony, can I ask you to unmute yourself and ask your questions?
Yes. Thanks for taking my question. I think for the outlook of ASMPT as SMT, do we have any expectations how the recovery pace in third quarter and in the second half of this year? Thank you.
Yeah. So I think, if you look at our guidance for Q3, we are getting HKD 400 million for Q3. That's down 10% year-on-year and 6% Q-on-Q, mainly due to SMT. As we said, SMT, unfortunately, the overall market has softened, and also the automotive market and industrial market, which is our stronghold, basically, typically, are also soft. So for that reason, we are facing some headwinds in terms of SMT business. However, for semi, we expect the Q3 revenue to be kind of flattish for semi. And for semi, we continue to be very positive for AP business, as I mentioned earlier.
However, you know, the recovery of the semi mainstream business is taking longer than expected, you know, due to reasons I explained in the opening remark, you know, tepid consumer spending. But we do see, to be honest, increasing, you know, wire bond and hybrid, wire bond and die bond business for consumer-related application. And this is a sign that, you know, hopefully it's a sign that it's some green shoots that should signal that the, you know, some volume is coming back in the next few quarters. We don't know for sure, but we do see some increasing momentum in terms of wire bond and die bond. So I think that's the picture I can give to you for Q3.
Tony, do you have a second question?
Yes. Do we have any color of the booking of the third quarter?
Yeah, we don't guide as usual, but certainly we can give you some color. Now, we think our group bookings for Q3 to come in at a low single-digit Q-on-Q decline. Again, we look at Q3, the decline in bookings are mainly due to SMT. But we expect semi to be flattish and but with an upside bias for semi. So there's a possibility that semi will perform better than flat Q-on-Q for Q3 booking. Again, you know, SMT decline really due to the overall weak SMT market, and in particular for SMT, our strong market segment for SMT in terms of automotive and industrial softening. Having said that, we also believe at this point that SMT booking is probably near the bottom already.
Now, something worth noting about semi, I think Katie mentioned before in the opening remark, is also that our Semi-quarterly bookings have shown year-on-year increase since Q4 2023, and this trend should continue into Q4-Q3 2024. So from that perspective, in terms of the good trend now we are seeing for semi, we believe semi bookings have probably reached a bottom, supported by the very strong booking momentum for AP and also the improving sentiment for the mainstream business. So this is, some color I can give, give you for Q3.
Thank you. That's all my questions. Thank you.
Next, can I request, Juan, can you unmute yourself and ask your questions?
Thanks, Raman. Thanks, Robin, for taking my question. My first question is just wanted to hear your thoughts really on, how you see equipment throughput for TCB improving, down the road, because it has a lot of implications on the market term and where ASMPT is positioned. Are you able to increase your prices to, if, say, a customer, you are able to increase your throughput as well? Thanks.
Yeah, generally, not just for TCB, but I think generally for equipment market, you know, the next generation of two typically have better capabilities in terms of placement accuracy, in terms of throughput. And from that perspective, customer typically are more forgiving. They, you know, they are willing to pay more for more advanced tools. I think that's a general perspective, applicable also to AP tools.
Sure. Okay. And then my second question is regarding in memory itself, is there a difference in intensity of throughput usage between a customer that's adopted a mass reflow MUF process versus a TC NCF process? Or is there really no difference in terms of the orders you expect from different customers in memory?
So I think for AP, as I said many times, you ought to-- I mean, it differs from customer to customer, right? So AP is very, AP solutions are very customized, so, so we cannot generalize, you know, what is the output for, TCB from one customer, with the other. Now, typically, you talk about mass reflow. Typically, mass reflow has a higher, throughput compared to TCB. The reason is that mass reflow are typically just pick and place, and they let go the die. Whereas TCB, you have to pick and place and bond the die, with what we call in-situ bonding. So obviously, TCB will take a longer time, compared to a mass reflow.
Got it. That's all my questions for now. I'll hop back to the queue. Thank you.
Next, Donnie, can you unmute yourself and ask your questions?
Yes. Thank you, management team, for taking my question. My first question is regarding to your forecast or your roughly outlook on the TCB shipment trend into next year versus this year. The reason is because I think this year, on substrate, TCB still accounts for the majority of our TCB shipment. You can correct me if I'm wrong. But for next year, if we're considering maybe we can get more on-wafer TCBs and, you know, HBM TCBs, should we expect that the volume will be trending up next year? Or, you know, there is a possibility that the on-substrate TCB shipment may decline, considering, you know, lots of offset companies have started expanding capacity aggressively in 2024 already. Thank you.
So, Donnie, just to summarize your question, basically, trends in terms of shipment and volume going from this year to next year for both our chip to substrate and chip to wafer, broken down, and which side can be higher. Plus, will there be any weakness in chip to substrate side going forward?
So Donnie, I hope I cannot be too granular, but I can give you a general picture. If you look at our trend line, right? So it's sloping upwards, and in fact, our trend line is now steeper than what we have presented in the last June presentation. So I think based on this, the outlook for TCB is indeed very bright, okay? So with that kind of trend line, we expect, you know, the TCB contributions in 2025 and beyond to continue to be strong compared to the prior years. Okay.
Okay, understood. And another thing, maybe just to follow up on this question, is like, I'm not sure you have provided any specific timeline that whether you can start shipping, you know, the mass volume of the on-wafer TCB to leading foundry. If not, could you elaborate more on that again? And also, have you seen any more meaningful on-wafer TCB orders coming from OSAT companies after, you know, you start, potentially, you start to ship to the leading foundries, maybe in the coming months?
So, as I said earlier, our ultra-fine pitch fluxless TCB for chip-to-wafer application is having this joint development exercise with the leading foundry. So, we hope that that will turn into some kind of order win in the second half of 2024, but we believe the volume would not be big this year, but it will be definitely more substantial going into 2025. Because, you know, looking at the demand for AI architecture, you know, so, so purely based on that, I think the volume will increase. What's the second question from Donnie?
I think-
Yes.
Yeah.
After you successfully ship into a leading foundry sometime in the second half this year, do you see other OSATs will follow TSMC's suit to procure your on-wafer TCBs?
Yeah.
As I think TSMC is aggressively asking some other OSAT companies to expand the capacity, you know, in terms of the outsourcing the customers', products to OSATs.
Yeah, Donnie, I really cannot answer this question on behalf of a customer. Maybe you can ask them. But looking at the chip to substrate, you know, the leading foundry has already subcon to the OSATs, and we are also winning orders for TCB chip to substrate application from their OSAT partners. Yeah.
Okay, thank you.
Next, Sunny, can you unmute yourself and ask your questions?
Yeah, sure. Could you hear me okay?
Yes, yes, yeah.
Thank you very much. Well, so my first question is to follow up on TCBs into HBMs. And so you talk about these fluxless TCBs, how likely do you think fluxless would be required for HBM3 12-high and HBM4? I understand from a technology point of view, fluxless definitely provide better performance and better bonding position. But based on your engagement with the key customers, do you think they will certainly move into fluxless into 2025?
Yeah. Good question, Sunny. I think for ASMPT fluxless solution, we are rather unique because our tools are, in a way, backward compatible and forward scalable. In a sense that we can, not literally, but we can switch off, switch on the fluxless function. So if a customer doesn't need, you know, the fluxless capability with switch off, they can, they can just use it as a normal TCB. But if they need it for a 12 high and above, or 16 high or even 20 high, they can switch on the fluxless function. So, so we have a rather unique solution for our customer in terms of HBM application.
Got it. And so I have a follow-up on the competitions for Fluxless TCBs. My understanding is that the key know-how for the tools is in terms of how to deal with the oxide on the surface of wafers or interposers. And every vendor has its own recipe.
Yes.
So how should we think about the competitiveness of your offering? You mentioned you have a joint development with the leading foundry. If your solutions turn out to be preferred by both logic and HBM customers, will it be easy for the others to replicate your solutions?
Yes, certainly. I think the short answer is yes, Sunny. So that's why winning, you know, fluxless TCB for the first two tools, you know, with two customers, is important. It sort of opened the gate for us to establish ourselves as a serious player, serious player in the HBM market.
I see. And so you say that these orders of two tools for fluxless are for two different customers for HBMs?
Yes. Yes, yes, for-
I see.
Different customers, yes.
I see. And so, earlier, when you mentioned that for Q3 bookings, semi is flattish but with upside bias, is that relevant with the possible orders that may come through later in the quarter with regard to HBMs?
We think so. We hope so. So as I said, once we get into HBM, hopefully this, you know, this will become a better known in terms of HBM application. I mean, we are already well known for logic, so with this win, we hope it really open the gate for us for HBM application. Yeah.
Got it. Thank you. Sorry, just one last question, also on HBMs. I guess, before, there are some supply chain feedback, that your tools have pretty good, bonding position, but throughput maybe, is not as good as some of the of your competitors. And therefore, you couldn't really ship, many tools, to your customers. So I wonder, on that part, are your tools improving? How are you resolving the issues?
Yes, as I said, we don't comment on competitors for sure. But we are confident of our own tool, so I hope you are comparing like to like. Don't compare TCB tool with an MR tool, which I mentioned earlier. So that's not a fair comparison, right? So if you have to compare TCB bonder with a TCB bonder. And also, please note that a TC bonder can also be used for MR function, for example. Right? So please compare it on the like-to-like basis. That's my advice to you.
Okay, no problem. Thank you, Robin.
Yeah.
Next, can I request, Simon, can you unmute yourself and ask your questions?
Yes, great. Can you hear me well or-
Yes.
Yeah, great. Yeah, so far we discussed well with the, all the TCB team. So very quick, quick follow-up question is, going forward, the memory, I mean, the TCB for memory portion, could it be bigger than the logic then, going forward?
Market size-wise, yes, definitely, Simon. As I said earlier, in answering Gokul's question, I believe, right? Just purely looking at the number of interconnects for HBM, you know, you started with 8 high, 12 high, 16 high, 20 high, and then multiply by the number of HBM that will be employed, you know, in the AI, 2.5D kind of packaging structure. Right now, maybe 6 HBM, 8 HBM. I think in the future it can go up to 10 or 12. So looking at the HBM trajectory, certainly, I think the potential for HBM TCB application is larger than the logic side.
Yeah. But the issue is like this: when you look at the HBM history, 8 high for the past few years, and then we are talking about 12 high-
Mm-hmm
... 16 high, even more than 20 high. So the question is, you are previously sold or shipped the TCB for 8 high, 8 high, should it be discarded once the memory makers focus on 12 high, 16 high? Or any chance to reuse all the TCB bonders, which was for 8 high, 8 high?
Okay, Simon, maybe I correct you a little bit here. So we do have a few tools at one of the major HBM players, HBM player, but our tools are actually for 12 high. Our tools, TCB tools, they are not really for 8 high. But maybe I can ask Robin if he wants to comment more, you know, how migration would be from 8 high to 12 high, and the relevance of TCB there.
Yeah, Simon, I... Earlier, I also talked about the, you know, some unique feature about HBM tools, right? We are, as I said, backward compatible in the sense that, you know, our tool can use for 8 high as well as 12 high and above. So, for 12 high and above, we think that the fluxes, going forward, fluxes will be used more and more for 12 high and 16 high, depending on as the industry migrate from HBM3 to 3 to 4, right? So but our tool are fungible... Right? So, if customers want to use our tool for a lower height, they simply, not literally, but, and just simply say, they can switch off the fluxless function, and they use our tool for 8 high.
So from that perspective, we offer a very compelling solution for our customer in HBM application.
Yeah, yeah. Yeah, very clear. So just to clarify your previous comment, overall, the currently you are working with the three different HBM memory makers, not two, right? One, and then plus two means three-
Yeah.
different HBM memory makers.
Yeah.
Correct?
Yep.
And then if these guys rush your great TCB bonder, what's the lead time? Like, three months is enough, or you have to spend five or six months? So what's roughly the lead time-
For-
-to deliver your tool from?
Yeah, for TCB tools, typically, 6 months, 9 months, depending on the configuration. Typically it's longer than the traditional wire bonder and die bonder tools, which is 3 months, right? So for AP tool typically long. Yeah.
Yeah. The pre-order should be very important to predict your maybe two or three quarters later earnings, right?
That's right. That's right.
For your TCB.
Yes, sir.
So overall, so far, so good. So maybe, I know every quarter, earnings call, you update the long-term trend. So time to, you know, I think to hear again, your long-term trend. So, so far, up to last year, 350 units TCB bonder shipment.
Mm-hmm.
And then the year, maybe another 50 unit or any rough idea the number, because, you know, we are very number-oriented people. Thank you.
Yeah. We don't give really forward-looking, but looking at the trajectory for TCB, I think the future for TCB is bright. Yes. It will continue to increase in terms of application.
Okay. Thank you very much, Robin.
Thank you.
Can I request Daisy to unmute yourself and ask your questions next?
Yeah. Hi, Robin. Can you hear me?
Can, yes.
Yeah. So, my first question is regarding the geopolitical tension. So, last week, I think you must also read that the U.S. wants Netherlands, Japan to further restrict the transfer of chipmaking equipment selling to China. And, as of first half of this year, China is the largest market, 36% of your total revenue. So, may I ask the impact of the potential stricter rules of selling equipment to China, and any mitigating actions that taken or will be taken by the company?
Let's put it well, I'll answer you in a very, in a more generic way, right? The Group, you know, we ensure that our global operations are in compliance with all applicable laws and regulations. And, if you recall, there was an update sometime back related to semiconductor manufacturing equipment, explicitly mentioned that, you know, front-end IC production equipment and component assembly accessories as the scope. So back-end equipment, which assemble and test or package ICs, have been excluded. So as you're aware, we are at the back end, we are SMT. So, I think this is a positive development for ASMPT from that regulatory standpoint.
Oh, got it. It's very clear. And my second question is regarding the glass substrate. So glass substrate is kind of the new technology in advanced packaging, and compared to the organic substrate, glass has an advantage of, like, lower warpage risk. So, will TCB still have a like, like a shift, TCB can still be used for the glass substrate or other equipment can, will be used?
Yeah, I think that's a good question. I think going forward... Yes, the short answer is yes, TCB can bond on glass as well as other substrates, so, pretty flexible tool for TCB. I think going forward, you're right. I think a glass substrate probably would be used for some kind of panel bonding rather than wafer bonding. This is a trend that we also see. Yeah.
Okay, got it. That's all my questions. Thank you.
Can I request Leping to unmute and ask your questions next?
Hello. Thank you for taking my question. So, first question about your Hybrid Bonding tools. So can you share some color on, you mentioned that 2 Hybrid Bonding tool for the next generation tool. The magnitude of the Hybrid Bonding tool compared with the current TCB, and is it for 1 customer or for 2 customers? Thank you. This is the first question.
Yeah. So the Hybrid Bonding tool that we won recently is for one customer, two tools. But I must say that, you know, HBM, right now, the way we see it, TCB will be, actually the, the sweet spot. So for hybrid bonding, for HBM, we believe these customers are very forward-looking. You know, they probably one or two to do some kind of research and development, more than those kind of high-volume manufacturing. Yeah.
Okay. The second question is about your... Can you share some color about the revenue guidance by countries? So I saw your second quarter, the China proportion increase quite significantly versus last year.
Yeah.
Does it mean the recovery of China is stronger than the rest of the world? Thank you.
... Yeah, good question, but, I can't tell you quarter by quarter because it does fluctuate. But, I think for the near term, as I said earlier, right? So, we see more green shoots coming from the consumer area, and typically, these are more Chinese-based customer. So, we won't be surprised that, in the near term, the contribution from China will increase vis-a-vis other areas. Now, I mean, we talk about relative, right? So if you look at SMT, if SMT has softened, typically SMT is very strong in Europe and America. So, when we look at the near term, in terms of geographical contribution from our revenue and booking, we see more, probably more, is in Chinese. You know, China contribution will come up, relative to, say, U.S. and the European region.
Yeah. So related, is a follow-up. Do you see some order from the Chinese memory customer, since we see other equipment vendor from in Japan, they see some quite strong order from the Chinese memory. Do you see similar things? Yeah, thank you.
Yeah. Unfortunately, due to confidentiality reasons with our customers, we observe this very strictly, and customer ask for that, so I can't comment, you know, where these customers are coming from.
Okay, thank you.
Arthur, can I request you to unmute yourself and ask your questions? Hi, Arthur, can you hear me?
Hi, yes. Can you hear me?
Yeah. Now, can.
Yeah.
Go ahead.
Yeah, thanks, Robin and Katie, for sharing the color on TCB. So I want to switch gear to the high-precision photonic opportunity. You just comment that there's a meaningful order flow.
Mm-hmm.
Can you also share with us the machine content value compared to the TCB? So can we expect this is also a pretty high content value product line?
Compared to traditional tools, yes. But it's not, the ASP is not as high as TCB, for sure. Yeah, in terms, because TCB, you know, has higher, much higher volume requirement compared to a photonics tool. Hopefully, that answers-
That's good.
Yeah. Thank you.
Yeah, and when you say the meaningful order flow, when the timing will kick in? Is the end of this year, or is next year?
It has already kicked in, Arthur. Actually, yeah, it has already kicked in, in the last two quarters.
Mm-hmm.
But, but we see increasing momentum in this area. I think that's not surprising. With so much, build out of, a data center, right? You, you need, you know, the transceivers, you know, to transmit the information.
Mm-hmm.
Yeah, so we see the trend. So it's a good trend. That's why we want to call it out this time around, to let you guys know that we are a major player in the area. And we believe we have a and we've been winning market share in terms of photonics application for 800 G and above.
Got you. And my second question would be on the board level structure trend you mentioned. You believe that the country will add capacity onshore? And then can you share with us which segment is more like a SMT or is more like in the semi solution?
Sorry, Arthur, the first part of the question wasn't very clear. Can you repeat your question, please?
Yeah, so it's on the board level structure trend. You mentioned that a lot of countries will add capacity onshore, right?
Yeah.
So which would be on the SMT side or would be on the semi solution side?
Both, Arthur. We see both happening, right? So for both SMT as well as our semi business, yeah.
Okay, okay.
Mm-hmm.
Probably a quick follow-up on SMT. We see there's some lag from the TSMC from the SMT, but there's also a big opportunity on the HBM. Maybe in the longer term trend, can you see the HBM opportunity actually offset the SMT weakness?
Okay, so you're asking whether the HBM opportunity can offset the SMT weakness? And first part, first part, what you're saying, let me clarify, I think. Our SMT definitely is also relevant for say, you know, AI or those kind of applications, but at the moment, not so much on the HBM side. So, whether it can- HBM can offset the SMT side, I'll let Robin comment-
Yeah
... because this is a bit forward looking, yeah.
Yeah. So interesting question, but, you know, Arthur, let's put it that way. As I mentioned earlier, HBM, the way we look at it, HBM potential is gonna be significant going forward, you know, because from 12 high to 16 high to 20 high, and also the... We believe there's also a pull-in by all the HBM makers, you know, in terms of expansion, due to the, you know, surging demand in the AI architecture. So a lot of HBM are required. So I hope I answer you in a more generic way. I cannot answer you in a specific way, the way you ask.
Thank you. Thank you. And thank you for your color. I don't have any further question. Thank you.
The last set of questions, can I request, Gokul, to ask your questions?
Yeah, hi. Just going back to HBM. So Robin, you mentioned the fluxless TCB tool will be quite flexible.
Mm-hmm.
Are you anticipating that your HBM customers will change from mass reflow to local reflow at some point in 12 high or 16 high, which requires a lot more TCB tool usage? Or you're thinking that they are still going to be remaining at the mass reflow, and still you can design in your TCB tools even for the mass reflow process?
Yeah, Gokul. It's our view that, you know, the MR, the mass reflow solution will face some kind of a challenge as the industry continue to stack from 8 high to 12 high to 16 high. Primarily because the mass reflow tool, as I mentioned or alluded earlier, is purely a pick and then place, you let go, you know? So TCB can handle, you know, warpage better than MR2, right? So especially for two, I think we have very good capability to handle very thin die, you know, so we don't break the die when we pick up the memory die, which can be as thin as 30 micron, right? So the way we look at it is this migration of technology from Mass reflow to TCB is quite inevitable, you know, due to the technical requirement for 12-high and above.
Okay.
In terms of thin die handling, in terms of chip gap, requirement, I think all, all this adds up, you know? So, so TCB is really in a sweet spot for this.
Okay, so are some customers already telling you this is going to happen, or, you still need to get confirmation from customers at, 12 high?
Yeah, I think, based on our engagement with customer, of course, customer don't tell you outright, okay? But based on engagement, I think this is some of the signals that we're picking up.
Understood. One question for Katie is on the gross margin side. We still see quite a bit of operating leverage on, this time on the SMT solution, SMT Solutions business. We saw that in second half last year for Semi Solutions, and you took some cost actions there. Do we expect to see some cost actions being taken in SMT Solutions, given the downturn seems to be more protracted, and it's starting to affect segment margins? And lastly, I think, given your ample cash balance, do you consider any stock or share buybacks or dividends are still going to be the primary way of shareholder return policy?
Gokul, I think there are two questions if I unpack those. On the margin question, for Q2, as you noted, SMT had basically a negative impact due to the lower mix of automotive industrial versus communications. As we go forward, we always keep cost measures in mind, but at this point, I don't have anything to announce, but that's, you know, just typical operational muscle that we have. And then the other question you have is on share buyback, et cetera. So we announced the interim dividend, and we do have the dividend policy of 50% of earnings. Usually what we do is we look at this at an annual basis.
So, as we go through the year, we will consider the total dividend payout for the year. At this point, in terms of share buyback, I do not have anything to share. I hope I answered your questions, Gokul.
Yeah, thank you. That's, that's clear. Thank you very much.
Thanks, Gokul. With that, we will conclude our Q&A session. Let me request Robin to say a few words and then officially we'll conclude the call.
Thank you, guys, for all your questions. Allow me to quickly highlight, you know, the key takeaways, you know, from today's call. First, semis quarterly bookings have been increasing year-on-year since Q4 2023, which is a positive trend. Second, the group's AP solutions continue to be a bright spot. The overall TCB market is looking brighter, and our best-in-class TCB capabilities place us in a strong position to capitalize on this. In HBM, we won orders for both TCB and hybrid bonders. In logic, our TCB continues to maintain a commanding position in that space. Last but not least, photonics is an area to watch with its high growth potential. This concludes our call, and I hope to see you all in the next quarter. Thank you very much.