ASM Pacific Technology Ltd

HKEX:522

[Audio Gap]

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 guidance for the next quarter, while Katie will provide details on the financial performance.

With that, let me hand the time over to Robin now.

Thank you, Romil. Good morning and good evening to everyone today. It is a pleasure to have you all on our earnings conference call for the third quarter of 2024. Before we go through the details of our business and financial performance, let me take this opportunity to give some highlights on the semiconductor industry.

Recovery continues to be rather uneven for the overall semiconductor industry. At one end, non AI-related cyclical semiconductor demand is recovering at a slower pace than anticipated. This includes consumer, computer and communication end market applications. In addition, automotive and industrial end markets continue to remain sluggish, owing to ongoing inventory digestion. These trends continue to impact the group's semi mainstream and SMG businesses.

Let me provide some color here. Even though our semi mainstream business had quarter-on-quarter bookings growth this quarter, we still feel that the order flow remained rather sporadic with volumes that would reflect a broader-based recovery. Overall, where our customer utilization rates have seen some improvement this quarter, these utilization rates are near or at levels that would trigger volume equipment orders.

Looking at SMT, this overall market continued to experience softness with low booking levels. Despite this, our SMT team maintained its leading market position this year. So there is a snapshot of the AI-related cyclical semiconductor demand. At the other end of the spectrum, demand for generating AI was [ vigorous ] driven by significant capital spending from major AI players globally.

The accelerating adoption of AI continued to boost demand for advanced logic and memory packaging applications. Against the rapid proliferation of this AI megatrend, advanced packaging or AP solutions continue to benefit, with AP bookings remaining robust this quarter, supported mostly by our thermal compression bonding, or TCB, and photonics solutions.

I need to reiterate that our unique and broad-based portfolio continues to be an advantage for us. As the group businesses follow different cycles, the slowdown in one can be compensated followed by momentum in the other. Moreover, as our AP and mainstream businesses have different industry exposure, AP's strong momentum helps to partially mitigate the impact of weakness in mainstream. With that overview, let me now move to advanced packaging solutions.

It's worth reemphasizing our firm belief that we have the industry's most comprehensive suite of AP solutions that serve a diverse range of applications. As I've mentioned, the group experienced strong demand fueled by high growth from generative AI and high-performance computing application. And this demand was across multiple AP solutions in our portfolio.

Let me share some highlights. First up is our star performer, our TCB solutions. TCB continued to make the highest contribution to both the group's AP bookings and AP revenue for this quarter. Looking at TCB for logic applications, we continue to win orders this quarter for chip-to-wafer application from a leading IDM customer. For our next-generation fluxless TCB that caters for ultrafine fish chip-to-wafer logic applications, the joint development with our leading industry foundry customer is still ongoing.

For chip-to-substrate applications, we continue to have a commanding position as we serve our leading foundry customer and a supply chain partner. There were many full TCB orders in this quarter from this leading foundry's OSAT partner. We have also commenced high volume shipment of our chip-to-substrate [ TCB tools ] to this OSAT customer this quarter. Lately, our TCB momentum intensified for high bandwidth memory, or HBM, winning a number of TCB orders from several HBM players in this quarter.

In October, we had a significant breakthrough, winning a bulk TCB order from a leading global HBM player. This volume order will cater the customers to high demand ramp for HBM 3E and the TCB tools will be delivered in the coming quarters. These promising developments affirm our TCB leadership in HBM and position us well to partner with a wider reach of customers for more order wins.

Let me also provide some detail in our unique TCB capabilities that help secure these HBM order wins. You may recall that the last quarter, we mentioned our TCB can handle team memory die with thickness of less than 30 microns and can deliver cheap requirements of below 10 microns. In addition to this, our TCB has capabilities to enable seamless upgrades to fluxless application for 12H and beyond stacking requirements.

Last but not least, our TCB can provide fungibility to handle different packaging processes, which include NCF and both flux-based or fluxless MUF.

Next, let me touch on photonics as it continues to be another exciting area in our AP portfolio. After TCB, our photonics and silicon photonic solutions meet the highest contribution to both AP bookings and revenue for this quarter under the semi segment. Our market-leading photonics solutions continue to have meaningful order wins this quarter, in line with robust demand for 800G optical transceivers for data centers. Our silicon photonic solutions have the best-in-class placement accuracy and this saw order wins for high-end optical communication-related applications. We expect other momentum to continue forward in our photonics solutions.

Lastly, a quick update on our hybrid bonding solution. This quarter, we marked an important milestone, shipping our first hybrid bonding tool to a logic customer, and more shipments are scheduled in the coming quarters. We remain confident of winning more orders for our next-generation hybrid bonds in the quarters ahead.

With those highlights, let me now pass the time over to Katie, who will talk about our group and segment performance.

Thank you, Robin. Good morning and good evening, everyone. Let me take you through the group financials. This covers the group's key financial metrics for the third quarter of 2024.

At the beginning of the quarter, we guided a revenue range of between [ USD 330 million ] and USD 430 million. The group delivered revenue at the high end of this guidance and was flat quarter-on-quarter and down year-on-year. While SEMI registered revenue growth during this quarter, SMT revenue was adversely impacted by the ongoing market softness. SEMI revenue contribution to the group was also higher than SMT in Q3. This demonstrates advantage of our broad-based portfolio as SEMI and SMT sales for the different business cycles and provide some stability at the Grupo.

While group bookings continue to grow year-on-year, it grew slightly quarter-on-quarter as well, mainly driven by SEMI and partially offset by softness in SMT. The group ended the quarter with a backlog of about [ USD 806 million ]. Backlog was down quarter-on-quarter as SMT continued to consume its backlog, while there's growth for SEMI. Group gross margin improved on both quarter-on-quarter and year-on-year basis. Better gross margin, coupled with stable operating expenses provided growth in the operating margin.

Operating margin was 5.3% in the quarter and improved for both quarter-on-quarter and year-on-year. Adjusted net profit for the group was HKD 29.5 million and it was a decrease for both quarter-on-quarter and year-on-year. This decline is mainly driven by a foreign exchange loss of about HKD 108 million. Excluding the ForEx change effects, adjusted net profit would be stable quarter-on-quarter and would have increased by 73% year-on-year. The group continued to have a healthy balance sheet at the end of the third quarter, with cash and bank deposits at HKD 5.47 billion, while bank borrowings were at [ $2.58 billion ].

Next slide, please. In the third quarter, group revenue of USD 428.5 million was flat quarter-on-quarter and down by 3.7% year-on-year as SEMI registered revenue growth while SMT's revenue declined. Group bookings of USD 406.1 million had a small increase quarter-on-quarter. Bookings were up 7.1% year-on-year, driven by SEMI and partially offset by softness in SMT. In Q3, the group's book-to-bill ratio was at 0.9% in and improved on both quarter-on-quarter and year-on-year basis. The group's gross margin was [ 41.0% ] for the quarter. It improved by 94 basis points quarter-on-quarter and by 683 basis points year-on-year. This significant year-on-year margin increase is driven by SEMI.

Next slide. SEMI revenue of USD 229.4 million increased quarter-on-quarter by 0.7%. It contributed a higher proportion to the group's Q3 revenue, at about 53.5%. The IC/Discrete business unit had a quarter-on-quarter increase in revenue, mainly due to mainstream die bonders and wire bonders.

Optoelectronics BUs revenue also increased quarter-on-quarter and was mainly driven by photonics-related applications. Revenue for CIS BU declined quarter-on-quarter with lower revenue from high-end smartphone applications due to seasonality that is front-end loaded in the first half of the year. SEMI bookings of $237.9 million were up by 7.0% quarter-on-quarter and were driven mainly by demand for mainstream wire bonders and die bonders. AP bookings remained robust for the quarter.

The book-to-bill ratio was 1.04 in Q3. It has remained above 1 since Q1 2024. It is also interesting to note that SEMI's quarterly bookings continue to show year-on-year improvement since Q4 2023. The bookings recorded a strong year-on-year growth of 40.1% for this quarter and was mainly due to AP. The segment delivered strong gross margin of 48.6% for the quarter. Gross margin increased by 406 basis points quarter-on-quarter and was mainly due to higher manufacturing utilization driven by TCB production ramp. This ramp is in line with the growing demand for our TCB solutions, as Robin mentioned earlier.

Next slide, please. For the quarter, SMT's revenue was USD 199.2 million, a decline of 7.5% quarter-on-quarter. Bookings were at USD 168.2 million, down by 5.4% quarter-on-quarter. Both revenue and bookings continue to be adversely impacted by the ongoing softness in SMT's overall market. However, our SMT business maintained its leading market position this year. Segment gross margin of 32.3% was down 337 basis points quarter-on-quarter. The decline was mainly due to unfavorable product mix and volume effect.

Next slide, please. As for the announcement made on October 23 last week, the group has proposed to dispose of a stake in its strategic joint venture, Advanced Assembly Materials International Line, or AAMI, to Shenzhen Original Advanced Compounds Co, Ltd., or SOAC. This disposal will be in consideration of new shares to be issued by SOAC and the remaining consideration in cash.

SOAC is a company listed on the Shanghai Stock Exchange. Besides the immediate cash flow that the group would receive on completion, this proposed transaction may potentially create additional value for our shareholders as the group would receive no less than 20% of the shares in SOAC with further potential to grow its value in the semiconductor materials field.

With that, let me now pass back to Robin for next quarter's revenue guidance.

Thank you, Katie. Looking to the near term, we are confident about the prospects of our AP business. For SMT, we believe that its bookings are bottoming out. For SEMI mainstream business, it is on the recovery path, but the recovery remains slower than anticipated. Considering these near-term trends and factoring seasonality effects, the group expects Q4 2024 revenue to be between USD 380 million to USD 460 million. And midpoint of [ USD 120 million ] is down 3.5% year-on-year and 2% quarter-on-quarter.

This concludes our presentation for the third quarter of 2024. Thank you, and we are now ready for Q&A. Let me pass the time to Romil to facilitate. Thank you, Romil.

[Operator Instructions] We're going to have the first round of questions. Can I request Donnie to unmute yourself and ask your questions?

My first question is a housekeeping question. So wondering if Robin can provide some booking outlook for fourth quarter this year, considering you seem to have some new orders from TCBs.

Thank you, and thank you, Donnie, for your question. We expect Q4 group bookings to be flattish too. The way we look at it is that semi will be up Q-on-Q, attributable mainly to but mainstream could be down due to seasonality. Now for the AP, we believe or we expect an increase Q-on-Q and year-on-year bookings for Q4 to be largely driven by TCB memory or HBM memory demand. I mentioned earlier in our announcement also in my opening remarks, we want a bulk order from a leading HBM player for TCB application HBM. However, for SMT, we expect it to come down due to the continued softness of the SMT market and also due to seasonality. But having said that, we believe the SMT bookings bottoming already. I think this is the color to give you for Q4 2024 bookings.

Rob. Very clear. So the second question is just, as you mentioned, you have received a sizable TSB order from a leading memory company. Wonder how sustainable it is into 2025? And do you have a rough idea about whether the volume could be continuously increasing in 2025. And another thing is all related to TB is for leading foundries. So now it's already end of October. So wondering if you have any estimates on when exactly we can win the order or whether we can still get the older but need to split with our competitor.

Let me answer the first question on the HBM first. Now if you look at industry reports and independent research, the AI momentum will continue well into 2025. So the big cloud service providers are making big bets in terms of investment. And from that perspective, we believe the memory space, especially for the HPC space that go into the AI chip will continue to grow. So with this win for us, this [indiscernible] us on the TCB memory market, it's really a breakthrough for ASMPT. It shows that we can really fight and win in this HBM market. They're not the incumbent, right? So it is continued investment in the data center, AI data center. We believe the HBM market looks very interesting in year 2025.

Now for your second question on the -- I suppose you're referring to the chip wafer applications that we are now ongoing, this ongoing evaluation together with a competitor. We're the leading foundry. As far as we know, they are still in evaluation mode. And we don't know when the results are exactly come up, but probably within the year could be the kind of timing.

Thank you, Don. Next, can I request Gokul? Can you unmute yourself and ask your questions?

My first question is again on the PCB side. Congratulations on your HPM win. Could you talk a little bit about what is your estimate of the TCB potential shipment? Is it possible that we can exceed 100 TCB tool shipment next year, this memory win? Secondly, memory versus logic now that we have this HBM model win, when do we expect -- do we expect that memory tools could actually be bigger than logic tool for TCB in 2025. Is that a very likely outcome?

And lastly, I think on this HBM customer win. Is it still the similar kind of TCB tools that you're shipping to logic? Or is there any particular customization that is required for memory given the application seems to be slightly different? And does that have any implication on your market share, given that there is a pretty large incumbent player within this customer?

Okay. Gokul, I'll break your questions into 3 parts and then Robin will answer all 3 of them. For the TCB one, I think you want to know the estimated or rather potential shipments for TCB and especially going in 2025, can it cross the level of, say, 100 tools? Maybe Robin can give some color on this.

Gokul, thanks for the question. We don't comment on winter, we believe, as I said earlier answering the first question, is an exit is an exciting market for us. So we potentially will continue to grow in this market segment. Second question, please.

Okay. So second one is more regarding TCB for memory or HBM versus logic. Will memory tools take over in terms of volume than logic in 2025? Any indication on that?

Yes, definitely. We have been saying that HBM is a bigger market for simply because talking about 12H and the efficient 16H. And the number of HBM in the AI architecture will continue to increase from 4 to 6 and 8 in the future. So the potential for HBM TCB tool, the demand is better than the logic side.

Thanks, Robin. The last one is more from the HBM side in terms of our TCB tools. Are they very similar to logic in terms of what we supplied to HBM or there is some level of differentiation or customization required from memory? And last part is how does it change our market share, given that we have won this bulk order and what is the difference between the present incumbent on the memory side?

Largely, I think our TCB tools are pretty fungible. Probably the only different -- the difference, I would say, is really for memory die it's rather thin, right? So the ability to handle a thin die within the die is the magic that we have in our TCB tool. So there's a difference between memory tool and a logic tool. Logic tool tend to handle also bigger die, for example, for cheaper substrate. We have handled a bigger die to the size of maybe 70 mm by 70 mm, right? So I think there is a little bit of technically different, but largely TCB tools are pretty fungible.

Now in terms of how we look at our TCB tools or HBM going forward, the potential, as I said, we have -- this is a significant breakthrough for ASMPT, right, despite not being the incumbent. So we are confident that our tools are differentiated in a number of ways. We -- our tools can upgrade. We can upgrade quite seamlessly from flux-based to fluxless using plasma technology. And I think our tools are also pretty fungible. It can be used for NCF. It can also be used for flux-based MUF as well as fluxless MBS. I think these are some of the attributes that customers for when they choose ASMPT tools. Next question, please.

Got it. So next question, just staying with TCB. On the, this foundry, leading foundry partnership on chip-to-wafer, Robin, any indications on what is the extent of this migration? Do you think that they basically move completely from flip chip to TCB, flip chip to wafer? Is it for the current 2.5 kind of process itself? Or is it going to be for some future like [indiscernible] related process? Just some idea about when you really expect to start shipping to this customer?

And also on your wafer-to-substrate W2S business that is ramping up right now, is your understanding that everything migrates to TCB for wafer to substrate? Or is it still like -- there is still assumption of the business which is still like using the old flip chip tools and there's some new capacity or new tools, there's new products that are migrating to wafer to substrate for this 2.5D [ W2S ]?

Yes. Thanks, Gokul. Let me answer your first question first on the chip-to-wafer. We Look at it and also in the conversation and discussion with customers, a chip-to-wafer migration from current technology to TCB may take a little bit longer because the current process using MR is rather established. And I think for 2025, the demand for chip-to-wafer tools may not be as big as the chip-to-substrate, but progressively, with the advent of more finer pitch and also a bigger die, so I think inevitably, TCB will be the tool of choice compared to the current MR process.

Now for chip-to-substrate, it's already quite established. We have been shipping the volume shipment chip-to-substrate tools to the leading foundry and OSAT, so clearly, this TCB process has become the tool of choice compared to mass reflow primarily because we have been saying many times before, the reason is this because the chief substrate in the die is huge and the MS has a severe limitation in this aspect. So TCB can handle a bigger die. And also going forward, we believe customer may even have to upgrade to ARR solution as well because of flux-based can put a challenge from [indiscernible] at the substrate level. It can post a challenge of flux-based kind of application. So with our capability to seamlessly upgrade from non-AR to plasma A, I think that put us in a very good position to continue to entrench our position in this space for chip-to-substrate application. Next question, please.

Maybe one follow-up. Is your understanding that, Robin, you are pretty much the sole source for this foundry and OSAT on substream? Or are there competitors in this business as well?

I think the foundry competitors are knocking our here. But as far as we know, we are the sole supply at this point in time.

Thanks, Gokul. Next, can I request [indiscernible] to ask your questions, please.

I have 2 questions. My first is just a follow-up on High Bandwidth Memory TCB. Given that the leading memory high-bandwidth guys probably going to ramp up capacity aggressively to defend its share gains, the second guy in Korea. Is it fair to say that any upside to your orders may only -- will likely come in the next 2 quarters, and they will be premised on the incumbent TCB HANMI under delivering? That's my first question.

Well, difficult question to answer. But we're working hard. Now that we have broken through this particular applications with this leading player, HBM player, we are very hopeful that subsequent orders can flow to ASMPT.

Just trying to understand the dynamics. Did these orders obviously need to install? The production ramp profile takes some time. And once they try our tools, say, I mean, there are some articles that there saying over 32 over the next quarter or so, they all need to try those tools first before they actually get comfortable and order incremental into the second half of next year. So my concern is, I mean, do you see things differently? And by that time, given HANMI has already received your bond orders in the second quarter of this year, the critical inflection point will be over the next 2 quarters. Is that, sorry, the right way to look at it?

Yes. Now that we have one foot in, basically, right? So we're in a much better position before. So this breakthrough gives us the opportunity to capture more opportunities to come from this leading HBM player.

Sure. And then my second question is just regarding the -- HANMI said they are essentially expanding for 400 tools class by the end of December 2025. Do you see any risk of overcapacity in the space given there have been similar expansions by other competitors as well?

As I said earlier, I think HBM tools, HBM, sorry, in the market, has a lot of potential going forward in light of the expansion in terms of AI data center. So on our own, when John, you probably know that we also said we have been expanding our internal capacity. We're actually doubling up in 20 compared to 2024. So we are optimistic, basically.

Got it. Sorry, just one final question on foundry chip-to-wafer. I'm just curious to hear your thoughts on mass adoption at each devices such as smartphone and PC. Obviously, there are no announced road maps, but the TCP throughput is much lower than flip-chip, Obviously, the stats will probably be much lower as well. But -- is this something that's commercial into 2026? Or there's still some things to work out in terms of raising the throughput, getting the use up? Because obviously, there will be a big upside to your to the industry's current estimates on TCB Logic?

Yes. I think AI, using TCB for each devices, probably not so soon, the way we see it. I think technology, we are ready, okay? It's a matter of CEO, right, whether it makes sense to use a TCB package AIG. Maybe a different form of TCB in wafer panel, maybe the solution, but this probably will not happen in 2025. Probably a longer time frame for panel TCB to materialize.

Thanks, [ Juan ]. Next, can I request Kevin? Kevin, can you mute yourself and ask your questions?

My first question is on the next-generation fluxles TCB. I know the progress is still ongoing. I was wondering how confident are we in terms of winning. And just wondering if this could still be a split order for the rest of the year? And also, if I recall correctly, this is a joint development effort with the leading foundry. So just in case we didn't win this quarter perhaps this year or next year, can we still market this fluxless solution to other customers?

Yes, definitely, we can. We're always confident sorry. We're always very confident our TCB solution which ever a few we are talking about, whether it's logic or memory, right? So we'll continue to find, right? The -- I mean the juries are still out there. We will win this bid for the chip-to-substrate application for the leading foundry.

Now I can't really answer you the question whether is it going to be just an eventual one winner or are you going to speed order? I think this question is best directed to our customer. Thank you. And yes, yes, these chip-to-wafer tools certainly can be can be marketed to other customer needing this kind of solution. So there's really no limit there, limiting to just one foundry player.

Kevin, do you have another question?

Okay. Yes, sure. The next one, I would like to ask about the TCB breakthrough at the HBM maker. So just wondering how much -- what's the pace of we're expecting the rev contribution to come in next year? And also, in your opinion, would this breakthrough at this leading HBM maker helping improve the changes penetrating HBM tool for 12H into other HVM makers?

Okay. Sorry, Kevin, I'll just repeat it. Okay. Kevin wants to know what will be the pace of revenue recognition for this bulk order, which we received in October. And then the second part is more on the penetration that how -- how can it be for other, say, TCB potential for other HBM players. Can we actually supply to other players for 12H and higher as well. Kevin, for your first part of the question, revenue contribution, maybe I'll request Katie to give you an update.

Yes. Kevin, this is Katie. So on the revenue contribution, so as Robin was mentioning in my opening remarks as well, we are ramping our production to prepare for the shipment and just to meet the customer requirements. So we'll follow their pace. And as we start the shipment this quarter Q4 and into Q1, our revenue recognition will follow.

Okay. Your next part, in terms of potential chances and engagements with other HBM players, I think Robin can give some color.

Sure. I think with this, the breakthrough win that we mentioned a couple of times this call. It really put us in the radar stream for many other HBM players. I think that will increase potential to serve many other HBM players in the future.

Next, can I request Leping. Leping, can you unmute yourself and ask your questions?

The first question is about the photonic solutions. So we see a lot of new technology in the data centers communication side, like the 1.60 optical module and the chip to -- the CPO solutions. What's your view on the timetable of those -- the commercialization of the technology. And what are the magnitude of this equipment market for the data center photonics versus your existing other -- like the HDM Alcoa solutions? This is the first question.

Yes. I think Photonics, we mentioned a couple of quarters already that we are seeing very healthy demand for photonics tools. Those are mainly for optical transceivers up to 800G at this point in time, some are going from 1.60, but most of them are 800G. So if you look at some of the research houses, 800G will be the main applications going forward for a couple of years. So we are well positioned in that particular area.

Now in terms of CPO, CPO may take a little bit more time because at this moment, it's still at infancy, I would say, right? So -- but we are also -- we're very well positioned in this particular area. Our outfit in micro, which is based on Germany, they are really the #1 player in this space for silicon photonics. So we have the different shape between photonics and silicon photonics. Photonics the take off in terms of demand is already there. But for silicon photonics is still at the infancy stage at this point in time. I hope I answered your question anything.

If you guess what will be the, 2025 or '26, what would be the initial order for the equipment you can see for the silicon photonics?

Yes. Silicon photonics will not be material or significant in 2025, probably 2020 will start to pick up. But Photonics, it's already in high demand. Yes.

Okay. So the equipment size for per unit, what should I say, will it be bigger or will be how the magnitude if silicon photonics pick up?

Well, if you compare silicon is smaller than photonics and photonics will be smaller than TCB related.

The second question, I think, do you have any comment on this KKR news article? So do you have any overall since you already disclosed [indiscernible] what are your overall strategy to maximize the shareholders' value?

Leping, I think there are 2 different things. First you talk about the news which were circulating on privatization. The second thing, you talk about which we -- announcement, which we announced on our joint venture, which is AMI-proposed disposal to it. So do you want a comment on the first one? Or both?

Yes, both. Yes. Of course, first one is more sensitive to share criteria. Yes.

No problem. I think Robin will highlight to you on both.

Yes. I think the first one, I think, we can't comment because we already made an announcement under the [ 3.7 ]. So no further comment on our side. On the AMI, this is -- yes, I mean, we made announcement on the 23rd October that we're going to dispose our stake to a listed company in Shanghai. So this is another step of really unlocking value in this particular investment. So recall the first unlocking of value happened in 2020, when we divested our shares to a strategic joint venture. That was a private enterprise, right? So this step of disposing our shares into a listed company is a further unlocking our value. So we are making our stake in AAMI, our investment more liquid work. So in exchange for AAMI, our shares in AAMI, we're going to have some shares in the listed company and also in return as cash -- immediate cash return. So that full portion, there's an immediate cash consideration as well as further investment in shares of the listed companies. So I think this step is really the step to further unlock the value of this particular business that we have.

Thanks, Leping. Next, can I request Simon to unmute yourself and ask your questions?

Very high-level question first, the ASM International. They really confer the Board and then they also influence your maybe some quarterly annual strategies?

Sorry, are you asking...

Issue with ASM International.

Yes. As our largest shareholder, we do have ASM representing the company at the Board level. They have 2 people sitting on our Board. In terms of influence on company strategy, okay, I will let Robin comment on it.

They are basically, I think as initial capacity when you sit on the Board of ASMPT, right? So we run our own show. We run our own strategy. So there's no interconnection with AAMI per se. So when the 2 board directors sit on onboard, they are acting in their [indiscernible] duty as ASMPT Board of Directors.

Yes. So if they say they are maybe [ 25 ] take to some investors, Board Members have not influenced their peers, right? ASM International has own decision, right?

You are technically correct. And I think this kind of question is better if you direct to ASM rather than us because we can't give you an answer on their behalf.

Okay. Yes. And then back to the ATVM question, yes, congratulations on the new orders. But the CFO mentioned the revenue recognition may start from December quarter, but although order just took the place in October. So question is, once you get the new order, how many months needed to recognize the revenue? And also, what do you mean the definition of the order that should be the immediate delivery or maybe one year time horizon order and then the delivery is kind of the quarterly basis. Could you share some color for the definition of the order and impact on the revenue and the leading time to deliver the equipment.

Okay. So I think Katie will give you some more comments.

Simon, this is Katie. So this order is rather unique due to the size of it that way. As we're working with our customers really closely in prior quarters, the management had made a decision to risk build for in preparation for this order because of the urgency of the delivery. That's why you probably heard me talking in Q3 margin, right, there was even the ramp of production for this order. So we have started production for this order a while back. Okay?

Now in terms of the -- in general, the lead time for AP production is much longer than our mainstream products, somewhere around 6 months plus. So that's the typical lead time. But this time, because the size of the order and the significance of it and the fact -- due to the urgency of the requirement, we have started the production way earlier. Okay?

Yes, Katie. So you are saying normally, the delivery time from, let's say, today in the order, right? And then you have to deliver the equipment. The typical TCV is about 6 months, but the HBM TCV longer than 6 months, you are saying?

No, no, sorry. That's not what I said. Sorry, if I was not very clear. The -- actually, Robin mentioned about -- whether it's the HBM TCB or logic TCB right? From a production perspective, the lead time for these machines is actually very similar.

How many months?

Six to 9 months.

Six to 9 months. So usually, you get the order now and then to recognize your revenue, we have to wait 6 to 9 months.

No, we are talking about production lead time, not recognition of revenue. But those 2 different things here. So production lead time is about 2 months.

Yes. Okay. And lastly, another one is 12H and beyond. So if the team makers purchase your great TCB or HBM, of course, they can use 12H, but that means versus the 16H and then even the 24, there is any limit because after the TCB for HBM, the our next step is obviously the hybrid bonding equipment. So the question is what could be the cap number of the TDM die taking 12 or 16 as a maximum. After that, you will market a hybrid bonding equipment to the customers is the question.

Yes. Now the top of tool can be used, whether it's the TCB, it all depends on the high -- right now, sometime back a few quarters back, they have relaxed the hike from 720 to [ 725-micron ]. So that enabled a TCV that you use for a longer period of time, right? So from 12 -- right now, TCBs definitely 12 going forward 16 with problem because in the lab, we have done 16 high stacking already in our own lab. So we are capable of 16 high. So it all depends on the height requirement. If they continue to relax the height of TCV content, there's no limit. So the limitation of TCV is actually limited by the height, right? So we can continue to stack. As long as our height requirement is relaxed, we can continue to stack 16, 20 and beyond. But if there's a limitation to the height, then spot hybrid bonding may be the choice of tool if there's a limit to the height.

Next, can I request [ Tony ]?

So my question is regarding to the financial reporting, we have being impacted by the FX in this quarter. So can you give more details? Like what kind of FX movement will cause -- record this kind of a loss? And how the situation in the fourth quarter if the direction move backwards, will be reversed or record some kind of gain from the FX movement? This is my first question.

[ Tony ], it's Katie. Let me try to answer this question. So the -- this is not our first time actually FX has impacted us. But this time, the impact is rather large, especially considering the relatively low profit level. So as a global company, we operate in many different countries and have currency exposures in the major ones you can think of: RMB, Malaysia ringgit; Singapore dollar; yen; euro; et cetera, right? And overall, though, generally speaking, for the exposures in these countries is a net exposure in U.S. dollars. So in Q3, U.S. dollars actually depreciated against this basket of currencies. Therefore, we had a hit on the P&L. To your question, if U.S. dollar strengthens, then we should be expecting a certain level of FX -- positive FX impact.

Okay. So we have this kind of a hedge impact I think because our -- the full FX intact action level is the other direction, right? This means if U.S. dollar appreciation is appreciation or negatively impact our gross margin or operating margin?

So you're correct, actually. So as I was speaking about the exposure, these are the entities that we do not hedge. We do have certain businesses in certain regions that have hedged. So that would behave differently to what I explained earlier.

Next, can I request Arthur? Arthur, you can unmute yourself and ask your questions.

Congrats on the good SEMI margin. 49% is a good surprise to me. I said that investors actually underappreciate this 4% Q-on-Q increase. So my question is, is this because of the mix effect or because the -- is it sustainable? The reason I asked this question is because if we read through those front end and even your Europe peers, 50% margin looks quite common. So I don't know it's because you enter in a new market, so we should expect this margin for the future? That's my first question.

Thanks, Arthur. So let me maybe comment on the short-term margin quick, and I'll give you a little bit of color on the long term. So for Q3, as we mentioned in the opening remarks, we were ramping TCB in preparation for the bulk order. Therefore, our margin was favorably impacted due to very high absorption for certain parts of the plans. Actually, capacity was -- the utilization was running almost at 100%. So as we go forward on the SEMI side, we do expect that the AP mix will continue to be quite strong. However, I do want the usual caution out there to the group here that we have always communicated right, the margin, whether it's for SEMI or SMT could be impacted by volume, product mix or a combination of the 2 in certain given quarters, especially and also as we go forward. As AP content grows, you probably noticed that the long AP could be -- AP revenue could be quite lumpy, right? So the lumpiness of revenue from quarter-to-quarter may actually bring even more fluctuations to gross margin quarterly. And also for SEMI, if certain recoveries happens in the mainstream products, which have relatively low margin, especially for the CCC markets, right, that could impact the margin as well. So having said that, I guess, we do aspire actually to improve gross margin over the longer term, especially with the tailwind from us pivoting to AP markets.

Yes, the second question, I think, probably is a little bit out of box thinking. So do you and do your clients worry about the component or capacity shortage in '25, especially in TCB? The reason I asked this question is because I also cover a lot of Taiwanese and also TSMC. I sense that a lot of the vendor they actually worry about these issues, especially for the COAS supply chain SPE. So can you share with us your high-level thinking about this issue?

Yes, Arthur, no particular call-out. Actually, the way to mitigate some of this is to do risk buy, right? So when we do a risk buy, we secure the components supply chain. That's why we are able to [indiscernible] the TCB HBM in short time because we do a risk buy. So I think that's probably the medication, the stuff that we can take to avoid all this supply chain issue.

And Robin, can I follow up one thing? When -- what time you start to risk buy.

Depends. We have to judge, I mean the commitment from the customer. So once we have a solid commitment, we can do a risk buy. Yes.

Got you. Got you. So if based on the previous discussion and based on the HPM, your Korean clients come, they do expect the 12 high to reach the crossover before mid of '25. And then the lead time for the HBM bundler lead time is 6 months. So we pretty much can forecast where very strong ramp-up in the upcoming quarter, right?

I hope life is as simple as you speak. Not exactly. All depends on customer to customer, okay?

Thanks, Arthur. Next, can I request Gokul? Gokul, you can ask your follow-up questions.

Yes. So just shifting the focus to the traditional packaging side. Robin, any extra color that you can add in terms of what your customers are telling you, both in China. I think where I think last quarter, you were a little bit more positive in terms of how the momentum seems to be as well as outside of China? And lastly, I think for SMT, I think you mentioned that the orders, the bookings are kind of starting to bottom out. Anything that you're hearing from your customers on their own utilization rates and when we could start to see any pickup on the SMT side of the business.

Thanks, Gokul. We are hopeful that with the various stimulus package, we're hopeful that our Chinese customer base will start to feel a bit more confident. So I have nothing more to share at this point. Hopefully, we can share a little bit more good news when we meet again in the Q4 conference call.

Now yes, SMT bookings have been coming down for a number of quarters. We feel that it's really a bottoming out. Also looking back for the history of SMT, so the booking level has reached a very low level. And in terms of feedback on customer utilization rate, obviously, I think for automotive, industrial, still sluggish. We see a little bit movement in the consumer space for SMT as well as also for the SEMI side. I think this is some of the color that we give to you right now, Gokul. Yes.

Okay. And on the silicon photonics side, just following up on a previous question. Is that also something that's going to have significant TCB adoption? Or is it going to be some different kind of tools that you will be selling given a lot of the co-packaged optics and linear drive, et cetera, will also be some kind of 2.5D packaging eventually.

Yes, pretty similar, Gokul, pretty similar. We have a tool from a German outfit [ Amitra ]. Very precise. But speed-wise, it's not as fast as TCB. Of course, TCB is probably more expensive. So it all depends on customer, right? At the end of the day, customer preference. Do they want to use an [ Amitra ] tool for CPO or they want to use a TCB tool for CPO? So all depending on customers' preference and their technology.

Thanks, Gokul. We will just take one last round of questions. Simon, can you mute yourself and ask a follow-up question?

Just a question to Katie, the CFO. Any update shareholder return policy? Because these days are you guys EPS number is so low, but if we apply historical average of the payout ratio, the dividend amount seems to be similar. So would you update your overall target, the payout ratio and dividend? And then your CapEx burden, because you have to expand the newly growing TCV demand, you have to expand the capacity, right, for HBM, et cetera, so your CapEx burden and with current cash amount versus total debt. Yes. So financial question, overall, number one. Yes.

Simon, so let me answer the CapEx question first. So the group's CapEx rounds about HKD 400 million, HKD 600 million a year. And for this year, we definitely have invested in certain clean rooms, especially, again, in preparation for the AP ramp and that we're quite comfortable with that. But in terms of your question on the dividend policy, at this point, we do not have any new things to share.

Okay. Fair enough. And then the last question to CEO Robin. I'm feeling that the order ATBM related to CapEx spend among the mama already picking up this year because as you know, ATBM manufacturing cycle time usually half a year and then a lot of new orders is an annual basis. So I wonder why you think the TCB accumulated order for HBM will grow further. I'm feeling that already the growth is hitting the record to high level for now. And then next year, maybe growth could be lower. And then we have to wait the new equipment order or hybrid bonding. But today, you found that the TCB for ATBM, the new order will grow further and further through the next year. But for me, it may already hit the peak level for now.

Simon, 12 high will continue to grow. If you look at those industry forecasts, right, is 12 high will overtake 8 high. So 12 high will continue to grow and 12 high need TCB. 12 high cannot use the traditional master flow technology. So that's why we are confident that the TCB for HBM will continue to grow 2025 and beyond.

I see. And then the even 16 high, right? That's the great capacity.

After 12 high will be 16 high. Yes. So all these are already probably in the customer road map already. Yes.

So you think existing TCB for 8 high cannot be used easily for 12 high. They don't need a new one.

12 high and above for new TCB. Very simple.

So eventually, the long-term growth opportunity for hybrid bonding for HBM, do you expect a meaningful order maybe next year? Or we have to wait until 2026?

Continue. Continue. So we have the first batch in October. So we believe the 12 high TCB demand will continue into 2025.

No. I mean the hybrid bonding.

Hybrid bonding would not be so soon, sorry. Hybrid bonding, as I said earlier, so it all depends on the hybrid requirements. So far, TCB is still able to do it, whether it's 16 high or beyond. But if they don't relax them, probably hybrid bonding will be the solution.

Thanks, Simon. And with that, we will conclude our Q&A session. Before we end, let me request Robin to say a few words, and then we'll conclude the call. Thank you.

Thank you. So thank you all for your questions, and allow me to quickly highlight the key takeaway from today's call.

We are enjoying strong TCV momentum, which is intensifying, particularly for HBM. The bulk TCV order from a major HBM player is a solid endorsement of our ability to win in this highly competitive HBM market. We are the only supplier of substrate TCB in the leading foundry supply chain, and we'll continue to win orders there.

Orders also continue from a leading IDM customer for TCB, chip-to-wafer. Thus, we strongly believe that the inflection point for TCB demand is here. While SMTs would keep bottoming up, SEMI's quarterly bookings have been increasing year-on-year since Q4 2023, which is a positive recovery trend. Its book-to-bill ratio has also remained over [ 1 ] for this year.

With that, I conclude our call. Thank you very much.