Asmpt Ltd

Good morning and good evening, ladies and gentlemen. This is Romil from Investor Relationside, and I will be the moderator for today's call. On behalf of ASMPT Limited, let me welcome all of you to the group's fourth quarter 2023 investor conference call. I would like to thank you all for your interest and continued support 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 only after the management has gone through the entire presentation. During the Q&A session, priority will be given to the covering analysts. Let me go through the 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 the 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 this result can be downloaded from our website. On today's call, we have... our Group Chief Executive Officer, Robin, and the Group Chief Financial Officer, KT. Robin will cover the Group's key highlights, outlook, and first quarter guidance, while KT will provide details on the financial performance. And then we will open the floor for Q&A. So with that, let me hand the time over to Robin now.

Thank you, Rong. Good morning and good evening, everyone. It is a pleasure to have you all on an earnings conference call for the fourth quarter and full year of 2023. Before we go to the details of our performance, let me take this opportunity to give some highlights on developments in the semiconductor industry and the overall macro environment. We all know that 2023 was a challenging year in semiconductor industry. I will even add that it was one of the toughest years in our recent history. Overall, the macroeconomic environment was characterized by, one, persistent inflationary pressure, two, a high interest rate regime, three, escalating geopolitical tensions, and lastly, a slower than anticipated recovery of the Chinese economy. These collectively weakened overall consumer sentiment and electronics demand by a considerable degree in 2023. However, amidst the gloom, our unique broad-based portfolio provided a certain level of resilience. The prolonged semiconductor downturn adversely impacted our semi-business, with significant revenue contraction in 2023 as demand for PCs, smartphones, and other consumer electronic devices dipped significantly. However, our SMT business remained resilient, mainly due to continued demand from automotive and industrial end markets. And SMT delivered higher revenue than Semi for a sixth consecutive quarter in the fourth quarter of 2023. Another advantage of a unique broad-based portfolio is the diversified end markets that we serve. From an end market perspective, Automotive and industrial application demand remain robust and continue to contribute the most to the Group's 2023 revenue, even as demand from the consumer electronics and market remain constrained. Let me now share more about what we see as the most exciting part of our unique broad-based portfolio and the one with the highest growth potential, our advanced packaging solutions. We believe that we have the most comprehensive suite of advanced packaging or AP solutions in the industry. Being deeply embedded in the supply chain of major AI and HPC customers, we are experiencing growing demand from generative AI and HPC applications as these require more advanced tools. So, despite a down-sider year, The relative percentage contribution of our AP solutions increased year-on-year to about 22% of the Group's 2023 revenue of approximately $410 million. Looking ahead, we are very excited about the prospects of our AP business. We estimate that the addressable market for AP will progressively increase from about $1.7 billion in 2024 to approximately $3.3 billion in 2028, at a compounded annual growth rate of about 18%. This addressable market and CAGR estimates have increased since our last update a year ago, as we factored in prospects from the fast-growing global generative AI market. We are seeing more of our AP solutions gaining traction and further solidifying our entrenched system with major AI players. This gives us the confidence that our AP market share will grow. In the next couple of slides, let me highlight some significant developments within our AP solutions. You may recall the picture on the left of the slide. This is an example of a high-end generative AI or HPC device using 2.5D packaging. These devices are getting more complex as they incorporate an increasing number of chipless and more advanced high-bandwidth memory, or what we call HBM stacking. The key message here is that we have a range of high-precision bonding solutions within our portfolio, that are capable of handling the different interconnect requirements for such intricate devices, most notably our TCB, but also flip chip, hybrid bonding, and SMT placement tools. Let me comment on each of these. We will begin with TCB, or thermal compression bonding. We strongly believe that the majority of the complex interconnects of the future can be handled by TCB. The reason is simple. TCB solutions are in an optimal position, occupying the nexus between favorable total cost of ownership and the technical ability to handle increasingly demanding bank pitch and placement accuracy requirements. Our TCB solution delivered their highest yearly revenue in 2023 and contributed the most to ASMPT's overall AP revenue for the year. Let me shed light on our engagements in the logic IDM space. We have a solid foundation serving this market and our solutions have a commanding position in both chip to substrate and chip to wafer applications for HPC and AI. Looking at logic requirements required by generative AI demand, We want meaningful TCB orders for chip-to-substrate applications from a leading foundry in the third and fourth quarters of 2023 and continue to engage deeply with this customer for a next-generation ultra-5-page chip-to-wafer TCB solution. In addition, we also want orders from OSACs for both chip-to-substrate and chip-to-wafer applications as these OSACs expand capacity to support growing AI demand. Now let's look at the HBM market. Here, our TCP tools are already in production as a leading HBM player. And we continue to support meaningful engagement with multiple HBM players. Based on this ongoing engagement, we are confident of more order flow over the next few quarters. Let me highlight that as HBM packaging requirements become more demanding, it will increasingly require TCB processors. And this is where our technology leadership and expertise come in. As we are ready for both 12 and 16 high HBM with our next generation ultra-fine pitch TCB solution. To summarize our TCB story, We are a market leader in TCB with the largest in-stock base of tools across the globe. As the first mover in TCB, we accumulated extensive industry learning over the past decade and have come a long way expanding our customer base beyond logic IDMs into HVM, Foundry, and OSETs. I'm confident that TCB will continue to have tremendous growth potential and as its adoption accelerates, we are in the best position to capitalize on the generative AI boom. ASMPT is able to offer the most comprehensive and scalable TCB solutions, and we will continue to expand our TCB production capacity. Let me now touch on our flip-chip solutions. DIV has also gained traction from generative AI and HPC application. which require varying degrees of pitch and placement accuracy. There was consistent order momentum for flip chip tools throughout 2023, which is expected to continue into 2024, as we engage leading foundry, HBM, and OSAC customers for both chip to substrate and chip to wafer applications. Recently, there has been a lot of talk in the market about the potential in AIH devices. Let me add here that our Flipchart tools are capable of panel-level pick-and-place fan-out applications with lower form factors, which are very well suited for such AI edge devices. Taken together, both our TCB and Flipchart tools are already catering to AI players at the cloud and data center level. And ASMPT is in a good position to take advantage of the huge potential in AI edge devices when the demand surge happens here. Now, let me comment on hybrid bonding. We have a breakthrough year for hybrid bonding solutions, winning orders for two tools for 3D integration, which will be delivered in the second half of 2024. We are confident of securing more orders for hybrid bonders in the first quarter of 2024 and beyond as we continue to engage key customers in various air market applications for our next-generation hybrid bonding solution. And we are confident on intercepting the high-volume manufacturing ramp in time to come. In the device diagram on this slide, you can see that we also have SMT placement tools that help place integrated passive devices on the substrate. In summary, this slide emphasizes the wide range of interconnect solution ASMPT has available to handle the complex requirements of generative AI and HPC devices. Let me cover our photonics solutions next. On the left, of this slide, we show how packages are evolving with increasing bandwidth requirements and where our solutions are relevant for die placement and lens attached requirements. With the generative AI boom, there's an ever increasing need for higher bandwidth and data centers are expanding and upgrading to support this. This expansion is in turn fueling demand for higher bandwidth optical transceivers and co-packaged optics or CBO applications. For transceivers, we have market leading range of photonic solutions capable of handling bandwidth from 100G to 800G and beyond. In particular, we have a comprehensive range of solution for 400G and higher bandwidth transceivers and we command the dominant market share in the transceivers market. For CPO applications, our silicon photonic solutions have the best in-class placement accuracy and highly flexible system capable of handling multiple bonding processes. With our technology leadership coupled with demand growth, we won't repeat orders from leading AI players in 2023 and expect this order momentum to continue into 2024. I hope this couple of slides on our AP capabilities and footprint provide a clear message of ASMBT's strong and comprehensive technology position in the various types of AP solutions and their overall competitiveness. These are gaining traction in the face of strong demand growth from generative AI. Our automotive and market applications continued to deliver a robust contribution to the group, providing the highest proportion of group revenue for two consecutive years, at about 22% of ASMPT's 2023 revenue, or approximately $410 million. This was primarily fueled by growth in engagement with automotive players, and particularly electric vehicle or EV players that enable more of our solutions becoming the process of record for these companies. We witness demand momentum in the growing EV market, which was supported by the entry of new automakers and the launch of more EV models. An area of keen interest for us is silicon carbide application. As demand for silicon carbide related application rises, ASMPT has a complete range of solutions to serve this market, including laser dicing for wafers, die attach, pressure sintering, molding, and SMT placement. These value-added solutions have helped us to become the preferred co-development partner for our growing base of customers. Looking at an addressable market for automotive and market applications, we expect it to grow from approximately 1.8 billion U.S. dollars in 2024 to 2.6 billion US dollars in 2028, a CAGR of about 10%. As I mentioned earlier, our SMT business provided a certain resilience to our overall performance during the prolonged semiconductor downside primarily powered by automotive and industrial and market application, SMT delivered a relatively buoyant revenue performance in 2023 and strengthened its position as market leader. SMT performance was in large part due to robust demand for its high-end placement and printing tools, which came mostly from Europe and the Americas. SMT contributed higher revenue than Semi in 2023 and its revenue has staked over $1 billion for three consecutive years. In the second half of 2023, as automotive and industrial air markets started normalizing, SMT bookings softened. However, I must emphasize that our SMT business has exposure to various air market applications. And so even though automotive and industrial air markets were normalizing, there was also growing demand for SMT tools from AI-related server applications. In fact, we received orders for SMT tools from AI server customers and a leading founding player in 2023, as these tools have flexibility in handling varied board sizes with high placement accuracy. More recently, there was also demand for smartphone wearable applications, particularly for SMT system in package or SIP tools. For SMT's AP solution, we are also scaling up and engaging customer with our next generation of tools that have higher placement accuracy, multi-dye picking capabilities, and the ability to pick dyes directly from wafer for better performance. These tools are gaining traction across SIP, wafer-level fan-out, and embedded substrate applications, and we expect more orders for these AP tools in 2024. 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. This slide covers the group key financial metrics for full year 2023. As Robin has highlighted, the prolonged semiconductor downside echo negatively impacted our semi-business, with revenue declining 37% year-on-year to $812.9 million U.S., while SMT experienced milder impact, declining 10% year on year to 1.06 billion US dollars. Taken together, group revenue performance was 1.88 billion US dollars, a decline of 24.1% year on year, with SMT contributing about 57% of group revenue. You can see how our broad-based portfolio provides some advantage as our two segments follow different business cycles. For bookings, Semi's year-on-year decline was steeper than SMTs, mainly due to the ongoing semiconductor down cycle. SMT bookings were impacted mainly in the second half of 2023, as automotive and industrial end markets began normalizing. Our backlog declined to $846.1 million at the end of the year, compared with $1.15 billion a year back. Group gross margin was down year-on-year by 186 basis points to 39.3%. Our operating margin declined by 920 basis points year-on-year to 7.5% due to lower sales volume and a reduced gross margin. Similarly, group adjusted net profit declined by 71.5% year-on-year to $744.9 million. Adjusted earnings per share was 1.82 Hong Kong dollars, a decrease of 71.4% year on year. We had a healthy balance sheet at the end of 2023 with strong cash and bank deposits of 4.8 billion Hong Kong dollars. Our net cash was also at an all time high of 2.8 billion Hong Kong dollars. Before I go further, let me quickly touch on our strategic investments in 2024 and beyond. Despite the downturn, we'll continue to prioritize investments in R&D and infrastructure. As Robin highlighted, we firmly believe that AP is a strategic growth area with significant upside potential, and we're prioritizing R&D resources and capacity investments to further strengthen our leading position. In addition, we embarked on large-scale system rollouts that are focused on people development, IT, ERP, and other operational areas, and these efforts will intensify. Once completed, our infrastructure investments will make ASMPT even more productive and competitive. These investments are expected to incur additional operating expenditure of about HK$250 million in 2024, and this incremental expenditure will be on top of our relatively stable OPEX over the last two years. At the same time, We continue to remain mindful of costs and will continue our cost control and efficiency enhancing initiatives. For 2023, group revenue of 1.88 billion US dollars was a decline of 24.1% year-on-year. Both segments had declines, but it was much steeper for semi, in line with the prolonged semiconductor down cycle. Group bookings for 2023 declined by 33.5% year-on-year to $1.57 billion. For end markets, combined bookings from AP, automotive, and industrial remained stable year-on-year at about 60% of group bookings for 2023. Group gross margin for 2023 was 39.3%, declining 186 basis points mainly due to semi, whose gross margin declined 375 basis points year-on-year to 40.9%. This decline was partially offset by SMT's gross margin growth due to a favorable product mix, improving 73 basis points year-on-year to 38.1%. Operating margin declined 920 basis points year-on-year to 7.5% due to lower sales volume and loss margin. In the fourth quarter of 2023, group revenue of $435.4 million was marginally higher than the midpoint of revenue guidance, a decline of just 2% sequentially. Year-on-year decline was 21.4%, in line with prevailing industry weakness. Group bookings of $349.8 million declined 7.6% quarter-on-quarter due to seasonality. Year-on-year decline was 12.2%, mainly due to SMT, as its automotive and industrial end markets began normalizing in the second half of 2023. Group gross margin increased by 812 basis points sequentially to 42.3% due to gross margin for the previous quarter being exceptionally low. And from better product mix for both SEMI and SMT, gross margin improved by 87 basis points year-on-year, mainly driven by SMT. Group operating margin improved by 356 basis points sequentially to 5.5%. operating margin declined by 825 basis points year-on-year, mainly due to lower sales volume and a gross profit. For fourth quarter 2023, Semi registered a small increase in revenue of 1.2% quarter-on-quarter to $203.9 million, as revenue was down 15.3% year-on-year in line with industry weakness. The IC discrete business unit had a stable revenue quarter on quarter, with the highest revenue contribution from TCB. The optoelectronics business unit had a small amount of business growth sequentially. The business unit's advanced tools serving photonics applications grew in revenue quarter on quarter, and it contributed the most to its revenue. The CIS business unit's revenue continued to be adversely impacted by ongoing weakness in the global smartphone market. For fourth quarter, Semi recorded bookings of $158.9 million, down 6.2% quarter-on-quarter, mainly due to seasonality. However, please note that Semi's quarterly booking in 2023 declined at a slower rate year-on-year for the first three quarters and turned positive in the fourth quarter. Semi's bookings increased by 10.5% year-on-year in the fourth quarter, mainly from the growth in AP bookings. Semi's gross margin in fourth quarter was 43.8%, mainly due to a favorable product mix, as AP and automotive contributed the most of segment revenue. It improved significantly quarter on quarter due to low Q3 margin. Gross margin was down marginally by 66 basis points year on year. Semi's profit was 0.9 million Hong Kong dollars in fourth quarter. Our SMT segment continued to deliver higher revenue than STEMI for a sixth consecutive quarter in Q4, 2023. Fourth quarter revenue was $231.5 million, a decline of 4.7% sequentially and a 26% year on year. SMT's industrial and automotive end markets combined still contribute the most to segment revenue. As automotive industrial end markets normalized in the second half, SMT's bookings declined in the fourth quarter by 8.7% quarter-on-quarter and a 25% year-on-year to $190.9 million. SMT's gross margin was strong at 41% in the fourth quarter, an increase of 493 basis points quarter-on-quarter and 188 basis points year-on-year. SMT's gross margin improvement was mainly due to a favorable product mix. SMT's profit was 266.6 million Hong Kong dollars in the fourth quarter, an increase of 3.3% sequentially, but down by 49.4% year on year, mainly due to lower sales volume. This slide highlights the ASMPT's management's best estimates of revenue breakdown by end market applications for 2023 compared with 2022. These end markets highlight the extent of a broad-based portfolio and the our exposure to diverse ad market applications. Automotive remained in pole position and had the highest contribution to group revenue for two consecutive years, providing approximately 22% of 2023 group revenue. Even though automotive witnessed some softness, our comprehensive range of automotive solutions and engagements across a growing base of customers helped this ad market maintain its contribution levels. Next highest contribution was from industrial. This market also witnessed some softness, but its percentage contribution remained stable, providing about 16% group revenue. It was also the highest contributor to SMT's performance, benefiting from structural trends towards intelligent factories, greener infrastructure, enhanced automation, and digitization. the consumer communication and computing or triple C markets continue to experience softness due to weak consumer sentiment. The others category includes revenue from spares, services and other applications that cannot be meaningfully identified. And this revenue has remained stable year on year. This slide shows the yearly revenue contribution by different geographies year on year. Looking at 2023, China, including Hong Kong, continued to see a year-on-year revenue decline as its share of group revenue dropped from 42% to 31%, partially offset by year-on-year revenue growth from Europe and America. Europe's share of group revenue increased from 18% to 28%, and America's increased from 12% to 18%. Our diverse customer base is spread across the globe and includes IDMs, OSETs, Fabulous, foundries, high-density substrate manufacturers, memory players, EMS, and others. This helps us maintain a low level of customer concentration risk. For 2023, our top five customers accounted for approximately 17% of group revenue. We remain fully committed to enhancing shareholder value and returning to shareholders. We have an existing dividend policy to maintain dividend payouts at about 50% of group profit on an annual basis. For 2023, the board has declared a dividend per share of 87 Hong Kong cents in line with this policy. In addition, the board has declared a special dividend of 52 Hong Kong cents per share on top of the 50% dividend payout. With this special dividend, the total dividend for 2023 is 1.39 Hong Kong dollars per share, a payout of 80%. Let me now pass the time back to Robin for first quarter 2024 revenue guidance.

Thank you, Katie. The group expects first quarter revenue to be between 370 million to 430 million US dollars. At midpoint of 400 million US dollars, This represents a decline of about 20% year-on-year and 8.1% quarter-on-quarter. This decline is mainly due to lower revenue from SMT as its bookings began softening in the second half of 2023. Many industry experts have anticipated the semiconductor industry to recover in 2024, which could in turn fuel the next multi-year industry up cycle and the group remains optimistic about its prospects of optimism is further supported by long-term structure trends from automotive electrification smart factories green infrastructure 5g 6g iot and ai growth across cloud data center and ai devices on the broader level These structural trends are also moving in tandem with increased KPEC spend from nations securing their supply chains via more onshoring and organizations preparing themselves to deal with more dynamic global supply chains. This concludes our fourth quarter and the full year 2023 presentation. Thank you, and we are now ready for Q&A. Let me pass the time to Ron to facilitate.

Thank you, Robin. For asking questions, please either use the raise hand function or you can type your questions in the chat to SMPT Q&A. Please ask your questions one by one and limit them to maximum of two questions at each turn. Thank you. With that, can I first request Gokul to unmute yourself and ask your questions?

Yeah, thanks Pramil and hello Robin and Kiri. My first question is on TCB. Could you talk a little bit about your market share in IDM for TCB? Because there is some commentary in the market that you will be losing share from one of your competitors. So just wanted to understand your market position in your leading IDM customer. And secondly, congratulations on this wins with the leading foundry for chip on substrate and potential engagement on chip on wafer. how do you see this Foundry and OSAT TCB opportunity? Do you think it is going to become bigger than your IDM exposure in the next two to three years in terms of the number of tools that you could ship? And I think previously you had given us a guidance on doubling the number of tools or at least reaching the same number of tools that you shipped between 2012 and 2021, if I remember right. in the next two to three years. Given that you raised the advanced packaging growth guidance, is there any update to this tool shipment forecast for PCB?

Okay, I think, Gokul, let me break it down into a couple of sections. First, let me request Robin to comment on our TCB, particularly in relation to the market share when dealing with the IDM customer base and whether are we losing market share and what is the potential there?

Thanks, Rob. Thanks, Gokul, for the question. Now, I think we have been talking about this logic IDM space for many years. We started with this idea this business more than probably around 10 years ago. So it started with a chip to sub-chip applications. So we were the first mover in this particular space for many years. And then when these Logic customers used our tools for chip to wafer, recall we won a 100 million US dollars order in 2022. So we started to ship this chip to wafer tools in 2023, based on this $100 million that we got in 2022. Now, as far as we know, for chip to wafer application with this logic customers, we are probably the only one. And for cheap to substrate, we are aware that there is a second source for this particular customer. Next question, please.

Okay, so next one is basically on the foundry events. where our TCB has won orders for the chip to substrate part. And there's some potential for chip to wafer with the leading Foundry player as well. So can Robin, can you comment on this potential with the Foundry and how it is also sort of moving towards the OSATs or spilling over to the OSATs? And how does it compare to the IDMs?

We are obviously very excited. to be able to participate, you know, in this business we are leading foundry for cheap to substrate applications. Now, we called out that we want an order, meaningful order in 2023 for cheap to substrate application for generative AI 2.5G packaging kind of requirement. And we also alluded in 2023 that we continue to win orders in Q423, and we continue to win the order. So we back the order as well in Q423. So for two quarters, we have won meaningful orders from this Foundry customer for cheaper soft trade applications. Now, we believe that going forward, As the demand for generative AI chip continues to increase, there will be opportunities for TCB to be used for chip-to-wafer applications as well. Currently, for chip-to-wafer, the industry process is still using a 50-mastery flow. But fluid-chip mass reflow, in our opinion, has limitations as the demand for final pitch and final placement accuracy for chipless integration at the chip-to-wafer level increases. So TCP is the best place to serve this demanding technology and process requirement as pitch and placement accuracy continue to be tighter and tighter going forward. So we believe our current generation of tools, as well as the next generation of ultra-fine-page TCB tools, are very well placed to capture the opportunity for chip-to-wafer application for 2.5D packaging with this particular leading foundry.

So last part of the question is, I will acknowledge that we have previously indicated that the demand for our TCB, which was in the last 10 years till 2021, will sort of double from 2022 to 2024. So Google wants to know whether is this still valid or is there some upside since now we have actually, you know, increased our addressable market and Kegel growth. So can Robin comment on that and give some sort of guidance on the AP growth? Sure.

i think the recent orders from uh you know the chip to substrate orders from leading foundry as well as uh our growing base of tcp uh business with uh oses as well uh i think this this sort of validated our okay uh thanks uh thanks very much uh robin uh my next question uh is on hybrid warning uh could you talk a little bit about what is the progress on hybrid bonding

uh, we do hear that you are engaging with some of the boundaries for some micro bump based, um, uh, hybrid bonding tools. Uh, could you compare your hybrid bonding progress compared to the market leader? Uh, and, uh, is there a difference in terms of where your hybrid bonding tools are being applied, uh, compared to, uh, the traditional, uh, use case for hybrid bonding, which has been largely like logic to logic, uh, direct, uh, interconnect and, uh, Any further commentary in terms of beyond these two tools that you've already shipped last year? Does it go to 8 to 10 tools in this year or is it going to be a slower progress?

I will highlight here that especially for hybrid bonding, I think due to confidentiality, we are not really able to give more details in terms of customer and of course, going forward. We can't give an exact indication of the number of tools which is in the, you know, for our potential. But I think, let me summarize your question, then Robin can give more color. First thing, I think Robin can give you some comments on the progress of our work hybrid bonding tool. Secondly, engagement with, you talked about, is there any engagement with the foundry or leading foundry customers? So any comment on that? And lastly, how do we sort of compare against peers in the market in terms of our own tool and application? So maybe Robin can just give an overview on this.

Yeah. Now we had in the ND&A and also in the presentation, we did mention that 2023 was a breakthrough year For us, in terms of hybrid bonding, we won two orders, two tools, actually, in 2023. And these two tools are due for shipment most likely in the second half of 2024. Now, we are confident that we will continue to win more orders for Gen 1. We're still talking about Gen 1 HB2s. So, Gen 1, we're confident that we'll continue to win orders for Gen 1 in Q1 2024 and beyond. Now, Goku, as what Rom said, because of client confidentiality reason and also competition reason, we do not want to comment on the clientele that we are serving for HP. uh now for um you mentioned about micro bomb now that could be a little bit of uh uh i you know view that you know for for hp the for hp there's actually no no micro bomb when you talk about micro bomb typically you you the street is referring to or the industry is referring to tcp TCB has microbiome. So as I said earlier, when I answered your question on engagement with a leading foundry, as I said, we are engaging this leading foundry for next generation of TCB tools, which is ultra-fine. We have a good technology to address.

Thank you, Robin and Benjamin, for taking my question. My first question is regarding to some... Uh, I'll look so, uh, as usual, wondering if you could kindly give us some colors on the booking momentum across semi solutions and SMT in the first quarter, uh, this year. And also, uh, uh, when you mentioned about the advanced packaging sales was, uh, 22% of total sales last year. Um, I asked how much is from roughly from the TCP. So that's the first question.

Okay. We don't really break it down in terms of the different kind of tools for our AP. We did very clearly mention that for 2023, TCB was the highest contributor when it comes to as percentage of revenue for AP for the entire group. I think that's the only color we can give in terms of the breakdown in terms of for TCB. And for your portion on the bookings or bookings, some color. We don't really give official guidance, but I think Robin can provide some bookings color for 2024 Q1 for both SEMI and S&T. So we'll let Robin share more on this.

Thanks, Rob. And thanks, Donnie, for the question. We see bookings bottoming out for both SEMI and S&T in Q1. We expect Q124 bookings to be around plus or minus 20% better than Q423, which you know, looking at the numbers, which Q423 was a very low quarter historically. While we share the view of many others that semiconductor industry looks set for a gradual recovery in 2024, But I think at this point, it's still too early to have a clear line of vision on the rate or the slope of the recovery of our business in 2024. A little bit more color between the two segments. At the segment level, we expect semi-Q&Q bookings growth to be higher than that of SMT. because SEMI was coming out from a very low base in Q4'23 while SMT bookings only started to soften in the second half of 2023. So I hope I answered your question, Tony.

Yes, very clear. Thank you, Robin. Just one follow-up is I guess for the SEMI advanced packaging, Spooking should be growing faster than conventional packaging, right? Or it's not the case because conventional packaging base has been very, very low in the fourth quarter last year.

Yeah, right. Tony, you're right. That's the case for advanced packaging versus what we call more the mainstream packaging solution.

Okay. Thank you, Robin. And my second question is regarding to advanced packaging. So maybe from two angles. The first one is TCP. So Just want to have some directional comment from you is that do you think that our TCP business opportunities this year will be lying more on HBMs or logic ICs? And for hybrid bonding, I know you're not able to comment too much, but wondering what kind of applications will start to use your hybrid bonding first is that?

uh memories or or logic ics and is that the demo tool or is the or them before mass production thank you okay uh donnie let me break it down because these are two questions firstly on tcb you want more of a directional comment from uh robin that uh which segment either hbm or logic will sort of have a better potential for us so i'll let robin comment on this first

Tony, I would say the logic space for both HPC as well as generative AI. As you are aware, we are still very deeply engaged with the logic IDM for chip-to-wafer TCP applications, and we see it continue all the stream coming from that space going forward. For the chip to substrate the logic for 2.5D packaging from the leading foundry, we are also confident that this demand will continue to grow in 2024. As you know, we are probably at only the beginning stage of the generative AI boom. So the industry involved in this area is building a lot of capacity. So this will come on stream progressively. So we bring the, you know, so well positioned for this space, right? So we are confident that we'll continue to get more traction in terms of supplying chip to substrate tools, TCP tools for the logic side as well. And not forgetting, I also mentioned earlier that we are also engaging customers for chip to wafer application, which we are confident it will come because of the increasing demand, you know, in terms of bumpage and placement accuracy at the chip to wafer level, right? So at a certain point, mass reflow solution will face its limitation and TCP because of technological, higher technological capabilities, we have to, we'll replace, you know, the MR solutions for higher and higher AI chips. Now, in terms of HBM, I think you should be aware that we have already shipped four tools to a HBM player for for 12 high. We also believe that at some point as HBM continue to migrate from 8 to 12 to 16 high, increasingly I think the MR solution will also face some limitation, right, in terms of bumpage, in terms of placement accuracy. So again, I think at that space from 12 high HBM to 16 high, we believe that TCB will be in a sweet spot for such application. So in terms of timing, well, we looked at logic space for IDM, leaning foundry, offsets for 2.5D packaging, for chip to substrate application, application will continue the momentum in 2024. For HBM, probably will come at a later point, maybe towards the second half. and into 2025. That's the plan that we're looking at for the various applications for TCP.

Thanks, Robin. The next one is on hybrid bonding. Basically, can you comment on what kind of applications is our hybrid bonding tool more relevant for and is getting the order and traction for? Next is, can you also comment whether these tools are going for demo or production?

Yes. Of course, we, you know, sometimes we don't exactly, you know, a customer don't exactly tell us what they are being used, but we can sort of discern, you know, from the packages that these are for, we believe it's for 2.5D logic applications. They are not demo tools, don't need, they are, we believe these two, one of these two are already, probably already in low volume manufacturing mode. Once we ship to these customers in the second half of 2024, they will use it for low volume manufacturing. Yeah.

Okay. Thanks, Robin.

Can I request Dylan to unmute yourself and ask a question?

Yes. Good morning, everyone. And thanks for taking my question. So my first question will also be on TCP and two parts. First on logic. Yeah, so as you mentioned in the earnings release, we are engaging with the leading foundry on chip to wafer kind of variants. And yeah, I think this seems promising. But all these, if we translate to near-term kind of outlook, we also mentioned this year the TAM for advanced packaging could be $1.7 billion. And judging from that, how much of the market share do we expect to get from this addressable market for this year? And or as a reference, how much did we get during 2023, especially for TCB? because I guess most of the questions during the past earnings calls we have been discussing about TCB. So I guess it would be helpful to get a sense how much of a revenue have we got from the specific tools.

Dylan, I think I will comment one thing here that in terms of the exact market share or that detail, I'm sorry, we won't be able to provide. But let me request Robin to maybe give some color on what's happening with our TCB, especially we'd say this leading foundry and, you know, some other applications and what is the potential. Then I think you can gauge better that, you know, how we are progressing. And I believe we definitely have a dominant market position when it comes to our TCB site. We have the highest install base. And let Robin highlight more to you.

Sorry, sorry. Mel will add a little bit on the memory side as well for the TCB part. Sure. Yeah, so sorry for the interruption. So for logic side, sorry, for memory side, you also mentioned the HBM part.

Maybe we're looking at... Let me answer your first question first, Dylan. Now, if you look back for TCP 2.5D patches, it has been around for a while, actually. and the the first solution actually for 2.5 the packaging whether a chip to wafer or to the subject uh it was they were using a mastery flow 50 mass free flow but as the compound die you know for chip to substrate application gets bigger and bigger the mass free flow solution will face some constraint and that's where tcp is best placed to place large compound dyes for chips or substrate. Because for mass reflow, the bonding for the dye onto the substrate is offline because they still need a reflow process to do that. Whereas for TCB, it's in-situ bonding. So you pick out the large dye and you place very accurately onto the substrate, and then you bond it straight away, right? So, LushDyte, TCB is suited for LushDyte because of this, as well as LushDyte tend to warp. You know, so TCB has better handling of warpage compared to mass reflow. That's why going forward, as the compound dye gets larger and larger, TCB will become more and more relevant. you know, compared to mass reflux process going forward for large time. Now, that's why we want the orders in Q3 and Q4 of 2023 because of a warpage handling capability for TCB as well as, you know, the bumpage and displacement accuracy requirement. So we see this trend will continue because as AI chips become more more and more powerful that, you know, the interest would tend to pack more and more chips into a real estate. So, certain real estate. So, the time would get bigger and bigger. And that's where TCB has a sweet spot for the kind of large-time chip to substrate applications. Now, for chip to wafer applications, The picture is slightly different. For chip-to-wafer, the requirement is actually more on bump pitch and placement accuracy. So you need to put a lot of chiplets at the chip-to-wafer. So again, TCB has that advantage over the current solution, which is a mass reflux. right, because of the same reason. MasteryFlow is offline bonding, TCB is in-situ bonding, right, in-situ bonding. So in terms of accuracy, much better, you know, for TCB compared to MasteryFlow. So we believe TCB, that's why we are excited about the prospect of TCB because increasingly for 2.5D packaging, TCB will be the tool of choice for both chip-to-wafer as well as chip-to-substript. Now for memory, kind of similar picture because memory die are very thinned right so the TCB can handle thin die without breaking the die better than a mass refold process that's why we are also we are also excited about the future of TCB being used increasingly for 12 high and going forward into 16 high as well especially with a new generation of TCB tools, whereby we could, you know, the bumpage could go down to below 10 micron, accuracy down to below 1 micron, you know, with a kind of capability, TCB in terms of handling thin die as well, and placement accuracy will be very relevant, you know, for HBM, 12 height and beyond. I hope I answered your questions, Dylan.

Alright, maybe a follow-up on the mass reflow part, because we also mentioned our capability in Flipchip, and that's mass reflow, right? So, just to clarify, when we talk about our opportunity in Flipchip, the panel-level packaging and stuff, that's separate from the 2.5D advanced packaging opportunity, is it?

So, Dylan, can you repeat the question again?

Oh, so because we're saying that TCP could be better positioned in 2.5D advanced packaging kind of use cases and compared to mass reflow. And at the same time, we also highlight our capabilities in Flipchip or mass reflow technologies. So are these the two like separate applications or we are, you know, Flipchip is also in 2.5D advanced packaging right now, but at some point maybe TCP is that will replace it.

I got a question right now. Both MR and TCB will coexist in the future because not every AI chip require that kind of bumpage and the kind of precision that is required for TCP. Let me put it that way. It's a range of AI chip. The lower range of AI chip is still possible to go for MR solutions, but increasingly, as we demand more and more from generative AI, increasingly, I think the trend is to move towards a TCP solution. I hope I answered your question, Dylan.

Yes, got it. Thank you very much. So maybe my next question will be on R&D because we also highlighted we will spend an incremental 250 million of R&D for maybe our R&D capabilities, infrastructure and stuff. Can we try to quantify how much R&D we are spending for advanced packaging? Because I think as a group, We have quite some R&D budget compared to our peers, but when it comes to only for advanced packaging, how much resource are we putting into it? So like I mentioned previously, we are now progressing well with TCB and even we are trying to expand our market share in hybrid bond. And with all these opportunities out there, do we have sufficient R&D resource to support all these projects, especially for those so-called AA customers?

Dylan, this is Katie. Thanks for the question. Let me probably give you some context. And first, a correction of the $250 million investment that I mentioned in the prepared remarks. It is for both R&D and infrastructure. So of the $250, roughly half of that is for R&D, okay? So then coming back to your question, R&D, as Robin mentioned, S&P has the leading technical and product positions in several areas. So we'll double down on our AP investments in R&D resources and capacity. So the R&D resources basically will be deployed to engage our potential customers earlier and more embedded. They will work with our newly expanded customer base very closely, and they will continue our development in next generations of technologies, such as TCB, Hyperbond, Photonics, a couple of them you mentioned those already. So, other than the OPEX of half of 250, just so you know that we actually have a CAPEX investment planned as well. to ensure that we have the right capacity to meet the growing demand. So in short, we're very excited with our market-leading position, and we will double down. So in terms of the R&D resource, you had a question about percentage. We don't share the exact percentage, but I can assure you that for semi-R&D, a very decent portion of the R&D expense actually goes to AP-related technologies. Okay, so just a little bit more color on the other half of the OPEX, that's more so for the infrastructure side. When we say the infrastructure, these are the projects like the ERP overhaul, new HR and R&D systems, cybersecurity enhancement, et cetera, right? So those are multi-phased and multi-year projects. And they are at different stages of implementation right now, and it will intensify in 2024. Upon completion of these projects in the coming years, we expect to have better efficiency and productivity across different functions. So Dylan, I hope I answered your question.

Sorry, just to clarify one point. When you say double down the resources for AP, how much of the incremental OPACs we're talking about? Because we don't have a percentage of R&D or OPACs disclosed specifically for AP, but we're doubling those budgets, right?

Okay, when I say double down, I think it's a descriptive term to say that we are determined to invest more. It's not a double down in mathematical terms.

Ah, got it.

But again, I think the incremental R&D that we talked about, the half of the $250 million, you can't consider the majority of that going to AP, if not all.

All right. Thank you. This is all from me for now. I'll be going back to the queue.

Hi, Eric. Can you unmute yourself and ask your question?

Yeah, thank you. Good morning. Thank you for taking my question. I'm asking on behalf of Sonny and Randy. So, my first question is also around TCP. I think in previous interviews, we talked about 24 being the inflection point of TCP growth. and that HPM could be a more important driver than logic given more stacks. But I think from today's presentation, it sounds like you're even more upbeat about logic. So I was curious about whether the statement about memory better than logic in terms of TCP RAM with steel holes at this point. And that's my first quick question. I have a follow-up.

Yeah, Eric, right? Eric, let me answer your question. Yes, we are, at least for 2024, we are more upbeat in terms of logic application RAM compared to HPM for the reason I said a few times in the call already because at this moment the HPM are still focusing around the egg height kind of space So at that level, MR solution is still the preferred solution in our opinion. But as HBM migrate to $12,000 and beyond, that's where we think our TCB solution would be a sweet spot. Okay, so I think that's a short answer for your question. Any follow-up question from you, Eric? Eric, can you unmute yourself and ask your questions?

All right. Thank you, Robin, and management for taking my questions. Some of the HBM questions were addressed by my other peers. So I just want to ask a little detail on as we're going to 12 high or 16 high, how do you see the TCP capability compared to hybrid bonding? and as we are talking about we are having more and more confidence in getting more of the scoring more tools for hbm application are we talking about for tcp or hdn products you mean whether we are talking about hbm related application for our tcp and hybrid bonding yes okay

I think for TCP solution for HBM, there's a long way to go because we are together with our partners, we have continued to push the boundary of TCP for HBM solution going forward. Now, it's the perennial question and answer, right? So we have been saying at the end of the day, it boils down to the cost of ownership using TCP solution versus hybrid bonding. We mentioned this many times that I think between the two, at this point in time, it tends to favor a TCB solution because of the total cost of ownership position, favorable position for TCB versus hybrid bonding. However, going forward, increasingly, as we say, these three technologies, whether it's a mass reflow, whether it's a TCP, or whether it's a hybrid bonding, we strongly believe these three interconnect technologies will continue to exist and coexist at the same time. It all depends on the kind of requirement, right? So if you need very high-end performance, probably HP is a solution. If we don't need the kind of performance, maybe MR could be the solution. But we believe TCP is at a sweet spot, you know, overlapping these two areas, MR and hybrid bonded at this point in time.

Okay, thank you, Robin. Just a quick follow-up. For 16i, could TCP, our capability currently, support the specific spec requirement for 16i HBM?

there's really no any volume production for this particular requirement at this point. But looking at our next generation of TCP tools, we believe we are well-placed to serve the HBM for 69. Thank you, Robin.

My second question is just on the, I guess, the mainstream products that are not available and advanced packaging. Just wondering, in the past three months, do we see more of the, how do we see our customers' order trending? Are we seeing more order pulling in or sound of order pushing out?

Okay. Good question. Now, I think increasingly as we do our channel tracking, we go customer base for mainstream packaging other than AP, uh we are quite happy that to hear that the utilization the federal utilization rate has been uh increasing although not in a very sharp manner but at least increasing over time. So, you know, for tool supply like us, we monitor this matrix because when you hit a certain level, factory floor utilization, then you'll start to order new tools. So I think that is encouraging. But is it broad-based? Probably not. For Q1 so far, we see some uptick in demand from China for some of these mainstream applications. But these tend to be more, in our opinions, a little bit more sporadic than profits.

Yeah, thank you very much. So, yeah, we appreciate. Number one question is regarding the revenue mix trend, focusing on the 2023 results. And then the second question should be more TCV and hybrid bonding related. And then I have some suggestion for your PT material. So the revenue mix, a little bit confusion is the advanced packaging is a very promising growing area. But the 22% out of $1.9 billion, that includes the SMT as well, right? Non-SAMIs related.

Yes, it does.

Yeah. Any rough idea? What's the exact rough idea of the mix of the other ones, the packaging for pure SAMIs, which is mainly for bonding area versus SMT?

we don't really break it down. And as you know, that under Semi, we have quite a few solutions for AP. Then we have a smaller portion of AP, which is under SMT. All I can say is majority is from Semi, but we don't exactly break down the revenue split between the different AP products.

Yeah. How about this way to divide your maybe pure Semi's document for the advanced packaging? How about this way? Previously, I think you mentioned that the good track record for the TCV since 2012, so up to maybe sometime 2023, the cumulative Shinman unit about 300 units, and then the price seven-digit million, seven-digit dollars, so that means about $1 million per unit price. So would you please recap? Maybe already you closed the 20...

Now, between the two segments, we feel that the Semi will recover stronger than S&T because Semi has been in the downturn for two consecutive years, 2022 and 2023. 2022 was a steep one, right? So if you look at our booking for Semi in Q4, it was really low, right? So bouncing off from a very low base is kind of expected. In light also with the overall view that 24 will be a gradual recovery year. Now, SMT, as we mentioned many times, also in the NDA, that they follow slightly different cycle, which is a good thing for us, right? So we have two segments that have different cyclical pattern. So semi will recover. first, and then maybe one to two quarters later, S&T will follow suit. So this is the booking trend. In terms of revenue, you know, it takes time to convert our orders, right, into delivery. And this is also dependent on a few factors. First is what kind of orders are we receiving? You receive more AP orders, then the lead time will be longer than maybe around six months, nine months. But if you receive more mainstream orders, it can be converted quite fast. So there are a number of factors, you know, which will determine the conversion from booking into delivery. So we can't really see too far, but these are some of the metrics that we are watching.

And you think SMT could still be a growth year this year based on whatever we see right now?

I think from Booking's perspective, because of the recovery of the semi-year flow down to SMT, for the revenue side, it all depends how strong this look of recovery will be for Booking. If it's stronger... In the first half, we could see a recovery of revenue, right? But if it's not so strong in the first half, then we probably will push out to 2025. So it all depends. We can't see at this point in time.

Okay. Thank you.

Thank you.

I think we'll have to conclude the Q&A session there. And for remaining questions, we'll reach out to the analysts directly. And just before closing, can I request Robin to say a couple of words, and then we'll end the call.

So let me take this opportunity to highlight some key takeaways. Our unique block-based portfolio provides a range of advantages that include resilience during down cycle, our rich technical expertise and engagement with many different types of players. Together, this enables us to scale higher and more rapidly when the industry moves into a growth phase or when key application areas such as generative AI fuel a demand cycle. We are accepted and confident of our AAP potential based on a number of factors. One, our entrenched position with major AAP players. Two, our expanding AAP addressable market. Three, our confidence in capturing more market share. And fourth, our leadership in TCB that position us well to meet accelerated adoption for Logic and HBM. This concludes our call and I'll see you all in the next quarter. Thank you and take care. Bye-bye. Thank you.