DISPFDISPF
LOADING
|||
Yoshida: Thank you. I am Yoshida from CLSA Securities. Thank you. First of all, I would like to ask you to give us an overall summary of the actual shipment amount and the Q2 forecast, which we always ask you about, as well as comments by application for dicers and grinders, and also the composition ratio of major products. Thank you.
Koba: I will now answer regarding the status of the shipment. The amount was JPY111.1 billion in Q1, which means that we have achieved a record high for a quarter. The company has achieved well above its external forecast. First, I will briefly explain this upward revision to the forecasted figures in the external announcement. The overall increase was about JPY9 billion, of which more than JPY5 billion was for machinery and equipment, and about JPY3 billion was for other products. As for the other miscellaneous parts of this project, we have always been conservative in our forecasts each time, so there is no particular reason or background for them. The main reasons for the JPY5 billion increase in the shipment of machinery and equipment were that some projects that were expected in Q2 and later were brought forward, and there were some unexpected projects, for example, some evaluation machines had been delivered to customers and were recorded as shipments this time, which resulted in an increase of about just over JPY4 billion. We will then explain, as usual, the QoQ movements for each application. We have not separated the dicers and grinders for this presentation, and we hope you will forgive us for not explaining the equipment as a whole this time. To put it simply, this time there was a significant increase in ICs, mainly related to generative AI. On the other hand, the amount decreased in non-ICs, showing a clear division between the two. As for IC, OSAT increased significantly, in addition to the main generative AI-related products. In particular, the increase in memory was rather large compared to the logic. On the other hand, non-ICs, in general, showed a slight decrease. The decline was particularly large in power semiconductors. We had originally expected a decrease in this category, and the downward trend in power semiconductors is still continuing against the backdrop of weak demand for EVs and automobile-related products. In addition, the shipment value of image sensors and non-semiconductor electronic components also declined slightly. These are the actual results. In addition, the forecast for Q2 is almost the opposite of the increase factors discussed in Q1. In Q1, we expect a large decrease in shipments for ICs in Q2, partly due to the advance mentioned earlier and partly as a reaction to the considerable concentration of shipments related to generative AI in Q1. Dicers and grinders are trending in much the same way. However, we do not think it is that demand is slowing down in the generative AI areas, but it is more about the timing of shipments. Therefore, although it is difficult to foresee what will happen in Q2 and beyond or Q3 and beyond, we are not expecting a slow decline in shipments. We believe that Q2 may see a decline in the shipments due to its timing. For other minor applications, there are no major developments. However, unfortunately, we still do not foresee a recovery in power semiconductors in Q2, so we will continue to monitor the recovery of the EV market and the automotive industry. Also, the ratio. As we always talk about, the memory ratio in Q1 which we finished was just over 40%, which is about what we expected. The OSAT ratio is 30%, which is also almost as expected. As for power semiconductors, the figure is less than 10%. The ratio of power semiconductors has been considerably low from the current term, so we would like to make a nondisclosure regarding the silicon SiC category. Each is in the form of a single-digit percentage. In Q2, the memory ratio is expected to be 35%. Power semiconductors and OSAT are expected to account for less than 10% and 30%, respectively. That is all for now.
Yoshida: Thank you for the details. One more question. You made some comments earlier about this H2, but I would like to ask you again about the direction you see for H2. For example, do you see the level of orders as being about the same as in H1, or do you see it recovering a little more? I would like to ask for some more details on your thinking for H2. I would also like to know if there is any change compared to three months ago.
Koba: The first thing is that it is quite difficult to forecast the entire H2. We are now grasping a vague sense of the level of Q3, and our initial focus of what Q3 will be like is that it will be a little more positive than Q2 at this point. Unfortunately, there is still a very unstable area regarding Q4, which is why it is still difficult to see things through. In terms of changes, there has been an increase in the number of projects with short deadlines overall. There is some indication that OSAT is changing a bit. In addition, in the past, due to the COVID-19 pandemic and the subsequent demand for power semiconductors, many customers provided relatively long forecasts for their projects, but recently they have begun to feel that they do not need to issue forecasts that early and are returning to the old way of doing business, which means we have more and more projects with relatively short time frames. In other words, I think the point of change is that we are returning a little bit to the old-fashioned situation where customers are not so willing to make POs. What is the background for this? One of the reasons is that customers are no longer fighting over the equipment, and there is no longer the overheated feeling that existed in the past. Also, as I mentioned earlier, not all but some of OSAT’s customers are changing and many of them do not provide their forecasts so early. So, we believe that Q3 may also receive more projects in the future, given the timing of when they are recorded as shipments, and I believe that Q4 is even more difficult to forecast but there may be additional projects. The outlook for generative AI in general is difficult to foresee. Our comments on this point have not changed significantly from the previous quarter, but I still think that the willingness to invest in HBM varies from customer to customer. Therefore, my impression is that there is still a division between memory manufacturers that are willing to make additional investments and those that are not among our customers. On the logic side, the market is relatively firm compared to that of the memory, but we are seeing a slight decline in Q2, and I believe that the timing of shipments is also repeating itself, with concentrated shipments and slight declines depending on the quarter. At this point, we do not have a clear picture of the demand for generative AI-related products for the next fiscal year and beyond. Did I answer your question?
Yoshida: Yes, thank you. That is all.
Suzuki: Thank you very much. Now, please ask your question.
Tetsuya Wadaki: My name is Wadaki from Morgan Stanley MUFG Securities. Thank you. First of all, I would like to know how your strategic products are doing, specifically NAND hybrid bonding, WMCM, VCS memory, and PLPs. What is the situation like?
Koba: To choose from the items you named, let me explain the NAND bonding. I believe we are making some progress here. I think we have been talking about it in terms of R&D for a long time, but I am aware that some customers are gradually entering the mass-production phase. So, my impression is that we are making more progress here than we were three months ago. However, I do not think that all customers are shifting to bonding at the same time, so regarding how the market will grow in the future, we will keep a close eye on the demand for end products, as it will be closely related to the demand for NAND technology. As for WMCM, we are not yet in a position where we can clearly see a large demand for it. Basically, we are aware that the equipment we provide does not require any special equipment, so I think the key point is how much capacity will be expanded. In 2015 and 2016, there was a huge demand for Fan Out Wafer Level Packages, and some of the equipment for back-end processes was delivered at that time. So, I think the key point this time is how much additional capacity will be needed. As for PLP, my impression is that the number of projects we talk about with clients is also increasing compared to three months ago. However, in terms of schedule, we are still waiting for the timing for full-scale mass production, or the timing for shipments and contribution to our business performance, but we are receiving inquiries from various customers, so if the PLP market is firmly established, we will be able to capture demand. We are waiting for them with various preparations. As for PLP, I still think that trends in OSAT and others are also becoming more prominent. I think that OSAT have not been successful within the area of so-called 2.5D package, so I think that OSAT would like to take on a good number of projects for the PLP. I think there are a variety of advanced packages other than that as well, but we are not yet in a position to see them as a major volume. Various R&D activities are underway.
Tetsuya Wadaki: Just to confirm, how much of this memory, which says just over 40%, is flash and how much is NAND? What does it look like in a nutshell?
Koba: I don’t have the figures at hand, but there was a lot of DRAM, and I think the ratio was about 6:4 or 7:3, or 7 for DRAM and 3 for NAND, or 6 for DRAM and 4 for NAND, something like that.
Tetsuya Wadaki: Thank you very much. One more question. Sales of consumables seem to be a bit lackluster as of late, is there any change occurring? Is it because SiC’s situation is still like that?
Koba: I think one thing that has an impact here, especially when you look at the YoY comparisons, is the clear decrease in the amount of SiC consumables. However, I feel that the occupancy rate of customers is steadily increasing, although it may just be in line with seasonal trends, so I expect it to increase in Q2 or Q3.
Tetsuya Wadaki: Thank you very much. This was very helpful. That is all.
Suzuki: Thank you very much. Now, please ask your question.
Nakamura: Hello. I am Nakamura from Goldman Sachs. Thank you. In my first question, I would also like to confirm some points about the shipping. First of all, you mentioned earlier that shipments were brought forward from Q2 to Q1, but can you give us a hint as to the destination or its application? Also, can you confirm if there is any talk of equipment shipments being moved forward or backward due to tariffs, or if this is something that is being discussed currently?
Koba: In terms of the projects for advance shipments, there are various, several forms with no particular trend. We did have a memory project. Also, we also had projects with OSAT. It is also a bit scattered regionally. Some of the projects are in China and Western countries, so many of them do not have a clear background. For OSAT, there is probably considerable competition among them, so when a certain memory business is acquired, for example, companies prepare for the project as quickly as possible, and this is why some shipments are accelerated. We have also had demo units, or equipment that had been evaluated by customers, and such units were shipped in a project this Q2. This product was being developed on an R&D basis and shipped, and since it is difficult to incorporate this area into the forecast, it was recorded outside the forecast. As for the impact of tariffs, it is difficult to say at this point how this is affecting the schedule moving forward or the timing of investment. It would be nice if we could see something very negative or, conversely, very positive, but to be honest, we have yet to get a clear picture of the actual situation. I cannot say that there are none at all, but that is about it. With regard to tariffs, as you know, projects in North America account for about 10% in our company, so I do not think that there is any significant impact on projects destined for North America itself.
Nakamura: Thank you. In terms of follow-up on shipments, you mentioned earlier that the direction of shipments in Q3 is expected to be even stronger than in Q2. Can you give us some hints on the destinations and applications that will drive this? Can you give us a hint as to whether this is within the scope of the unstable shipping of generative AI, or whether things like the 3D NAND hybrid bonding you mentioned earlier will also be accounted for from around Q3?
Koba: Basically, I think it will recover a little because of the fluctuating generative AI-related issues, and as far as other uses, I think there is quite a bit of miscellaneous stuff. For example, we have been talking about power semiconductors being quite sluggish, but there is a timing in which shipments return to some extent in each quarter, so Q3 is factoring in a few of these projects.
Nakamura: Thank you. As for the hybrid bonding part, what do you think?
Koba: So far, I haven’t seen it yet in such large chunks.
Nakamura: I see. Thank you. My second question is about profitability. I would like to know your assumptions for gross margin and SG&A expenses for Q2. The gross profit margin for Q1 was a little over 68%, which seems to be a little lower than the same period last year, but is this due to foreign exchange or the product mix, or is there some other factor such as a decline in profitability within the same product group?
Suzuki: First, I would like to talk about the April-June results. The GP ratio for the April-June period was 68.1%, which was lower than the same period last year, but we understand that the foreign exchange rate had a large impact. The exchange rate for the April-June period was JPY144 to the US dollar, compared to JPY158 for the same period last year, and the appreciation of about JPY14 this year had a significant impact. If we apply the same rate of JPY158 as the previous year and calculate the GP ratio, we estimate that the GP ratio would be approximately 70%, so we understand that it was mostly due to the exchange rate. Regarding the plan for the July-September period, the GP rate is expected to decrease by less than 2 percentage points from the April- June period. The factor here is again the exchange rate. We have set our forecast at JPY135, so the yen is estimated to be a few yen higher than in the previous quarter. If the exchange rate is stable from April to June and from July to September, the GP rate is expected to be almost the same and stable if the exchange rate is stable. The concept is the same as usual, and the plan is basically based on the trends of the previous year, FY2024. Looking at the balance of uninspected accounts at the end of June, I have the impression that there are still some relatively high value- added projects remaining, so I think there is a certain degree of security in this area. Also, OSAT projects are recovering partially. Therefore, it is a question of what will happen in such an interaction. As for SG&A expenses, we expect about JPY27 billion for the July-September period. As in the past, the calculation of performance- linked bonuses is based on the amount of shipments, which in turn determines the level of SG&A expenses. In terms of the details, we have estimated the cost to be JPY27 billion, taking into account the fact that we have been making in-house fine adjustments to the bonus reserve from time to time. That is all for now.
Nakamura: Thank you very much.
Suzuki: Thank you very much. Now, please ask your question.
Koba: It is a little difficult to give you specific figures related to generative AI as the ratio varies from quarter to quarter, so please forgive me for not being more detailed. As for Q1, the ratio was high partly because HBM shipped a fairly large number of units.
Kamisaki: I understand. Thank you. Looking at the Q2, is it fair to say that the ratio is about 7:3, which does not deviate significantly from what you have previously mentioned?
Koba: I am sorry, that is a bit difficult to tell you at this point. However, as I mentioned earlier, in Q1, HBM is being shipped quite solidly, and I think we will see a reactionary movement in the Q2.
Kamisaki: I understand. Thank you. I know you may not be able to say this either, but could you please comment if you can on what you feel the ratio of generative AI in consumables should be right now?
Koba: I am sorry, but we do not calculate the ratio of generative AI-related products in consumables, even within the Company, as it is difficult to separate it from other categories, so we have not captured a specific ratio.
Kamisaki: I understand. That is all from me. Thank you.
Suzuki: Thank you for your questions. Now, please ask your question.
Koba: In terms of OSAT, if we limit ourselves to Q1, we have seen a significant increase in shipments of global OSAT, including Taiwan, which had been a bit weak. However, as we explained in the previous quarter, shipments have been fairly solid in this Q1, so we do not yet see the market growing all the way to the future. We have not yet reached the point where we are seeing a full-fledged recovery. Please note that what I mentioned earlier about PLPs is that the number of projects with customers is increasing, including various business discussions and does not mean that shipments for PLPs are increasing at the moment. #### I understand. Do you now surmise that the fact that OSAT in Taiwan is quite strong in the West at the moment means that the impact of tariffs and other factors are involved?
Koba: No, we cannot really analyze it that clearly, and OSAT does not explain that it is due to tariffs, so it is a bit difficult to forecast. #### I understand. Thank you. Lastly, I do not think you have given us a sense of the scale of generative AI for each quarter, but I understand that the scale has increased considerably in Q1 compared to last year’s JPY55 billion, on an annual basis. Is my understanding correct?
Koba: Yes, As you can see from the shipment by application and use, the driving force behind the increase in shipments for ICs was related to the generative AI, so the figures for Q1 were the largest when viewed by quarter. #### I understand. Do you have a feeling that if this continues, there will be an increase of maybe 20-30% or so on an annual basis, or do you have any comments in that area?
Koba: No, it is as I told you earlier. This is about Q1, and I see Q2 going down a bit, so I think there will be some fluctuation here. I do not think we should multiply the Q1 figure by 4. #### I understand. Thank you. That is all from me.
Suzuki: Thank you very much. Now, please ask your question.
Mikio Hirakawa: Thank you very much. I am Hirakawa from BofA Securities. I have two questions. First, I do not think we can determine it precisely, but if we divide the 30% or so of OSAT between memory and logic, what was the performance in the Q1? I would like to know if there is anything you can share with us about the QoQ performance and the overall sense of proportion and image.
Koba: Mr./Ms. Hirakawa, I am sorry. It is still hard to tell since we do not have the numbers with us right now, but both were strong in Q1. I feel that OSAT for memory was strong and OSAT for logic was also strong, so I do not have the impression that there is a bias toward one or the other.
Mikio Hirakawa: Thank you very much. Is my understanding correct that the OSAT for logic was strong because it was generative AI related?
Koba: I think that OSAT was not necessarily tied to generative AI, so it is a little difficult how to see it, but I think that it probably included projects for regular smartphones, for example. I believe that not only North American but also Chinese smartphones are included in the supply chain, so I believe those are included, too.
Mikio Hirakawa: So those are also included.
Koba: Yes.
Mikio Hirakawa: I understand. Second, you mentioned earlier that the DRAM and NAND ratio in memory was 60% to 70% DRAM and 30% to 40% flash, and I am aware that this was in Q1. If we look at last year FYE 3/2025, what would you say the ratio was?
Koba: I am sorry, I do not have the numbers there so I do not know, but for FY2024, I think there were quite a few DRAMs, so I would say it was still 70% or more. Perhaps there were a few more.
Mikio Hirakawa: Last year was strong where normally it would have been about 6:4, and although dealers are still strong, NAND is also getting stronger, with a ratio of 6:4 or 7:3 among them. Does that mean that the shipments in Q1 were such?
Koba: Yes, However, if we go back a bit further, to the past, there were many times when the ratio of NAND flash was higher, and I think that the ratio of DRAM has been increasing in the last one to two years with the advent of HBM.
Mikio Hirakawa: Okay. Thank you.
Suzuki: Thank you very much. Now, please ask your question.
Masahiro Nakanomyo: I am Nakanomyo from Jefferies. Thank you. Just one question. I am sorry to be very detailed, but I think SG&A expenses was JPY26.7 billion in the last quarter, and it is expected to be JPY27 billion for this quarter. I guess I do not properly understand what Suzuki has been able to teach us, but if we go by the fact that it is to some extent linked to the amount of shipments on a cash flow basis, I feel that it should be a little more for Q1 and a little less for Q2.
Suzuki: Thank you very much. As you mentioned, we have changed the method of calculation and provision of bonuses since the last quarter, and the calculation method is now linked to the amount of shipments on a cash flow basis. However, as I mentioned earlier, we are actually conducting detailed fine-tuning of our internal operations as needed. Therefore, the amount of shipments will not be allocated to the quarter because of an upward swing, but rather, a certain portion of the amount will be recorded as a sliding scale in the subsequent quarter. This is due in part to the fact that there was not such a large increase in SG&A expenses for the April-June period, but there was a certain amount of sliding adjustment in the July-September period. So, I think this is a bit of a detailed story, and it is very difficult for outsiders to see the details. Therefore, we will basically announce the level of SG&A expenses for the following quarter at the timing of the settlement of accounts.
Masahiro Nakanomyo: Thank you. In that sense, I think you had indicated that the SG&A expenses would be JPY27 billion or JPY28 billion for the JPY100 billion in the past, but I guess that kind of calculation is no longer necessarily true.
Suzuki: Yes. As you say, the actual situation is that we are not able to move neatly within the range that we have shown in the past. I apologize for the inconvenience, but we would like to inform you of the level of SG&A expenses as soon as possible when we release the forecast.
Masahiro Nakanomyo: Thank you very much. Very clear.
Suzuki: Thank you very much. Now, please ask your question.
Yoshioka: Thank you. I am Yoshioka from Nomura Securities. I would like to ask two questions. The first question is sort of a confirmation, and I think in the very first question, you commented on the level of shipments in Q3 and beyond. So, I understand the direction of the QoQ increase from Q2, but I am sorry, how should we think about its level again? In H1 of this fiscal year, the average for the quarter would probably be around JPY99 billion in terms of flat exchange rates, so I am not sure if I understand the level of this as an image, or if it is just an up-and-down kind of thing, and of course we are at a stage where there is no guide for calculation. Could you make any comments on the level?
Koba: As for the level, there are still a large number of unstable projects, so I cannot give an answer as to the extent of the upward swing at this time. However, at this point, we have been able to confirm that the number of projects has increased to some extent, so we believe that there is at least a high degree of certainty that the number of projects will increase from Q2. That is how we see it at this point, and as for the specific level of growth that is likely, it is difficult to say because we have not issued a guide for Q3.
Yoshioka: Thank you very much. The idea is that Q2 is low just because of the timing, and Q3 is just the timing to recover, but do you see any additional new elements or other contributions?
Koba: At the moment, we are not anticipating any new material, any major changes, or changes in market conditions. It may be that Q2 is a little lower, but to be more precise, I think in Q1, as I have been explaining, the level was a little too high. I hope you will consider that as well, and I think what you said earlier about looking at it as an average is a very solid view.
Yoshioka: Thank you very much. Secondly, this is a bit of a confirmation of the detailed figures, but I believe you explained that there was a JPY5 billion shipment backlog in Q4 of the previous fiscal year, and that JPY3 billion of this backlog will be included in Q1. For the JPY5 billion that was delayed, at what point in the end was this recorded as a shipment? A little update there would be appreciated.
Koba: We were told that the portion of Q1 that was delayed from Q4 to Q1 was about JPY5 billion, of which JPY3 billion was for Q1 and about JPY2 billion for Q2, and we are progressing almost according to that schedule. Q1 shipments were JPY3 billion, plus an additional JPY3 billion, and we have been able to ship some of those planned in Q2, and the rest will be shipped this July-September. As a result, we expect to be able to catch up with the JPY5 billion that is already overdue in H1 of this fiscal year.
Yoshioka: I see. I understand clearly now. Thank you.
Suzuki: Thank you very much. It is past the scheduled end time, but there are still people who wish to ask questions, so we will continue with them. Now, please ask your question.
Yoshihito Hasegawa: I am Hasegawa from Morgan Stanley. Thank you. We would like to have your suggestion on the status of the plant’s operation. I know that the shipping is subject to the convenience of customers, but could you give us some indication of the actual staffing and operating conditions at the plant? I understand that there are various situations, such as new employees who have joined it and not returned, but could you give us an idea of how the factory is operating?
Koba: First of all, we are not in a tight situation where we have to provide manufacturing support from our headquarters, as we had in the past during the COVID-19 pandemic. In the factory side, the situation of the operation depends on the equipment, and the slowdown in power semiconductors has had an impact, but the situation for related products such as grinders and laser saws has been relatively stable. On the other hand, we have been busy in the area of generative AI-related products mainly grinders, so it is difficult to express how busy we have been in total, but basically, we have been busy in general.
Yoshihito Hasegawa: January-March, April-June, July-September... You mentioned earlier that Q2 is subject to a rebound, but do you expect production in April-June and July-September, and toward the end of the year to remain relatively firm?
Koba: Yes, However, I am sorry to say that this also depends on the actual status of construction starts and inquiries from customers in Q3 and Q4 so it is difficult to clearly communicate this at this stage.
Yoshihito Hasegawa: On the other hand, there is no possibility of such a sudden drop.
Koba: Yes, As you can see from various news reports, the trend of generative AI-related products will continue over a few years, although it has been fluctuating in various ways. Although it has not yet been able to contribute to the business results, we have been talking with an increasing number of customers about PLP, which was mentioned earlier, and bonding such as the hybrid bonding in the future. I believe that this will be a factor that will make the factories busier in FY2026 and onward.
Yoshihito Hasegawa: Thank you very much. Regarding hybrid bonding, I think you mentioned that there is a lot of progress being made, but what exactly would you say the progress is in terms of what you are looking at?
Koba: Basically, our policy is that it is difficult for us to comment during the R&D phase. That is why we normally say that we are continuing the R&D. However, at the stage where the actual shipment amount is recorded, customers gradually shift from the R&D stage or small-lot mass-production to full-scale mass-production. I think it will be interesting to see if such a move will be made on the customers’ side in the future.
Yoshihito Hasegawa: I understand. Thank you. That is all from me.
Suzuki: Thank you very much. This concludes the question-and-answer session. Lastly, we have an announcement. We continue to receive meeting requests from numerous institutional investors and analysts regarding our proactive dialogue with investors, which is required by the Tokyo Stock Exchange in relation to our efforts to achieve cost of capital and stock price conscious management. Thank you. We have established the Meeting Code in order to make our dialogue with institutional investors and analysts more constructive and effective. By establishing introductory, market insight, long-time value insight, etc., in the press release and at each meeting, we will clarify in advance a common understanding of the purpose and progression of the dialogue. Through this initiative, we aim to deepen our dialogue in light of the Stewardship Code and Corporate Governance Code, and further improve the quality of our dialogue. Please check your closing schedule email for meeting code details. Thank you. With that, DISCO Corporation’s financial results conference call is adjourned. Thank you very much for joining us today.
Toba: Thank you very much. [END]