ADTTF - Advantest Corporation (ATEYY) Q2 2023 Earnings Call Transcript
2023-10-31 16:04:06 ET
Start Time: 03:00
End Time: 04:02
Advantest Corporation (ATEYY)
Q2 2023 Earnings Conference Call
October 31, 2023, 03:00 AM ET
Company Participants
Yoshiaki Yoshida - Representative Director, President, Group CEO
Yasuo Mihashi - Senior Executive Officer, CFO and CSO, EVP, Corporate Strategy Group
Douglas Lefever - Representative Director, Corporate VP, Group COO
Makoto Nakahara - Senior Executive Officer, CCRO, EVP, Sales Group
Yoshitake Kobayashi - Head of IR
Conference Call Participants
Yu Yoshida - CLSA Securities
Tetsuya Wadaki - Mitsubishi UFJ
Mikio Hirakawa - BofA Securities
Masahiko Ishino - Tokai Tokyo
Shuhei Nakamura - Goldman Sachs
Masahiro Nakanomyo - Jefferies Securities
Damian Thong - Macquarie Capital Securities
Presentation
Yoshitake Kobayashi
Thank you very much for taking the time to attend FY 2023 Q2 Earnings Call for Advantest today.
I would like to introduce our attendees today. This is Yoshiaki Yoshida, Representative Director, President, Group CEO; this is Douglas Lefever, Representative Director, Corporate Vice President, and Group COO; and this is Mihashi, Senior Executive Officer, CFO and CSO, Executive Vice President of Corporate Strategy Group; and Nakahara, Senior Executive Officer, and CCRO, Executive Vice President of Sales Group. I'm Kobayashi from the Investor Relations Department of the Corporate Strategy headquarters. I'll be the moderator for today.
For today, we'll begin with FY 2023 Q2 financial results by Mihashi followed by FY 2023 business outlook by Yoshida. After that, we will take any questions you may have. The schedule end time is 17.00. And please note that today's financial results briefing will be streamed with simultaneous Japanese-English interpretation.
If you like to use a Japanese audio channel, click on the globe icon at the bottom left of the Webex screen. And there will be a menu called My Interpretation Language, so please select Japanese. And also, if you move the slide bar on the balance menu below to the interpreter on the far right, you can listen to the Japanese interpretation while English is being spoken. If you want original audio in both Japanese and English, there is no need to change the settings. Please use the default settings.
Today's presentation materials are posted on TDNet and our Web site. If you're participating via phone line, please download them separately. Also, only Japanese materials will be projected on the screen at today's briefing session. If you need to view the English materials, please download the English version separately.
Prior to the briefing session, we have some information to share. The following explanation may include future outlook based on current expectations, but all of these involve risks and uncertainties. Please note that accurate results may differ from the forecast.
Now, I'd like to pass the floor over to Mihashi. Thank you.
Yasuo Mihashi
Now, I would like to share with you the financial results for FY 2023 second quarter. Please look at Page 4 in your materials. The second quarter sales and profit as you can see on this screen exceeded those of the first quarter. Although the yen’s depreciation against the U.S. dollar also boosted our performance, the business environment remained challenging as in the previous quarter, as customers continue to curb investment, especially in smartphone-related applications.
Gross profit margin fell below the previous quarter's level due to the deteriorating product mix. Operating income margin was higher than the previous quarter partly due to the impact of yen depreciation. Regarding the interim dividend, the Board of Directors today, October 31, approved ¥65 per share, the same as the forecast as of July. ¥65 is the dividend amount per share before the 4-for-1 share split as of October 1. Details will be explained sequentially in the following pages from Page 10 onwards.
Please go to Page 5. Sales by segment; Semiconductor and Component Test System SoC tester sales were ¥62.5 billion, an increase of ¥1.3 billion quarter-on-quarter. Sales for semiconductors mainly related to material processes such as for automotive, industrial equipment increased. On the other hand, sales for application processors which have been weak decreased. Although sales for high performance computing semiconductors where demand is solid also decreased. Deliveries of new tester equipment continue to be moderate, while excess capacity in customers supply chain is being digested.
Memory tester sales were ¥18.7 billion, an increase of ¥9.4 billion quarter-on-quarter. Although sales for non-volatile memory continue to be weak as in the previous quarter, sales for high performance DRAM including generative AI increased markedly. Mechatronic systems; sales of device interface products and test handlers increased quarter-on-quarter in tandem with an increase in tester sales. Sales of SEM metrology products also increased. Service, support and others, sales were on par with the previous quarter.
Please go to the next page, Page 6, sales by region. As you can see, for South Korea, sales of memory testers grew substantially. We have marked growth for DRAM. In addition, SoC tester sales were solid. Next will be about China. Sales increased in SoC testers and memory testers, respectively, due to delivery of testers for a variety of semiconductors.
As for Taiwan, the orange colored bar area, sales decreased namely for high end SoC data continuation of excess tester capacity related to smartphones. Testers, as I mentioned, testers that had been configured to test smartphones are now being converted to test HPC. As such, we expect Taiwan sales to increase along with an improvement of utilization ratio going forward.
Next page, please. This will be sales, gross profit and operating income. At the time of the first quarter results announcement in July, we had expected gross profit margin will show improvement in the second quarter and beyond, with the first quarter being the bottom. However, sales for high end SoC did not pick up as much as expected, resulting in a deteriorated sales products mix.
In addition, higher parts procurement costs resulted in a lower gross profit margin than in the previous quarter. SG&A was ¥37 billion on par with the last quarter. This includes other income of approximately ¥3.2 billion. This is mainly due to a settlement of a dispute with a counterparty. As a result, operating income was ¥21 billion as you see here.
Next page please. This is regarding expense and cash flow. R&D, CapEx and depreciation are largely in line with expectations. As for cash flow, in the first quarter, there was an investment cash outflow due to acquisition of Shin Puu. In the second quarter, free cash flow increased by ¥16.8 billion quarter-on-quarter.
Please go to the next page. This is the consolidated financial position. Trade and other receivables increased by ¥12.6 billion from the previous quarter due to an increase in sales. Inventories are at the similar level as in the previous quarter, despite the incoming delivery of strategically procured long lead time parts. We are currently working to establish a delivery system to meet customers' delivery deadlines in order to monetize inventories as quick as possible. We will make further efforts to strengthen inventory management.
Yoshiaki Yoshida
Yasuo Mihashi, thank you. From my side, I would like to talk about 2023 outlook. Please turn to Page 11. The current business environment is characterized by a slow recovery of the global economy, including the economic slowdown in China. In addition, business environment remains highly uncertain due to factors such as expansion of geopolitical risks, rising prices of natural resources and as commodities, and the risk of sharp exchange rate fluctuations.
In the semiconductor market, despite resilient and stable demand in certain areas such as automotive and industrial equipment, data center investment has slowed in addition to a decline in semiconductor demand in smartphones and other major consumer electronics applications, leading to continued capital investment restraints at related semiconductor manufacturers.
For the current year 2023 SoC tester market, we adjust our SoC tester market size estimate range to between $3.3 billion and $3.4 billion. We raised the lower end of the range compared to our estimate three months ago as tester for automotive, industrial equipment and DDIC showed a steady delivery progress while lowering the upper end of the range as we trimmed the potential for upside. As for the capacity ramp up during COVID-19, we believe it will still take some time before it is used up.
For the calendar year 2023 memory tester market, we adjust our estimate range to between $1 billion to $1.1 billion by raising the lower end of the range. While signs of demand recovering for non-volatile memory remain uncertain, test demand for high performance memory is growing as customers' investment appetite is increasing due to plans to increase production and strengthen quality assurance.
For high performance semiconductor such as those for HPC and AI, the effort to increase assembly capacity is going into full swing leading to an expectation for an increase in production volume from 2024 onward. There is also a stronger trend toward increasing the tester volume driven by the demand for higher performance in order to add value to semiconductors and for higher reliability. This trend is currently utilizing excess capacity in our customers and supply chains but we are also seeing our customers materialize their investment plans towards 2024 to 2025 based on a medium-term plan for demand growth.
Please turn to Page 12. This is the financial year 2023 forecast. In light of the first half results and the outlook going forward, we are revising down full year consolidated forecast. Sales forecast is revised down to ¥470 billion, which is ¥10 billion below the July forecast. Recovery of sales of testers for high end SoC is expected to take longer than originally anticipated due to the delay in recovery of demand for semiconductors for smartphones and other major consumer electronic devices.
Gross profit margin forecast which was originally projected at around 55% has been revised down to around 50% due to lower sales deteriorating product mix and higher material procurement costs. Operating margin is revised down to a level below 20%, although we are working to reduce non-essential costs. Exchange rates assumptions for the second half of the fiscal year are ¥145 for the U.S. dollar and ¥155 for the euro, respectively, to reflect the yen depreciation.
Our latest forecast for the impact of exchange rate fluctuations on FY 2023 operating income is a plus ¥0.9 billion per ¥1 of Japanese yen depreciation versus U.S. dollar and minus ¥0.3 billion per ¥1 of Japanese yen depreciation versus the euro. Due to the expected decrease in sales from U.S. dollar based transactions, we have lowered our estimate of the impact for the U.S. dollar by ¥0.2 billion from July.
Regarding the tightening restrictions on the export controls of semiconductor production equipment to China by the United States and its allies that had a direct impact on our FY 2023 earnings is currently expected to be limited, but we will continue to closely monitor the situation.
Please turn to Page 13. This is outlook by segment. First, FY 2023 Semiconductor and Component Test System outlook full year 2023 SoC tester sales forecast is ¥248 billion which is ¥20 billion lower compared to the July forecast. For mature processor products, customers’ investment is expected to trend solidly for applications such as DDIC, in addition to automotive and industrial.
On the other hand, demand for advanced process products is expected to fall more than expected against the backdrop of weaker demand for smartphones and other factors. As a result, SoC tester sales for the current fiscal year are expected to decrease significantly year-on-year. Memory tester sales forecast for the following year is raised by ¥6 billion from the July forecast to ¥76 billion.
In high performance memory applications such as DDR5 and HBM, including those related to generative AI, as a substantial increase in tester investment from customers can be expected going forward as they set to increase production capacity against a backdrop of medium-term demand growth and high reliability requirements.
Please turn to Page 14. In mechatronics and services, support and others, we have raised our full year forecast of FY '23 mechatronic system sales by ¥6 billion from the July forecast. Sales of device interface products and test handlers are expected to increase in conjunction with an increase in the full year sales forecast for memory testers. We have cut our full year '23 services, support and other sales forecast by ¥2 billion.
Therefore, maintenance services solid demand is expected due to the steady growth of an installed base. System level test business is expected to see a year-on-year decline in sales due to the impact of customers’ investment trends affected by the sluggish semiconductor market conditions.
Please turn to Page 15. So we have been talking about the business forecast for this fiscal year compared to three months ago. Our demand recovery is slower than expected compared to three months ago and ramp up is delayed. At the same time in the past three months, we also saw concrete progress in our business opportunities arising from generative AI.
There are two axes when it comes to factors which drive tester demand. One is enhancement of testing in response to technological changes, such as increased semiconductor complexity, and greater quality assurance, the other is increases in the conductor production volume. And that's what we are trying to show in one slide.
As semiconductor technology evolves, demand for testing is increasing in response to the adoption of the advanced technologies such as the migration to 3 nanometer processes and other miniaturization advances, which require more testing time to address new failure models and the increased importance of testing to enhance quality assurance for advanced packages that are stacked and mounted at the die level.
In terms of semiconductor production volume growth, major semiconductor foundries are expected to gradually increase their advanced package assembly capacity from calendar year 2024 onwards. In addition, customers that are developing and mass producing high performance memories such as HBM are working to increase their production capacity in response to midterm demand growth and high reliability requirements. Concrete discussions with customers about their plans for future tests or demand growth are ever increasing.
In addition, there is concrete progress for projects by several new players working on the design of semiconductors with diverse functions from service to end applications related to generative AI. The progress of those multiple projects is expected to increase the demand for testers over the medium term, starting in 2024.
Please turn to Page 16. So we've explained this several times in the past. We have drawn the ground design as you see on this slide. For AI related semiconductors, with the diffusion of advanced packages, test insertions in various stages to enhance quality of semiconductors is becoming important. These include wafer testing in a frontend process, package testing after packaging, and a system level testing.
Furthermore, for our customers to improve the yield of semiconductors, it is important to pursue quality improvement by utilizing our extensive testing expertise accumulated over the years in a series of processes from design to manufacturing. In the future, we believe it is highly likely that generative AI technology will change the very process of semiconductor development and manufacturing for our customers.
We will contribute to our customers' time to market, time to volume and time to quality by providing new solutions through Advantest Cloud Solutions, ACS, for the Air Force to utilize various data and utilize them in their development and manufacturing.
Please turn to Page 17. Here's the summary. We have lowered our full year sales forecast by ¥10 billion from the July forecast ¥480 billion to ¥478 billion. Customers continue to curb capital investment due to the slower recovery of smartphone related demand and digestion of excess tester capacity. It’s likely to take longer than what we expected back in July.
As mentioned earlier, there has been steady progress of business opportunities arising from generative AI. As such, our view remains intact for tester demand growth over the medium term. Against this backdrop, we will sustain R&D investment to address future technologies. However, in the current fiscal year, the business environment has remained sluggish longer than expected. We will strengthen our ability to respond to a demand fluctuation by establishing a delivery system to meet customers’ delivery deadlines through advancement of a supply chain management.
Furthermore, in light of soaring parts procurement costs, we will execute actions to improve gross margin by reviewing sales prices and implementing cost reduction activities as well as to set up commensurate cost reductions by reducing non-essential costs, cutting down headcounts, part of this is already done and the reviewing business processes. Additionally, we will focus on measures to improve profitability over medium and long term by strengthening efforts to improve company-wide operational efficiency through active utilization of DX and other initiatives.
Let me move on to the last page 18, other topics. We have published the integrated annual report 2023 as well as the sustainability data book 2023 on October 20. I hope you can refer to them for your reference. Also, we are planning to conduct the IR Technical Briefing again this year on November 29 to facilitate deeper understanding of our business. I hope you can participate in this meeting. That is all. This concludes my presentation.
Question-and-Answer Session
Yoshitake Kobayashi
We will now like to go into the Q&A session. We will take your questions in the following way. On the Webex screen, there is a participant icon. And when you click it, you will see a raise your hand button, so please click that. When you are designated by the emcee, you will see the unmute selection button, so please unmute and speak. Please name your company and your name and then ask your question.
Furthermore, we would like to take as many questions as possible. Therefore, we would like to limit the questions to one. If there is time left at the end, we will receive your questions until the scheduled time to end this session. We are streaming with Japanese and English simultaneous interpretation. So we kindly ask you to ask your questions at a moderate speed and in a concise manner. And we kindly ask you to refrain from asking questions related to individual companies. We will accept questions in both Japanese and English. Now we would like to open the floor for questions.
From CLSA Securities, Mr. Yoshida, please go ahead.
Yu Yoshida
Hello. At this time, FY '23 were the product mix change that had an impact towards the downward trend of the profit, but the SoC tester and memory testers and market outlook towards FY 2024. And if possible, do you think about the product mix? I would like to receive your comment including the breakdown of it as well. And also for memory towards HBM, if we just extract that part this year and next year as a way of thinking also if you can share that, will be appreciated?
Yoshitake Kobayashi
Nakahara in charge of sales would like to answer that question.
Makoto Nakahara
As we have explained previously, this fiscal year in terms of profit, we are struggling. We have an outlook for the next year’s market situation is that the high performance device is probably going to lead the market. That is because as mentioned in the presentation as well, generative AI related device that we are focusing on, this is a high performance and it is a large size logic device and that requires a high performance memory HBM. And the generative AI HBM number per generative AI is going to increase. So it is a challenging and difficult device. However, for us in terms of product mix on the memory side and also on the SoC side, we believe this situation is going to work in the positive direction for us. So we have high expectations towards this. Does this answer your questions?
Yu Yoshida
By the way regarding market size from this year’s perspective, is it going to increase? If you have a certain image, can you share that as well?
Yoshiaki Yoshida
This is Yoshida speaking. As of July or even before that, FY 2023 is the bottom here and we believe that that was going to recover towards 2024. And for that, we were expecting to recover to the level of FY 2022. However, this time, FY 2023, the degree of recovery is under what we have underperformed -- under what we have expected is this about six to nine months behind what we have expected. Therefore, whether we are going to immediately recover in the new fiscal year or not, in the calendar basis, I think that is going to be in the latter half of the year. Of course, we haven't given up but the number up till 2022 to come -- for the numbers to recover to that level, we did have the sense of crisis. Maybe that is going to be difficult [indiscernible] size wise, up to FY 2022, maybe we will not reach that as we have had expectations originally, but we believe that we will recover in a gradual pace.
Yu Yoshida
Thank you very much.
Yoshiaki Yoshida
Thank you very much. The next question, please.
Yoshitake Kobayashi
Mitsubishi UFJ, Tetsuya Wadaki, please.
Tetsuya Wadaki
Wadaki over at Mitsubishi UFJ. Thank you very much. This time around looking at the results, the gross profit has been on a major reduction, ¥20 billion and a 5% reduction. How much impact is coming from the product mix change and what about the impacts of raw material sourcing? Three months ago, we didn't have clarity on that. Was that the case?
Yoshiaki Yoshida
Thank you very much for your question. We made our announcement back in July, and this time around we updated it and you find ¥1 billion reduction in gross profit. And what is the background to do that? What sort of factors led to the duration of gross profit? That was your question as I understand it. Based up on our exercise estimates, idea, as we discussed in the briefing, the presentation, SoC tester [indiscernible], sales was down by 20 billion in memory, sales of [indiscernible] compensating by 6.6 billion. So this product mix impact is approximately 35% in terms of gross profit margin. And on top of that, the sales itself has come down. And because of that profitability, margin has declined by 25%. And FX impact -- operation cost impact is about 10%. And our processing cost has increased and it's about 25%. That is the breakdown of the impact. You're talking about operating income. I'm talking about operating income, but I was saying gross profit. My apologies.
Tetsuya Wadaki
Gross profit of 5% decline has happened this time. So could you talk about the breakdown of that?
Yasuo Mihashi
If you could wait a moment. So the reason why our gross profit has come down by 5%, I cannot discuss numbers. In principle as these sales came down and also product mix, as I mentioned earlier, ¥20 billion for SoC tester reduction and memory tester was up by 6 billion, so highly profitable 93,000 in sales was down by 20 billion. And in memory, on the preprocessing facilities and equipment, sales has been a major burden. There has been a deterioration of profitability. That is why gross profit margin was down from 55% to 50%. So the gross profit number has changed so much. I believe that because of the material costs and processing costs and so on. But why couldn't you grasp this three months back when you made the previous announcement? So, FX, the beginning of ¥135 was the assumption, but it became ¥145 and because of the FX situation and looking at the U.S. interest rate, this time around our planned rate has become ¥145. So, now the yen is a weakening only because of our processing costs has been partially impacted. That is for sure. Currency sensitivity, I believe it is natural that the gross profit comes down because of the yen weakening. So because of the yen depreciation, improvement is happening but we have a processing cost in the operating income, we talked about 25% impact, and this is also impacting the gross profit margin more than I thought but the sales should go up even more. So there is currency sensitivity working positively full year because of the yen depreciation. So I wonder why this is happening around gross profit.
Yoshiaki Yoshida
This is Yoshida speaking. Well, of course, cost itself has been going up more than we had expected. The impact is larger than what we had anticipated. So our time is limited. So I don't want to take up too much of your time. You could respond to this question later on. But gross profit when it comes to that, maybe there is some impact as the currency exchange perhaps parts [ph] close to increase and so on. So I wonder if you could elaborate on this.
Yoshitake Kobayashi
So I will contact you later. I am Kobayashi from IR. Part of the sales details I cannot really disclose on the foreign currency [indiscernible] been high. But the ratio of that has come down. And what about yen denominated sales as far as the -- some of the parts are sourced in a foreign currency dollar. So that's impacted us.
Tetsuya Wadaki
Okay, thank you very much. In the interest of time, I'll finish my question and I'd appreciate your explanation later on. Thank you.
Yoshitake Kobayashi
Next question, Mikio Hirakawa from BofA Securities.
Mikio Hirakawa
Thank you very much. This is Hirakawa from BofA Securities. Thank you for asking the questions. So having heard your presentation about the HBM, a new business opportunity in that area, currently, I want to know what your market share is in the business and gross margin. So if this business grows, what will be the impact on the overall gross margin? And within memory tester business, currently, how much of that is coming from this new business and what is expected for next year?
Yoshitake Kobayashi
We’ll have Doug to answer this question.
Douglas Lefever
Okay, thank you for the question. As far as HBM goes, we can't disclose our market share exactly, but I think it's fair to say that we're in a very good position. And we've worked very hard at this market for a long time to establish this position. As we go into next year, we're going to continue to see very large growth in HBM and we should benefit from our market position as there's other entrants into HBM. From a gross profit standpoint, we also expect to have improved margins as we continue to improve our marginal profit levels and other actions we're taking.
Mikio Hirakawa
And let me ask you a follow-up question about the gross margin. I have a concern. Like mentioned earlier in a frontend process, the wafer testing won't be coming in, right? So for the margin perspective may not be as strong, I am concerned about it. Would that be -- that's not true, still being a high value added process. So overall gross margin level, you can sustain high level gross margin either correct or not?
Douglas Lefever
Yes, I'll say for sure, the wafer test margins also are usually squeezed more than the final test. But our actions for margin improvement should cross both the wafer stored as well as the final test. And as these stacks become larger as we go into HBM3 and beyond, there's going to be even additional yield requirements in order to have a final yield level that is acceptable for the end stack.
Mikio Hirakawa
Thank you very much.
Yoshiaki Yoshida
And let me also elaborate. For HBM major manufacturers, we are positioned in a good -- we have a very good position. I think you may already know this. They are doubling or tripling the capacity into next year. So that's what they have. So that's what they're planning. So a large amount of volume is expected from next year beyond. So such a mass production impact or benefit as well to improve margin as a manufacturer, we have been brushing up our way of improving margins, so we can benefit from that to improve the margin. So we will see a substantial improvement in margin profitability from the current level.
Mikio Hirakawa
That's all for me. Thank you very much.
Yoshiaki Yoshida
We would like to take the next question.
Yoshitake Kobayashi
Tokai Tokyo, Masahiko Ishino. Mr. Ishino, please go ahead.
Masahiko Ishino
Yes, I can hear you. Thank you very much. Regarding the previous question HBM3 with your existing tester, I think the speed is not enough and also other companies have already made the announcement of their roadmap. And according to that, it seems that you will have to launch many products. And previously, you were saying that the manufacturing sides are HBM manufacturing capacity. In order to respond to that, are you having a sufficient level of manufacturing capacity or not is what I would like to know? Thank you.
Yoshitake Kobayashi
Nakahara will like to answer that question.
Makoto Nakahara
Regarding HBM, the HBM or HBM2 or HBM3 that are launched in the market already, recommending those devices the testers that we are currently holding the coverage is sufficient with existing products. So they are being used. And as you have pointed out, next year and onwards, HBM will evolve and HBM4 will be coming out. The interface speed will be increasing as well. And in order to respond to that, our testers need to evolve as well. Therefore, within our daily R&D work, we are continuing the work to respond to that. So at the timing that such tester is required. We are proceeding so that we will be able to respond that at a timely manner. And in regards to the production capacity, probably your question is related to how much is -- because you know very well how much of the volume is increasing on the HBM side in advance and quite a long time perspective, we have received a long-term forecast, and we do -- that is because we have a long-term relationship with them. And this year already, we have increased our manufacturing capacity. However, as we go into next year to get in line with our customer side capacity increase, we are working on to increase our capacity as well. Thank you very much.
Masahiko Ishino
Thank you very much.
Yoshiaki Yoshida
Thank you very much. Next question, please.
Yoshitake Kobayashi
Goldman Sachs, Shuhei Nakamura please.
Shuhei Nakamura
Thank you very much. The share SoC tester, memory tester, so each in the previous explanation, it is something like 55% to 60%. This year and next year, what is your outlook about the share?
Douglas Lefever
Market share for us we're forecasting to remain roughly flat year-over-year. So that's a simple explanation.
Shuhei Nakamura
So from FY '23 through '24, it will be flat.
Douglas Lefever
Yes, we expect to remain roughly flat. The market share in ATE fluctuates greatly with respect to single customers in some years. And so we expect year-over-year to remain at 50% to 55% collectively between SoC and memory.
Shuhei Nakamura
Thank you. Going forward there will be more applications around generative AI. And as a result, what about the evolution of that share?
Douglas Lefever
Yes, so we're very well positioned in HPC, high-performance compute, specifically in AI at all of our traditional HPC customers as well as many of the system and hyperscalars and also some of the ASIC providers. So we're very comfortable with our leadership position there. And so you're correct. As those applications increase in volume, we expect to continue to do well in that area.
Shuhei Nakamura
Thank you.
Yoshitake Kobayashi
Next question, Masahiro Nakanomyo from Jefferies Securities.
Masahiro Nakanomyo
Can you hear me okay?
Yoshitake Kobayashi
Yes.
Masahiro Nakanomyo
Thank you. I know it can be quite simple question, but the HBM was brought up earlier [indiscernible] so we first have to see the utilization improvement, all the existing tester to see an order recovery before the new orders. Would that be applicable for HBM because the new construction of factories, so HBM are separate? Are they moving already separately from the existing testers? Which is it?
Yoshiaki Yoshida
Okay. So let me answer that question. So that device is [indiscernible]. They're basically that large logic devices and also surrounding HBM’s memory devices. So there are those two groups. And the business models are slightly different between those two groups. As mentioned in the results briefing, so that excess equipment in the market are not fully utilized at this point yet. SoC tester chips are not growing as much as we expected. Logic device manufacturers are using the OSAT business model foundry. OSAT business model in 2021 and '22, OSAT had a huge install base with our SoC tester smartphone application processor, so they built such a huge install base and they saw the drop in the utilization. So now, with the AI device customers by making a change in the configuration, they are using these equipments right now. So they haven't got to the full utilization yet at this point. So there's a still delay compared to the expectation. But HBM, this is the business model customers, so they had a plan and they purchased equipment when they need them. So ahead of the SoC tester demand, we already see our recovery in demand in HBM. So foundry majors for 2.5D/3D package, I think they have made plans for them. So they are procuring new testers for them. So for logic devices for generative AI is exactly as mentioned that this is a cutting edge processes, a cutting edge packaging needed; 2D, 3D or chiplets. And those are the packaging method. Our testers themselves aren’t SoC testers are utilized, while such advanced technologies. In addition for the next generation testers, 3,000 and we had [indiscernible] and we can use them to be more efficient in testing. So we can optimize testing and we're making those proposals as well. So tester themselves -- our SoC testers are able to accommodate those new technologies in the future after packaging process, their measurements taking place after packaging, and trying to measure the devices that with a memory together there will be new measuring methodology that we have to discuss and come up with by talking to the customers. The current testing needed at the wafer level insertion being increased to improve the quality. That's the method they're taking to accommodate. Thank you very much.
Masahiro Nakanomyo
I also wanted to ask about whether it's HBM or whether it's 2.5 or 3D packaging logic foundries, so some specific new demand can be created. Can you identify them or how big the demand could grow into, can you tell?
Yoshiaki Yoshida
It will be difficult to identify. The last question -- sorry, can you repeat your question once again, sorry.
Masahiro Nakanomyo
For HBM, I think you already know the plan. The customer will be making an investment according to the plan that HBM driven new tester demand. How can you quantify that business opportunity?
Yoshiaki Yoshida
Let me confirm, so HBM or high performance computing related HBM, so for memory?
Masahiro Nakanomyo
Yes, the memory HBM. And the same question for [indiscernible]. If there are new lines being built, any additional -- instead of utilizing existing equipment, there could be a demand for new equipments and how big the potential will be? That's what I wanted to ask.
Yoshiaki Yoshida
For HBM, as you know, DRAMs could be 8 to 16 DRAMs, could be stacked or layers, then logic surface interface, logic controller will be implemented at the very bottom to create HBM. The biggest driver for demand will be the capacity for overall HBM, it will be larger. So the customers -- those layer DRAMs, including logic, this whole thing will need to be tested as HBM. So the huge capacity that makes the patterns [ph] to be much more -- I guess time to be larger, these patterns to be also larger and that time takes longer. So that's why investment into HBM is expected to grow drastically. And HBM itself along with the growth of AI service, we expect a 30% to 40% per annum growth expected for HBM along with the capacity increase, the necessary tester would be identified.
Masahiro Nakanomyo
Also the amount in the yen terms, can you quantify the increase? What do you mean by yen amount? That's the tester market or --
Yoshiaki Yoshida
On a shipment basis, currently, assuming what we're seeing from order taking, we cannot get rough image what the size would be. But HBM3 towards HBM4, the capacity is going to be increased from eight pieces up to 16 layers. So we need to consider that. So let me get back to by looking into them thoroughly.
Masahiro Nakanomyo
Okay, understand Thank you very much.
Yoshiaki Yoshida
Let me elaborate or add one more thing. So the mobile phones -- the testers prepared for those mobile phones and others that could be used for AI. Right now we are in a transition at this point. But the other restraints we see will be at the foundry side. If they are really producing the volume that is really needed by the demand, the Taiwanese foundry saying that for next year and beyond that they have a plan to continue to increasing the capacities. If that increases the total volume, the utilization is going to recover rapidly and that will lead to a new tester demand. And it's going to be -- we're hoping that could happen in early next year.
Masahiro Nakanomyo
Thank you very much.
Yoshitake Kobayashi
We are running out of schedule time. So the last question from Macquarie Capital Securities, Mr. Damian Thong, please go ahead.
Damian Thong
Hello. Can you hear me? This is Damian from Macquarie Securities. You are reexamining the numbers. How should we -- for your company's SoC is low, but it's a new area. Sorry, the audio is quite bad. The next fiscal year’s outlook, what is it?
Yoshiaki Yoshida
So your question is the system level test business outlook. Was that your question?
Damian Thong
Yes.
Yoshiaki Yoshida
Okay, understood. Doug, would you like to answer?
Douglas Lefever
It has been to address a lot of the high volume mobility market. So to that extent, there's still softness for our traditional SLT business. What we do see now is with the growth in HPC and AI, and as those volumes increase, we expect to see more demand for HPC level SLT, and that's something new. It still will be small for the foreseeable future compared against the traditional mobility SLT. But long term, we see that as a growth market. We also see some use in ADAS for SLT. And as that continues to grow, we expect to see some opportunities for SLT in that space as well. But for the next year, roughly, we expect things to be relatively flat until consumer and mobility comes back.
Damian Thong
[Technical Difficulty]
Yoshiaki Yoshida
Damian, I'm sorry to say but your voice is really breaking. Could you repeat your question again?
Damian Thong
I'm sorry. [Technical Difficulty].
Douglas Lefever
So I think the question is whether HPC market for SLT can grow to the size of mobility eventually?
Damian Thong
That's right. Thank you.
Douglas Lefever
So the answer I believe is yes, it is just going to take time. One of the key aspects of SLT is that in order to take advantage of automation, you have to have high volumes and large test times. And so right now, as the applications are growing, there's a lot of use of native boards for SLT. But as those volumes cross a threshold, we'll see the use of automation to optimize equipment for SLT. The other factor is as there's things like training, model workloads that want to be used to fully do SLT that's going to put additional requirements for that insertion to guarantee good yield before those devices get shipped.
Damian Thong
Thank you very much. I look forward to it. Thank you.
Douglas Lefever
Thank you.
Yoshitake Kobayashi
Thank you very much. I believe that we have received more questions. But with this, we would like to conclude today's session. Thank you very much for attending Advantest FY 2023 second quarter financial results briefing despite your busy schedules. Thank you.
For further details see:
Advantest Corporation (ATEYY) Q2 2023 Earnings Call Transcript