Advantest Corporation (6857) Earnings Call Transcript & Summary

Thank you very much for joining Advantest Corporation's IR Fireside Chat despite your busy schedule. We are excited to present this session and would like to introduce our panelists and our moderator. From the left side of the slide, Mr. Douglas Lefever, Representative Director, Senior Executive Officer and Group CEO; Mr. Juergen Serrer, Senior Executive Officer and Chief Technology Officer; Mr. Yasuo Mihashi, Senior Executive Officer and Chief Stakeholder Relations Officer. Mr. Mihashi will serve as the moderator for today's fireside chat session. I am Oike from the IR department, and I will be serving as your facilitator today. The theme of the discussion is be the most trusted and valued test solution company in the semiconductor value chain. This is Advantest's vision statement, which we defined last year. Following the fireside chat, we will open the floor for questions from the audience. We plan to close this session at 6 p.m. Japan Time. Before we begin, we would like to remind you that today's session contains forward-looking statements, all of which are subject to risks and uncertainties that may cause our actual results to be different from those in such forward-looking statements.

Good morning, good afternoon and good evening. Thank you for joining Advantest Fireside Chat today. Let me introduce my -- or our CEO, Douglas Lefever, sitting beside of me; and also our CTO, Juergen Serrer. We will conduct today's fireside with all of us three.

Yes. Thank you, Yas. Great to be here. We're in a beautiful setting today. It's autumn here in Germany, and we're amongst the Mercedes-Benz area where many things started in the automotive industry. Myself, I came from a family in the automotive industry, not in Germany, but in the Detroit area. And that's how I started my career, but I quickly moved to semiconductors, and I have been in that industry now for 27, 28 years now. So happy to be here. Yes.

Hello, I'm Juergen. Here, Boeblingen was the first site of Hewlett Packard outside of United States. And I'm now 34 years in the ATE industry, Hewlett Packard, Agilent, Verigy, and since 2011 with Advantest. And yes, I'm happy that now the entire team is still in Boeblingen, and they can show everything what we are doing.

Let me start touching about our MTP3, I think. So it's been 1.5 years after we initiated this, the mid-term plan 3. I think kind of maybe we should start touching about what is the highlights about MTP3. And we still have 1.5 years more, and we will focus to expand or to extend our businesses. Doug?

Yes, I can start, Yas. But maybe before we even talk about the mid-term plan 3, we can just go back even a couple of years further. When we started talking about mid-term management plans, we did that around 2017 when Yoshida-san became CEO, we implemented these 3-year management plans, realizing that we needed to start planning not just 1 year out but take into account all the cycle effects. And so we started putting some strategies in place along with some key performance indicators to measure ourselves over a 3-year period rather than single years. And so we did that from '18 to '20 and then '21 to '23. And now we're in our third 3-year tranche and halfway through that. And each of those plants has been very successful. I think we've built new strategies upon ones that are still core to our success. And so as we look into the second half of this third plan, we're doing a lot of things right. And I think -- and the company has executed in a way that the management really was expecting, and we've been able to grow from plan to plan at a very high rate. So as we look into this third plan, a couple of the themes now that we're having to adapt even more is around the concept of velocity and scale in our industry, that the time-to-market aspect is becoming more and more critical. The device generations are at a cadence of 1 year now as opposed to a couple of years or even longer in some cases in the past. And so we're having to shorten our development cycles. I'm sure Juergen can speak to that. But then we're also having to scale out at a level in our production that we never had assumed in the previous plans. And so we've done a good job. I think in the first part of this mid-term plan, and the second half is going to be a lot more of that scale, investing in our supply chain, investing in our production capacity. And as we reported in our last disclosure, we've upped the full year numbers, and we've updated the 3-year plan numbers now significantly, and that's a result of all the growth that we're experiencing along with our ability to respond to that growth.

I see. So as you know, we are doing many things, where -- it could lead our business to be -- or our position to be extended to a success, I think. But our success coming from its complexities, that's what we are talking in the market, I think. So what you can say about these complexities?

Well, it's a complex question, I've to think correct. So the AI market is enormously driven by performance and power. So all our customers are pushing all technologies to the limits, whether that's high bandwidth memory now becoming very fast, even more stacks, whether that's the I/O speed going 224 gigabit, now moving to optical integration, whether that's new process technology, whether that's 3.5 D packaging technology. And this is all stacking up and combined with what Doug said that velocity, the time to market pressure and the increased quality levels, this comes altogether and generates an enormous test challenge or an industry challenge. We call it industry problem. And that's where we are focusing it and where we are investing to resolve it.

So used to be testing has been caught as a cost because we're going to need to judge whether a device is okay or not. But now testing becoming something different than where we are, I think. And I think we are showing more volumes than what original testing is, I think. So is this trend that will be kept for at least in the next few years?

Yes. I'm very sure because the technology which is used, they always move to the edge. So our job in the meantime is not just no, no go, but we have to trim and calibrate every device, sometimes even subsections of every device to its optimal performance point, and having enough margins to the individual circuit. And that's not something we are here doing since many years, and it will continue that we trim and calibrate, that we make device better and that's a necessary step.

Yes. I would just add, yes, agree with Juergen. And kind of the equation has changed a little bit now with these very complex high-performance computing, AI devices and the packaging technology. And the cost of that packaging technology now is driving a focus on yield and quality. And that -- there's always that aspect in the past, but now these devices have such high ASPs and they're so costly in terms of the packaging, that the test content naturally is going up to make sure they're getting high-yield devices because it's just too expensive to throw away these expensive packages now.

I mean, another example is power pairing. So the process variation is so such -- and everybody wants to squeeze out the last piece of power. People want to pair chiplets in regard to best combination for power. And that's where we help. That's one of the areas where we are really good to provide that data to optimize the yield and the performance level of the bins.

It is really good for us, I think, test becoming so much of a value on the way to produce very complicated devices, including miniaturization technologies and heterogeneously integrated technologies. But, meanwhile, efficiency is always a requirement, is being required on testing, I think. So what is our approach to efficiency gain with such a complicated testing philosophy or testing methodologies in the future?

I mean let's start with a couple of things. And first of all, we are convinced that you have to extend towards a design because that is where the DFTs are being set. So that's part of the reason why we engage in our new initiative for our new initiative or product called SiConic, so close -- to close the gap between EDA and ATE, to prepare, to have a good sign off, to also allow that the designers can talk to the test engineers, test engineers can move the results back much better and that flow is closed. Second, we certainly invest tremendously into automation tooling, ATE software, AI tools ourselves to make our users much more productive. And third, and I think that's what we call as distribution of test. Sometimes, our customers don't know where to put what content and what content they have to add. And there will be -- we have a strategy of allowing the distribution of tests along all test insertions, whether that's engineering at wafer sort, hot/cold package test, burn in, system level test. So that's just a few highlights.

Yes, I would -- the other thing to add is I think in -- on the previous mid-term plans, we've done a good job of adding some of those pieces in that back-end flow. So to Juergen's point, as the distribution of test moves around depending upon the type of device or where it is in its life cycle, our approach is being able to cover those different insertions. And so I can remember actually -- just going back to the 2017 time frame, I still have a picture of the whiteboard that we made of this kind of moving to the left and integrating EDA to ATE and going right with system-level test and having cloud-based data analytics to tie them together. And that was all the way back now 8 or 9 years ago. And throughout that first and second plan, we've made some inorganic moves and acquisitions. We've done some organic development on things like die-level probe. And now we're putting those things together, and we're seeing the effects of those investments and strategies.

Yes. Indeed, we are making good kind of progress, I think. So I think my next question is since we've been doing great, I think. But meanwhile, we must consider where the risk is hidden so that we better do consider some risk factor on the way over our company growth and our company strategies. Juergen, where you see some risks we should think about. So we are doing right on our production capacity up, which is one of the biggest risks what we are having, so that we are working in. So what are the risks?

Yes, I get this question a lot, because everyone want to talk about bubble and what happens in a downturn. And I think this is -- for one thing, I believe this is different. In some ways, it's just a bigger wave. In some ways, it's similar to what we've seen in the past, where initially with a wave of technology change, there's initially an infrastructure or enterprise level growth and that typically gets followed up with something more at the consumer level or in this case, people talk about the edge. And I firmly believe that, that will happen, the timing of when the handoff from enterprise to edge is still a question, I think, that's debatable but not about if it's going to happen, it's about when. And so we're fairly comfortable in that, that will happen. And those volumes could be orders of magnitude higher than what we're seeing at the data center level. So we have high confidence that the future will kind of track with what we've seen historically. That being said, we've done a couple of things, and maybe Juergen can talk about maybe the platform approach, which the company has really been disciplined about, that helps us through kind of the ebb and flows of the industry. But from a structure standpoint, we're being very disciplined in our cost, particularly in our fixed cost level. So I think we're pretty well prepared for any kind of change in the market temporarily. And then the last thing I'll say is that while AI and HPC have driven the market, our portfolio solutions can adapt to wherever there is good segments and -- especially when the consumer comes back, we have a very, very robust portfolio of test and measurement products to address those segments.

Yes. And let me maybe interject here. So I think the risk of capacity, I think, is even an opportunity also. So meaning we have been investing so much in our flexible supply chain so that we can scale and handle the uncertainty that we believe that our capacity or manufacturing and supply chain in the meantime is the competitive advantage.

Sure. And in the materials on the way of revised MTP3, we stated we will increase our R&D expenditures because so many different technologies is coming up and so many complex test flow so that we need to build something really surprising in the market, I guess. So CTO, you're having so many plans, I think.

Yes. Well, I think that's the nature of AI right now, that everything is so fast. They are various technology options the market may take. And so it's answered. So what you have to do is, I think, multiple things. First of all, you have to find out what's coming, what's not coming. And we have established a really tight relationship to industry experts, customers but also to leading partners to find it out. And if you need to develop, a new technology may take 5-plus years. If you start a product, it may take 3 years. If it's a software enhancement, 0.5 year. So the most critical one is the ones where are far out and you need a long lead time. So our ability to predict trends, I think, is really good, and we have a good hit rate. And then you need to invest into the different technology options. Third, that's more the scaling aspect, as Doug mentioned. You also have to then think if you identified the industry product, how to map that on our platform in a scaling way. So that at the end, everything has to fit together and become effective. That's usually the second step that we build scalable solutions which are compatible to what's already installed. That's a big value for our customers so they can flexibly shift capacity, and we use the assets, but also allow that platform to be extended and scaled to future needs.

Yes. I would add to that, too. I'm always challenging Juergen on the R&D because I think more right now is better. But the other thing we can't do it all ourselves is another point, especially now we have to approach things not from just a tester, but a test cell perspective. And so we've engaged with very critical partners in the industry. And a lot of times, we are kind of co-investing with those partners to develop that full test cell solution. And so sometimes, I view that as R&D dollars or yen or euro in those cases, that doesn't necessarily show up in our direct R&D, but it's certainly focused on development of critical solutions to address these industry problems that Juergen spoke about.

Yes. So I think we are really extending through your CTO networks, and Dough, your networks, on how we can expand the area and the technologies that needs to be combined with all the technology which is coming, I think. But internally, our operation is getting bigger along with our business size. So I think we have some challenges internally too. How we should try to further engage? So from internal CTO viewpoint, what are you focusing?

A couple of areas. First of all, we are continuously investing into what we call refreshing the platform and the architecture. And those things take a long time, so you have step-by-step continuously invest. Second is we run programs that we can run now our projects for V93000 in various R&D labs. I mean it started all here in Germany, but now we also -- we have extended that to China, to Japan, and we're also now starting some projects together in U.S. So this gives us the flexibility to allocate resources. And combine things and leverage and get these synergies out of it. And next is most likely also -- most likely is investment into more software, automation and AI to make us internally more productive. That's not only on software area, but also on ASIC design, thermal simulations, mechanical simulations, and those are just a few points.

Yes, having those, we have very large R&D hubs here in Boeblingen and over in Japan. Those are our main, our 2 big hubs. But I think what we'll see in the future is, some level of local R&D, specifically placed around customers because the solutions are so complex that there has to be some touch of locality to help implement new solutions. And so I see us heading in that direction, into the next couple of years.

So next question is, it's been, I think, 14 years after company Verigy and then company Advantest has been together, I think. So there are lots of -- many discussions, I think, we've been through, and now we have really became big parties, one big family now. So on our way or from here to future, I think, so we really need to think about cultures inside of the companies, how we can -- how we should express as a management of our company cultures. So we -- you in as CTO taking care almost all BUs, so Juergen, you start, I think.

Well, I think we have this nice culture of teamwork and collaboration. And this is important because people need to talk, not just within R&D, but to our salespeople, to our application engineers to really find out what is important or not. And this is very important in this time to market at least. Second is this is also a collaboration with partners, that we are openly discussing what are the industry problems coming, and then turn it into, let's say, solutions. And as I mentioned, part of the solution to address this complexity is the distribution of test. So now we have to address that the different BUs or different products are linked together. Therefore, this collaboration or skill or culture of collaboration and teamwork gets more and more important, not just outside, but also inside. And this paired with curiosity, the passion, I think that's, I think, a unique characteristic of Advantest.

So I think, we have initiated our activities or our philosophy, integrity specifically. So it was -- I think we have communicated this in about 2018. And now it's been becoming an almost more than 5 years, and we have been expressing at how it's important to integrity is. This activity along with our daily communication operations, was it helping our commonality of what we care the most?

Well, I mean, the result show it, right?

Yes. I think this -- the concept that we started with integrity, each letter of integrity has its own meeting that we focus on for different parts of the company, is super important to Advantest. I mean our reputation is a company of high integrity, and our vision is to be the most trusted and valued suppliers. And I think we're living up to that. We're not perfect, but it's not because we don't ever have high integrity. I feel confident in saying that our culture is very well established, and it's something that we protect and hold very dear to us. I mean, we're a 71-year-old company that now is a blend of Japan, who has this very long-term view, very service oriented, excellent engineering technology, the German precision capabilities and approach to platform discipline and the teamwork and then the Cowboys from the U.S. But we've -- honestly, we pulled this off as a company, a lot of people would have doubted. We've been able to take that -- the best parts of those 3 cultures and then you add into it the Southeast Asia, Taiwan, Korea and other parts of Europe. I think we've done an excellent job in the company of taking those different cultures and putting them together inside of the Advantest culture, again, which is built around this concept of integrity.

I can see. So, Doug, you've been a CEO like 1.5 years, I think. So I think what you said -- I think is supported after you became a CEO, I think. So our company passed more than 70 years, I think. So there are many Japanese culture-based philosophies, I think. So now 1.5 years and more in futures, I think. So I think we're really being diverse, like what you said, I think. But how we should expand our company cultures to be more aggressive or more humble, I think. Any comment?

Yes. Well, it has been 1.5 years, a little less hair and a few more kilos I'm sure, but I have enjoyed it. And one of the best parts of the job is going to the different offices around the world and spending time with our employees around the world. And for sure, the world -- it's not just device complexity. We often talk about the world that's more complex now in so many ways. And so, to have a network within Advantest that can communicate clearly and be on the same page strategically is not an easy undertaking. And so we spend a lot of time in our management team, and we have this really wonderful I would say, 3-way matrix where we have our business units, which are driving the P&L, we have the functional groups that are supporting the different functions, service support, sales, administration, operations, these things. And then we have sort of a third layer, which is our regional-based entities that then all cooperate together and gives us the local blend on top of those business unit and functional unit matrices. And so I'm sure other companies have tried to do this, but I think it for us, it's really worked quite well to address how we scale if the business moves in one region versus the other, depending upon geopolitics. Our footprint is already in, I think, 16 or 17 countries. So it's easy for us to scale depending upon where the business ends up happening. So that's something I think we'll see more of, and we should be proud of, I think, that 3-way matrix that's really led to a lot of success.

Okay. Let me touch again about our automation of test, I think. So we are -- or what we need to do aiming for in future about these automations, Juergen?

Well, as I had already said, things become much more complex and the time-to-market pressure is continuing to dominate everything. We have to enable what we call automation or an entire flow. And that's where we are working on. And this also is our internal R&D, but also working with partners to have all pieces lined up and interacting efficiently. I think that's the main direction. And this means we need to work with EDA companies to prepare the front end, the test content, then we need to -- our solutions need to be able to efficiently manage and control the test content and move content around to the different insertions. And certainly, the results we need to feedback, to process, to design for improvement. This also includes test cells. I strongly believe that die level probing will need to come. The thermal is one of the key challenges in the industry. So you also have to think about the test cell spec, whether that's wafer wort or whether that's in a final test, if not kilowatts of the thermal control, vast thermal control, but also in SLT, the economics as a last quality gate. All those pieces have to interact and allow our customers to achieve the time to market, the quality levels and the productivity. So this is the main direction that we are heading.

Yes. I would just add. If you look at our history now going back again to the early mid-term planning, we basically been marching toward this ultimate goal. So we took the decision in 2017, 2018 to really double down on semiconductor tests. We felt like that was going to be a place there's enough there for us to focus on that. And then as we went forward, we picked up different pieces, the solutions. Right now, we've been integrating and enabling this automation of solutions. And ultimately, there'll be this back-end flow, and it'll be different for different devices and even different customers. But at that point, Advantest will have successfully put this capability in place where we can enable this automation. And so ultimately, the goal of the company is to really be able to provide this enablement of the back-end flow that is so much more complicated now than it was just 5 years ago.

Yes, I think -- well, we keep getting complex. It's been so compressed. So I think this complexity is really the thing -- well, it's going to keep driving our activities. But also, this is one of the biggest hurdles for our customers. So that we keep focusing on this kind of complexed semiconductor, I mean, technologies and the industries. Well, it was a good communications and any insight and inputs. Thank you so much. And I think we now moving to Q&A, but Dough, any other comment at the end?

Yes. I'll just say that there's never been a better time in our careers for test. And we've lived through some -- so we've seen lots of changes and ups and downs. And we never fully expected things to get to where they are already now. But the company has just been investing continuously for this time. And like I said, there's -- we're kind of in this period of accelerated growth where test is just going to continue to be a value-added part of the semiconductor value chain, and we feel positioned going forward as a company.

Let's now move to questions and Q&A.

Goldman Sachs Japan, Mr. Nakamura. [Foreign Language]

Thank you very much for taking my question and thank you very much for a wonderful session today. My question is regarding to the die-level probing. I think there were some remarks made several times during this session. And I just wanted to understand how you're looking at this opportunity going forward. We understand that the next device of the merchant GPU may potentially start to adopt the die-level prober. So I wanted to see when you would potentially see the contribution in terms of earnings and how we should quantify that impact?

[Foreign Language].

Okay. Yes. Thank you, Nakamura-san, good afternoon. Thank you for the question. So I'll go first and maybe between Juergen and I, we can expand on the die-level probe opportunity. We definitely -- from a high-level standpoint, we definitely are seeing traction built for die-level probe insertion for known good die and for known good CoW use. And so this is a very dynamic area because it's different. And the value proposition seems like it started to be fairly compelling around this insertion. In terms of timing, we see kind of later in next year, really there to be some definite ramping potential for this business, followed by several years of pretty high growth. And I'll just put a condition on that, that again the use case and the value add for this insertion has to be very clear for it to have that kind of aggressive ramp.

[Foreign Language]

Well, so let me give you some background why we believe that this is an opportunity for us. The power levels at the wafer test are so high that thermal control and warpage at the wafer level -- the entire wafer level is too high and difficult to manage. So people really want to do real known good die and want to do matching of chiplets later on. So they need to perform very precise and accurate known good die test at hot temperature.

[Foreign Language]