Okay, good afternoon and good morning, everyone. Welcome to Accton's 2025 Investor Conference Call. My name is Derrick Yang. I'm the Tech Hardware Analyst at Morgan Stanley. Accton is a leading networking company here in Taiwan, and has a very strong position in the AI deployment globally. So it's our honor to co-host this call with Accton's senior management. With us today, we have the President, Jackal Lee, Vice President of R&D, Michael Lee, CFO, Fanny Chen, last but not least, Spokesman Edward Lin. So for today's agenda, Accton's management will go through the financial highlights as well as company strategies and the technology dynamics. After that, we will have a Q&A session. So without further ado, let me pass it to Edward to kick off the call. [Foreign language] 好，谢谢，谢谢 Derrick, Derrick, 各位投资人、媒体朋友，大家午安。我是智邦科技的智理长及发言人 Edward Lin。感谢各位参加本公司 2025 年法人说明会。本次法说会是与全球投资人连线，所以我们会全程使用英文。感谢各位的合作。 Good afternoon, ladies and gentlemen. Welcome to Accton 2025 Investor Conference. This is Edward Lin, Spokesman of Accton. We are hosting today's conference with the support of Morgan Stanley. And so for the global investor consideration, today's presentation and discussion will proceed in English. Thank you for your cooperation. Next. Before our presentation, we would like to remind everybody today's discussion may contain the forward-looking statements that may be subject to the significant risk and exposure, and they may cause actual results different from those contained in such statements. So please make sure you fully acknowledge the safe harbor notice here. Thank you. Next. Here we have presenters include our President, Jackal Lee, our VP of R&D, Michael Lee, our CFO, Fanny Chen, and myself. Next. Today's agenda will include the financial highlights and company highlights and technology highlights and Q&A section. Next. Now we'd like to turn this call to our CFO, Fanny, please. Thank you, Ed. Good afternoon, everyone. Let's start with the financial highlights for the third quarter 2025. The third quarter revenue was TWD 72.94 billion. As compared with the Q2, growth rate is 20%. Three quarters revenue was TWD 176.3 billion and 147% growth year-over-year. The third quarter gross margin was 17.2%, 0.7 percentage point decrease compared with Q2. The three quarters gross margin rate year-over-year decreased 2.4 percentage points. The major reason is the product mix and the unfavorable foreign exchange rate impact in this year. The operating expense was TWD 3.5 billion, 28% increase from the second quarter. Three quarters year-over-year operating expense increased 42%, major from the increase of people cost and the increase of the project expense. Operating margin in the third quarter was 12.3%, 1.1 percentage point decrease compared to the second quarter. The third quarter net income was TWD 7.8 billion. EPS was TWD 14. Three quarters EPS was TWD 32.18. Next page. About the balance sheet. End of Q3, our cash balance is TWD 27.3 billion. Inventory was TWD 31.8 billion, 22% increase compared with Q2. The mainly due to the increase of the finished good and the material preparation for the Q4 shipment. Q3 inventory including 14.7 billion finished good and 14.6 billion raw material. Let's move to the key financial ratios. The Q3 end of Q3, the debt ratio is 63.93%. Compared with Q2, cash turnover days decreased 12.83 days to 10.83 days. It's mainly due to the reduction of the inventory turnover days. Next page. The revenue by segment. Our major business comes from network application products. It's 63% of total revenue for the three quarters of this year. Increased 361% year-over-year. The switch product revenue is 34% of total revenue for the three quarters of this year. Increased 46% year-over-year. Non-AI switch revenue is 2% of the total revenue for the three quarters of this year. Next page. The revenue by region. Our major revenue contributes from America. It's 81% of total revenue for the three quarters of this year. Revenue from Taiwan and the Europe areas increased. Taiwan's revenue increase is due to the customer change to the Bill-To entity. Now I will turn it over to Ed. Yes, thank you, Fanny. Now we'd like to turn to company highlight presented by our President, Jackal Lee, please. Yep. Hi, everyone. I think you all know Accton is starting from 1988, and we are the ODM OEM custom companies. We started to join the OCP in 2011 since the OCP start. We started to working on the open networking area. Then we launched the first 100G switch in 2016. It can make the 10 years steady growth. So you can see from the 2016 to the 2024, we have started steady growth with the healthy growth. With the 10 years, we are not only the focus on the manufacturer, but also focus on the R&D technologies. So you can see we have a good return in this year, 2025. It will not cause the short term period. It will be long term. We focus on what we can do. So looking forward the next several years, the most of the hyperscale, the demand and the investment is very strong. You can show the public news from everywhere. So the most important for us will be excellent execution for the growth. So for us to see future several years, the most important is how we execute the demand. The not the normal, not the good, not the best. We needed to do the excellent executions. So we will keep to our, we will do our best and do excellent execution for future growth. Thank you. Okay, thank you. Thank you, Jack. And now we'd like to turn this call to Michael for the technology highlights. Michael, please. All right, thank you, Ed. Okay, ever since the rollout of ChatGPT in 2023, AI has become a household name in the world. For R&D, we have been reviewing how we can keep up with the industry and capture new opportunities without breaking the bank. In the high-tech industry, changes are inevitable. As the company continues to grow, we embrace changes and identify the key areas to focus on to achieve operational excellence. Next page, please. Let's start with strategic direction. A goal without a plan is just a wish. To achieve a goal, we need both a vision and a strategy to execute it. In Accton, we added a few new items into our core technology portfolio, including 224G series and 448G series. These are the foundation of the 1.6T and 3.2T switches. And the rack integration, these are the obvious ones. In addition, we have OWS and WSS. These are related to optics. OWS stands for optical wavelength switch. WSS is wavelength selective switches. The basic concept of OWS is to use wavelengths instead of port for switching. The WSS is the core technology behind a wavelength switch. The potential use cases is for scale across data center networks. We expect optics will play an important role in high-speed interconnect. Next page, please. Okay, moving on to innovation. Innovation distinguishes between a leader and a follower. In Accton, we are in good position in 800G. To lead in 1.6T, we continue to promote innovation and encourage people to think out of the box, starting from the basic ones such as packet switching versus circuit switching or port switching versus wavelength switching, retimers versus redrivers, et cetera. We are looking at everything, without taking everything for granted. We want to innovate with the new AI era. Next page, please. The next topic is related to execution efficiency. Efficiency is about doing the right thing, not just doing things efficiently. In Accton, we train people and build mutual trust through teamwork and plan ahead of time every day to do the right thing. The Operation Midnight Hammer is a good example of what we are shooting for. We work hard and celebrate success together to move a miracle. Next page, please. For people development, leadership is to do the right thing at the right time. Management is to do things right. In Accton, we acquire many new talents to build new teams and work on new technologies. As more people join the Accton family, we develop knowledge base and bootcamp training materials for all newcomers to become our next generation leaders. Next page, please. For quality, quality is never an accident. It is always a result of intelligent effort. Quality means doing it right when no one is watching. In Accton, we continue developing processes by using new tools and automations to minimize human errors. As more AI tools become available, we encourage people to take advantage of the AI to build smarter tools for all works to improve the quality whenever possible. Next page. Moving on to the customer satisfaction or CSAT. A CSAT is the best business strategy of all. The most unhappy customers are the greatest source of learning. In Accton, we review customer feedback regularly and constantly looking for improvement opportunities. Next page. For business scaling, as we grow into a global footprint, the business scaling is about subtraction, not just addition. We are adding a lot of sites, but this site will cost more money. It requires a shift from a business to strategic leadership to achieve growth without a proportional increase in cost. In Accton, we constantly need to find new ways to do the old things. Human resources are the most expensive item in R&D. Being able to use tools and increase productivity is essential to support business scaling. Next page. The last one is about organization alignment. If everyone is moving forward together, then success takes care of itself. To move together, we need to align everyone to behave like a team. Accton is a teamwork-oriented company. Not only do we need great leaders, but also we need great executors. In Accton, we promote people based on merit and contribution. We use bell curves to motivate and reward high performance. We believe this is the right way to achieve long-term success. In summary, we believe the focus of R&D is operational excellence to become a world-class organization. Thank you. Okay, thank you, Michael, for your presentation. Now we'd like to turn to the section of Q&A. I think Derrick can help us to select questions in the box. [crosstalk] Okay. Thanks for the presentation. Now we will start the Q&A session. If you would like to ask questions, please use the raise your hand button in the system and wait for your name to be read out, so our first question will be from Carrie Liu. Carrie, please unmute yourself and go ahead. Yes, hi. Thank you for taking my question. My question is actually about demand and your capacity planning. So first of all, just wanted to hear your thoughts about how do you see the AI demand evolving over the next three years? And given that growth trajectory, what strategic steps you are taking to optimize the global capacity and ensure sufficient supply across key regions? Thank you. Okay, so let me answer the questions. Currently, more and more people say AI will bubble. But in other angles, we have a lot of hyperscale customers. They have put a large investment there for the future several years. For us to see, either something like a bigger 2000, 2020, 2000s, the 2000s, the dot com, either same situations, some company will still become a big and a great company, but some company will be broken in this time. For our point of view, the AI demand will be really strong in the future because AI is the future technology, and we need to use for them, either something like a networking and the dot com. The demand currently we see is very strong. We needed to execute well, then can support the demand, the first one. The second one, due to the geopolitical and most of the components shortages and the capacity issue and the labor limitations and the talent limitations. So global footprint will become the most important for the future several years. Now we have already planned for the most of the countries. Of course, either it's very difficult once we work in a different country because different country have different people, different culture, and a different environment needed to stay up. So in the different country, the strategy will be different, but the most important will be how you work with the local country partners. If you can have some partners in the local, then he can help you to speed up the overall development and the construction in those areas, and you will get a lot of return. So for the different country strategies, we will work with the local government, local partners to help us and also help them to grow in their countries and make a success for the future. Okay Okay, thank you. Next question, Derrick. Sure. Next question will be from Paul. Paul, please go ahead. Paul, please unmute yourself first. Okay. Hi. I'm Paul from KGI Securities. And there are two questions about Accton's rack-level development, and I would like to discuss further. The first one, what is Accton's current progress in the level 11 rack-level integration and the testing, particularly in its engagement with the North America customers? How does the company plan to participate in this area? And what's the strategy and the outlook that can be shared with investors? This is the first question. And I have the second question. The second question is regarding the key components used within the rack, such as interconnect cables, whether copper or optical. What Accton's plans and the strategies additionally for the air cooling and the liquid cooling components? Could you share more details about Accton's current preparation and the R&D progress in these areas? Thank you. I think that the question, we did our R&D VP, Michael Lee, to answer. Thank you. All right. Okay, so to answer the question on L11 rack level integration. So Accton started looking into a rack level integration about early 2023. We recruit several talents starting at R&D office about as early as 2023. Then we start our growth the team, then we need to start looking for a new location to house more engineers. And L11 used to be a server space. And we saw that L11, L10 is more and more become the basic unit for AI compute. And typically networking is not part of it. But it took the trend is everything needs to move to the rack. And the reason we are investing early is because we see this the demand about two plus years ago. So we start investing on rack level. The reason we need to go to rack level is because if you use the air cooling system, you have the built-in fans and within the box. So everything is controlled under the box level. As we move on to the liquid cooling, the full liquid cooling, there's no more fan within the box. And to replace the fan is by the CDU. We also developed the CDU is a liquid cooling. So the CDU is not in within the box or outside the box. So it must be the rack. As we move on to a more AI oriented product, we realized the need for liquid cooling, the need for L11. We have been investing in L11 for more than two years. Where we are at, we think there are three parts of it that need to be answered. One is our development can be ready for audit. The second is when is our manufacturing will be ready. The last one is how can we support the L11 in the field. So we have planned all three different phases a few years ago. So the target timeframe for us to get L11 ready is roughly. It's no guarantee. It depends on the progress. We also need to have some relationship with the customer and the progress of AI product lines that require liquid cooling and L11. R&D about the L11. They're shooting for ready by Q1 2026. For manufacturing, we're shooting for Q2 2026. And the fields are ready about Q3 2026. These are regular timeframe of our readiness. I hope that answered the question. Thank you. Thank you. It's very clear. Thank you. Your second question is the key component regarding copper versus optics. Okay. We have been working on this for a long time. About the same timeframe, we've been invested, heavily invested in optics. Copper has been our traditional signal integrity space for high-speed internet. We feel very comfortable on that. But as the speed goes higher, we think it's going to bump into a wall that the copper cannot travel too long in the rack environment. So we think in the scale-up environment, the copper is still dominant. We need to deal with either cable tray arrangement or the cable within the tray. Okay. As the speed moves on to higher or lower distance, optics is inevitable. We are heavily invested in optics. As I mentioned earlier, we're not only looking at the scale-up. We are one of the early pioneers looking at the scale-up cost by using a OWS. Okay. So we have made some progress in this space. I think we are in good position if we have to compete in the industry. Thank you. Thank you. It's crystal clear. Thank you. Okay. Thanks for the answer. Our next question will be from Jackson. Jackson, please unmute yourself and ask your questions. Okay. Seems that there's some technical issue on Jackson's side. So maybe let me ask one question here. Could you share with us some of your views regarding this 800 gig network switch migration into next year and also maybe longer term, the 1.6T migration, when is that going to happen? And do we see any change in the competitive landscape because of this migration? Thank you. [crosstalk] Thank you. Okay, this is Michael. Okay. In 800 gig, it's in the phase we try to capitalize. I think everybody has the 800 gig now. We are lucky to, we are one of the early ones to be able to ship in volume in 800 gig. So we are not starting from 800 gig. So we continue looking to 1.6T. And 1.6T is very different from 800 gig. There's lots of other things changed. The main changes are from the 400 gig to 800 gig, the product cycle is about four to five years. But from 800 gig to 1.6T, you can see the product cycle is reduced. So it is important to be the first one. Otherwise, the time to capitalize your investment is getting shorter and shorter if we are to the market late. Many new technologies required and need to be invested for 1.6T. Mainly the hardware changed. And the standard committee is still working on 1.6T. So in order to win on 1.6T, what our strategy is to invest on 1.6T infrastructure to make sure when the silicon is ready, that our infrastructure is ready for the 1.6T switch. So we think 1.6T switch is real. And in 2026, there will be a lot of companies trying to roll out 1.6T switches. But the 800 gig switch is still the mainstream, unless the 1.6T price can be justified by the, the capacity. Thank you. Okay. Thank you, and maybe my next question is regarding more of a new technological CPO, co-packaged optics. What are the implications to Accton from this, from this new technology from a switch ODM's perspective? And when do we think that this, this technology will be adopted widely in the industry? Thank you. All right. Thank you. That's a great question for CPO. Okay. CPO term, this term is around for at least four to five years. It has been very bumpy. Sometimes the CPO comes up, sometimes it dies down. But every time it dies down, the next, next, the CPO topic will show up again. And one of the reasons we have this kind of roller coaster is because every time we think we need a CPO, there's always somebody coming up with a bug or solution. So the CPO timeframe will be pushed out. The original definition of CPO is integrate the optical engine into a switch. But more recently, the CPO definition has been slightly changed. Any silicon and optics integration within a substrate, and we call CPO. What that means is CPO is not only referring to the switch, electrical switch with optical engine. It also can apply to a GPU with optical connection. You also can apply to a DSP with optical integration. So today's definition of CPO is, it's a broader one than a few years ago. But in Accton, okay, if we start with a CPO using a switching chip plus optical engine, it's going to take a longer time. So we have several programs. Okay. Without waiting for the switching chip and optical engine to become available, we start with a DSP with optical engine, but we start with transceivers. So we're looking at a transceiver CPO to help us to learn the manufacturing technology. If the silicon and optics integrate as a system vendor, the manufacturing capability will become very important. So we don't want to waste any more time. We can start with what's available today by working on the DSP-based CPO to help us learn manufacturing technology and get some of the key technology ready. We are in the process of working on switching chip with optical engine. That's what is in the works right now. That's our next step. The next step will help us to learn how our manufacturing line can deal with both electrical and optical capability and be able to build those products, not only the design of products. Okay. We think we are on track with what the industry trend is. So we'll be participating in the CPO when the time comes. Thank you. Okay. Thank you. And then, let me check. Our next question will be from Helen Tsai. Helen, please go ahead. Hello. Can you hear me clearly? Yes. Please go ahead. Thank you. I have two questions. One regarding networking, the other one regarding the AI rack assembly. So the first question is, what's Accton's position in scale-up, scale-out, and scale-across networking? What's the growth rate we expect for networking for the next three years factoring in new customer wins? And the second question is, what's our competitive advantage for AI rack assembly given the competition from server assemblers? What's the chance of getting a new AI rack customer? Thank you. [audio distortion] [Foreign language] scale-up, scale-out, scale-up. All right. Okay. This is Michael. Let me try to answer the first question. Accton's position in scale-up, scale-out, and the scale-across, we have planned for all three. The scale-up is mainly within the REC and the interconnect still using a copper solution. So we have a good relationship with all the major system vendors, to design the switching tray, for scale-up. Major one is the major silicon company, they have the new ESUN. Those networks are coming out, and we are part of the scale-up network solution. Okay. For scale-out, we also have a VOQ-based switch, based on the spraying technology, for the load balance. So in the load balance intelligence within the switch, we already have the solution. For next generation, it's mainly the speed upgrade from 100 gig to 200 gig. So we think we are pretty covered on both scale-up, scale-out solution. In addition, okay, we think we are ahead on the scale-across. We have acquired four technologies from other companies. We also recruited talent. It's mainly on the wavelength switch. So the concept of wavelength switch is to aggregate multiple ports into one port. So before we transmit, we can aggregate multiple high-speed ports into one fiber and then transmit from one side to the remote side. And the remote side, then we can recover from the 32 ports. I think this technology is viable. It's a very competitive solution for future scale-across network. I hope that answered the question. Thank you. Thank you for answering. So Helen, can you repeat your second question? Yes. So what's our competitive advantage for AI rack assembly given the competition from server assemblers? And what's the chance of us getting a new AI rack customer? So I think the L10 in the AI core, the AI core, the product is totally different compared to the original power switch product line because when we use the Broadcom, you can see the Tomahawk 1 to Tomahawk 6. Maybe currently is working on Tomahawk 7. It will be a stable technology. Then we can follow some guidance and follow some layer reference. Then we can decide this kind of the product in our by ourselves. And this is the for the networking is around the 25, more than 20 years, the development. But for the AI, it's very new. So either will be another standard design and most of the chip will be the different. You can see most of the customer have a scale. They want to have their own chip. Yeah. Different chips will cause different designs and the service and the technologies. I think we are not the first year to work with the AI core customers. We have long experience. We have several years experience in this field. I think that we are better than others because we know how to design and how to support this new technology. This is the first one. The second one is the speed. Our iteration speed is quicker than other company. In the AI world, the change will be faster compared to the original network switches. The speed will be more important there. We have a strong R&D, which is led by the R&D VP, Michael Lee. I think that it will be the core value for Accton for the AI technology in the future of several years. And for the second one, what is the chance to get the new AI customers? I think we are working with some new opportunities. Yeah. We will keep in spend more resources in this area. Thank you. Thank you. Okay. Thank you. I think, next question, maybe I can ask a couple of questions here. So, my, this one is more of a higher level one. So, what would be Accton's growth opportunities beyond the current network switches and this AI core businesses? Beyond that, what are the potential addressable markets that we're looking into that could help us grow faster in the coming few years? Okay. So I think the networking suite either will depend if we can execute excellent or well to catch the demand for the AI and the networking solutions. Yeah. Currently, we need to try very hard to catch the demand growth. This is the first one. So if we can execute well for several years, then I think the most we will go through the opticals. I think the optical will be the future, the technologies. So here is currently what we plan. Yeah. Good. Thank you. And then, as you mentioned, demand has been very strong. So, we are probably already tight in our capacity. And, I think our board of directors yesterday approved three projects. One is an investment in our Vietnam factory. And the second one is a capital injection to our U.S. subsidiaries. And the last one is the plan to set up a new subsidiary in Malaysia. So, are these plans all related to sort of new capacity additions in the future? Could you share more colors on that? Thank you. Either you may say, currently the capacity, I think only the manufacturer capacity, but there are also R&D capacity because currently in Taiwan, the talent is limited. So we need to have more countries to set up different manufacturers and the R&D office to have more talent for future projects. So this is why we have several countries who set up our future office, future company there and to hire more people to support the future growth. We were, we for this several activity join. I think that in the few, it will also depend on how to quicker return the resources to help us for those kind of the demand. Yeah. So, please allow me one more thing that not just for the talent limited or the opportunities outside North America or Taiwan, we believe there's some demand for the local market or local demand to be satisfied. So I think the way we work with the local partners, local talents that can, you know, help our CSP customers to expand their market in other countries. So as you can see, not only the switches and the ASIC, we will have, like, our Jackal just mentioned that will be optical or Michael mentioned the cooling stuff. So there were more integration, the different products across different countries. So, not only the, I just want to remind it will not only the expansion in the capacity, but also the potential expansion in the market and the customers as well. Thank you. Okay. Thank you. And then I think, during the prepared remarks, you mentioned a switch technology based on this wavelength selection technique. Could you share a little bit of a color on that, regarding the development of that technology and what's the implications to Accton as a switch maker? Is that a positive or negative to the value added, from your perspective? Thank you. Let me share some of the background on wavelength switching. This technology has been used for telco industry for more than, say probably, two decades. The originally is coming from called ROADM. Basically, it's an add and drop circuit. But the speed is so slow for telco. If you look at the AT&T fiber at home, you are looking at the one gig, 2.5 gig, or 10 gig at most. But we are using, trying to use the same concept. So there's no problem for the underlying technology. That technology has already been proven in telco. So we are taking the same technology to improve it, to make it run a lot faster, to make it suitable for data center applications. So we took the same technology, modified to make it faster, to meet the data center requirement. We have the prototype, POC stage. We are not ready to do a deployment or ready to talk to our customer yet. Those are still in the works. So far, it looks very promising. The basic concept, it looks like it's going to work, but we still have a long way to go, because most people may not be familiar with optical wavelengths. Most people know about OCS, optical circuit switching. The optical circuit switching is still port switching. The wavelength switching is wavelength switching. Basically, it's a similar concept, but instead of using a port, you're using a band, using a different wavelength to switch within one port. The technology has been proven in other market segment, but we're trying to bring it to the data center space. So there are some things we still need to validate to make sure that the whole theory has sound before we can roll out to the wider audience. Thank you. Okay. Thank you. And then maybe, a little bit on the, also pretty popular topic, recently, especially since the OCP last month, that's E-SUN, Ethernet for Scale-Up Network. So that's been promoted by Broadcom and the likes of AMD and other hyperscalers. So what would be the implications to Accton from that initiatives? Is that a positive or you think that it could be some headwinds for us? Okay. So ESUN is pretty much, if you look at the OCP, made an announcement. ESUN, it's a huge announcement, but Accton is positioned as ODM customer. So our name will not be on the list, but there are companies on the list that have embedded our technologies, somewhat related to Accton. So we are in ESUN network, we designed for our customers so that we can enable them to run application software. Okay. So we are in both, basically for scale-up, we have two types of endpoint, switched fabric and switch switched fabric. We have many products, we can fit in the ESUN network. Thank you. Okay. Thank you. Well, actually we have been talking a lot about this industry dynamics and the technology roadmaps. So, why don't we switch gear a little bit into some financial outlook? So looking to 2026, do we have any preliminary view regarding our revenue growth and also maybe the seasonality between the quarters? Thank you, Derrick. Just like we discussed, we see many strong demands from different customers, especially from the CSP, but there's still some challenges, including the capacity, the talent, and also certain key components, so looking back, 2025, we think the growth is really a surprise, but we are also very proud for our team that we can capture and meet the demand of our customers, but next year will be a bigger challenge, and even with the growing demand and also with a very promising CapEx for our customers, but we believe next year will be a very busy year. Definitely we are optimistic about the growth, the business growth, but we're still very cautious because I think the secret and the winning bottom line is whether we can overcome those issues, just like Jackal mentioned, not only in Taiwan, also in the U.S., Malaysia, and other countries. Like I just said, what we can say right now is we're still very optimistic about the growth next year, but we still need to be very cautious in the execution, in capacity, in talent, and in key components. Hopefully I can answer your question. Derrick, thanks that. Yeah, no problem. Okay. We're currently in the Q&A session, so if you would like to ask a question, please use the raise your hand button in the system, and please, ask your question after your name is read out, so our next question will be from Helen Tsai. Please go ahead. Hi. Thanks for taking my question again. I want to ask about the server rack contribution and also how it would impact our overall margin profile. Do we have more color about that? Because I think in the previous Q&A, the CTO mentioned that we should start manufacturing sometime second quarter next year. So I'm wondering the revenue contribution, and also margin profile for the business for next year. Yeah, this is Michael. I think maybe the message may not be explained clearly. We want to get our technology ready in manufacturing. It does not mean that we start doing produce something in manufacturing. So basically we want to prepare our infrastructure before actually build a product so that our schedule promise to our customer can be met. So this is one clarification. Okay. The other one is a server rack. Probably a server rack could be misleading. We usually refer to a rack level integration because within the rack, it's hybrid. There are servers, AI servers, there are networking switches, there are power shelves, there are CPU and cooling devices. If you're referring to servers, Accton is not designing servers, but we are heavily involved in the rack level integration, anticipating working with any AI server company or any future product demand, to be able to take on AI servers. Thank you. Sorry, a quick follow-up. So in terms of manufacturing ready or if we are going to manufacture rack in the future when there are orders, what part will we actually assemble and manufacture? Does that include server boards or we just do the integration of some other key components and also including networking? All right. Thank you. So within the rack, there are many types of components that can be put on the rack. For example, the networking tray or the computing tray or power tray. Okay. So we may not design all of them. Most likely we will not design everything. Okay. So, but if we ask, okay, we may design the networking tray, but we must integrate other suppliers, computing tray. In this case, we need to have the capability to do a rack level integration. Okay. We just want to make sure we have the capability if we have to do that. Okay. So that we don't miss out any opportunities. Thank you. Thank you. That's very clear. Thank you. Okay. Thank you. So, in the interest of time, we might need to conclude the call here. Before that, let me pass it to Accton's management for the closing remarks. Okay. Thank you. Thank you, ladies and gentlemen. Welcome. And also thank you again for your participation today and for the needs of the shareholders and our customers. As I, as we just mentioned, we will be positive in the business and also cautious in the execution. And also hopefully our sustainable growth will be continued. Thank you. Thank you again. Wish you all have a nice day. Thank you. Okay. Thank you everyone for joining. See you next time. Thank you. Thank you. Bye-bye. Bye-bye. Bye.