Earnings Call: Q3 2025

Nov 13, 2025

Good afternoon, everybody, welcome to Asia Vital Components' third quarter earnings call. Today joining us are Eric Chen, Asia Vital Vice President, Matthew Shen, Asia Vital Deputy Spokesperson, and Bill Chen, Asia Vital Deputy Spokesperson. Mr. Shen will summarize the third quarter operations and financial results. After that, we will open the floor for Q&A session. Now I would like to turn the call to Matthew for the third quarter operation and financial results. Right. Hello, everyone, and good afternoon. Actually, Bill is our finance manager, and he will be taking us through our Q3 2025 results before we jump into Q&A. Okay. Good afternoon all investors. I will now present our financial results for the third quarter of 2025. First, here is our disclaimer. Okay. Here is the consolidated income statement for Q3 2025. Our Q3 revenue reached TWD 38.9 billion. Gross margin for the quarter was 26.12%. Operating margin was 20.77%. Net margin was 13.71%. For EPS, the third quarter earnings per share reached TWD 13.67. All metrics improved compared to Q2 2025 and the same period last year, mainly driven by the continuous increase in AI server shipment, which significantly boost revenue margins and EPS. Moving on to the year-to-date consolidated income statement. Revenue for the first three quarters reached TWD 91.8 billion. Gross margin was 25.5%. Operating margin was 19.09%. Net margin was 13.66%. Cumulative EPS reached TWD 32.25. Compared with the first three quarters of 2024, revenue, margin, and EPS all record significant growth. The compound annual growth rate for revenue and EPS shown here are based on last year's data. We will move through this session quickly. Here we have AVC's revenue breakdown. In Q3 2025, our core business, thermal module and chassis, generated TWD 71.1 billion, accounting for 77.4% of total revenue. System integration product contribute TWD 12.3 billion or 13.4%. Our subsidiary, Fositek, generate TWD 8.4 billion or 9.2%, bringing total revenue to TWD 91.8 billion. What are you thinking about over there? That I'm an existential crisis. Compared with the same period in 2024, revenue from our core business grew 80%, mainly due to the strong demand for AI servers. Next is the revenue mix within our core business. Server product benefits from the rapid expansion of AI application and strong end market demand. In Q3 2025, server revenue reached TWD 46.6 billion, accounting for 65.6% of core business revenue. Communication, consumer electronic, and the other category also saw growth, but their weightings were slightly diluted due to the significant increase in server revenue. Overall, core business revenue grew 106.7% compared with the same period in 2024, primarily reflecting the strong customer demand and pull-ins. Okay. The charts show the quarterly revenue trend, and you can see that in Q3 2025, thermal module revenue continued to benefit from strong server demand and reaching TWD 31 billion, representing substantial growth over last year. Other product categories show relatively stable performance. Here is the quarterly revenue breakdown for our core business. Server revenue in Q3 2025 reached TWD 21 billion, showing significant growth both QOQ and YoY. Revenue from networking, 3C, and other categories remain relatively stable. It is evident that server products continue to be the main driver of our revenue growth. For non-operating items, most of the contribution came from others, primarily due to the higher other income. As demand for the liquid cooling module increases, customers will require more one-time tooling fees. In general, as the number of projects grows, this category will continue to expand. Here is the balance sheet. The total assets in Q3 2025 reached TWD 139.3 billion, including TWD 43.8 billion in cash. Total liabilities were TWD 97.7 billion. Shareholder's equity was TWD 41.6 billion, and the book value per share was TWD 93.32. Compared with the previous quarter and last year, the most notable change was inventory, which increased due to the stronger overall customer demand and the request for early stocking. Moving on to the financial ratios. Debt ratios and account receivable turnover remained stable. The decline in current and quick ratios, as well as the longer inventory turnover days, were due to the increase in inventory levels. Here is the cash flow statement. Operating cash inflow increased significantly, resulting in a substantial rise in free cash flow for Q3 2025. Net cash inflow for the quarter reached TWD 17.6 billion. This demonstrates that the company cash flow remains healthy. While expanding capacity, we continue to maintain financial stability to cope with future global uncertainties and challenges. Okay. Thank you, everybody. We will now move on to the Q&A session. Okay. Thank you, Bill, for taking us through our financial results of Q3 2025. We will now move into Q&A. The Q&A section will be answered by myself, Matthew, Bill, and Eric. We have collected questions ahead of this meeting, so we will be answering those first. If you have any questions, please submit them in the chat area, and we will follow answering them post answering the pre-submitted questions. We have the first question from Matthew Six, which is to please comment on AVC's 2026 outlook. This question will be answered by Bill. Okay. Thank you, Matthew. We believe that order demand for 2026 remains strong. At the same time, we are collaborating with NV, ASIC, and AMD. As long as their demand continues to grow in the future, we will remain one of their key suppliers. We also expect the penetration rate of liquid cooling to keep increasing in the coming years, and the company remains highly optimistic about the future outlook. Thank you. Okay. The next question is from Chartered Asset Management, and it's for us to comment on our growth drivers for the next three to five years. Over the next three to five years, AVC sees multiple growth drivers supporting its long-term expansion. AI-related demand is expected to continue growing beyond 2027, with customer orders and forecasts remaining strong. Beyond AI, the company is actively developing thermal solutions for emerging applications such as autonomous vehicles and humanoid robotics and space technologies. As long as technology continues to advance, the thermal industry will evolve alongside it, and AVC is well positioned to lead this transformation as the industry's foremost innovator. Thank you. Okay. The next question is from Eastspring Investments. The question is: What's management's strategy to capture the next wave of growth trend for thermal management given the uncertainties with the final solution? To answer this, I'll be answering this question. AVC continues to invest heavily into our R&D. We're spending 5% of our revenue on R&D every year. Our leadership team is also particularly focused on continuously evolving their own knowledge by speaking to experts in the thermal industry consistently and looking out for all new types of technologies to use for AVC's future projects. Some of the solutions that we are designing ourselves with clients with ODMs in preparation include immersion cooling, microchannel lid, and two-phase cooling. We've ventured to make aluminum cold plates, titanium VC lids, et cetera. We are always looking ahead and trying to prepare ourselves for whatever the next solution may be in the thermal space. This ensures that we will continue to always have the capabilities required for any future development and to continue to deliver our value as the leaders in the thermal space. We also have the team experience and resource to give us the confidence to do so. Our customers are actually working with us up to three years ahead of mass adoption to develop solutions, and we believe that this gives us sufficient time to develop and mature our technologies. As an example, we've already been working on liquid cooling, which has only recently become widely adopted in the AI space. This is a technology that we've already worked on for more than seven years. We are always looking ahead, and preparing ourselves to continue to extend our lead as the leaders in this space. The next question is from Brock Milton Capital. How has the competitive landscape changed lately? Are we expecting it to change materially in the coming years? I'll also be taking this question. We believe that the competitive landscape has not changed significantly, and that we are still the leading designer and manufacturer in the rack level thermal solutions. Although we have seen multiple companies branch into the thermal solutions that we are producing, we believe that the customers choose us time and time again because of our ability to design, manufacture, and provide end-to-end thermal solutions, what we call our total thermal solution, at a very high quality. Our customers are still coming to us with their latest projects, and we continue to deliver exceptional solutions. Our main competitors in this space remain the same players who have been in this space for many years and have built the expertise to design, not only to manufacture. We believe that the competitors that we're working against are still the same players from more than 5, 10 years ago. The next question is from FIL. How do we maintain competitive lead with NVIDIA? Once again, I'll be answering this question. Actually, well, I'll let Eric answer this question. Thank you. Speaking of business, it's always a combination of different angle of your competitiveness. We are not only been long partnership with NVIDIA. We have been, and we will be continuing to demonstrate our capability in terms of technology advance, in terms of geo-diversity manufacture capability, in terms of high quality, super bigger volume production capability, also good commercial cooperation. From different angle, we all try build ourself, become a tier 1 vendor, tier 1 partner to cooperate with all the key client we have. We will also continue to penetrate our advantage and leverage factors, and continue to work with NVIDIA to serve them, to let them feel satisfied with our service. Thank you. Thank you, Eric. The next question is from Polar Capital. What is your estimated market share in AI server cold plates? Currently across the CSPs GB products, we have more than 50% market share, in the more like 70%-80% range. This is actually abnormally high for a component manufacturer. Longer term, we are looking to target 50% market share. We are very confident to continue to be at least one of the main suppliers in every large customer project because we have the best resource, experience, and also more importantly, out of China capacity. For our Chinese customers, actually, our share is even higher as we have worked with all of the main CSPs for a long duration of time, and we have been well established as a trusted partner in the Chinese tech space. Next question from Nippon Life. How do you view competition from China in cold plate/CDU? We have many competitors in China, but since thermal is a fraction of the cost of the entire rack, customers are still prioritizing quality and reliability over cost. All of our Chinese competitors are good competitors, but we still feel the most presence from the established Taiwanese companies who have been in the space as designers, not just manufacturers of thermal solutions, rather than comparing with the cost-based suppliers. The next question from Bank of America and FIL, what is our capacity expansion for 2025, 2026? We are still aggressively expanding in Vietnam, and our plan currently is to quadruple our current footprint in Vietnam by 2027. We will be completing our V5 and V6 facilities by December, and to start bringing in equipment. Our V7 and V8 will be completed by Q1 of next year and brought into production by Q2 of next year. The main focus for our next phase of expansion will be on chassis and liquid cooling modules. Our V9 to V12 facilities will be completed by Q2 2027, and the focus is to continue to deliver our value in the AI server space. Our capacity plan quantitatively is still as outlined before. We are approximately targeting 50% growth in our capacity from our next phase of expansion in 2026. Our 2027 capacity is still to be confirmed. The next question is from Carmignac, and it's on our current mix of GPU/ASIC in our AI servers. Alla tänder är unika, och Oral-B är på ett uppdrag att hålla alla barn kariesfria. Vår elektriska tandborstar rengör bättre än manuella tandborstar, och når svåråtkomliga ställen där hål börjar. Musiktajming gör tandborstningen rolig. Godkända av tandläkare. Barns perfekta rengöring börjar med Oral-B. Lindahls kvarg. Krämigt god med högt proteininnehåll. Lindahls. Intensigt god. Our current capacity and sales revenue in AI space is still driven by GPU. It's more than 60, more than 70%. In next year, GPU will continue have a good growth. At the same time, we see dramatic growth. We will have a very high growth rate for ASIC chip thermal solution, they are all driven by cold plate module, cold plate product. Okay. Thank you, Eric. The next question is from HSBC, and it is on the impact of ASIC adopting liquid cooling. What does the content value look like? Actually, if we look historically between Hopper and GB200, we have seen content per chip grow by three to five times when we have made that transition from air to liquid-based solutions. Liquid cooling penetration for our largest ASIC customer is still low on the aggregate level currently. We are expecting to see liquid cooling penetration increase significantly over the next two years. Not only is our existing customer going to increase in liquid cooling penetration, we are also expecting to win new customers who are default liquid cooling. We will still continue to have volume for our air-based solution. We don't believe that it is going to be completely obsolete, and this is still a part of our business that we're excited about going into the future. Next question by Chartered Asset Management, key management member succession planning. AVC continues to train up our key employees and encourage them to continue to grow and promote upwards. When we look on the longer term, our key roles will continue to be filled by employees who have been in the company long term, are familiar with the company, and who has worked their way up and built deep relationship and trust within the firm. Across our teams, they're all working very closely, and our managers have all been held responsible for their own KPIs to drive sustainable success of the company. We as a company are very passionate about manufacturing and the thermal industry. Our key management see this as a critical factor when picking a successor in the key roles. Okay, the next question is from HVC on the progress and adoption timeline of microchannel cold plates. Eric will take this question. Okay. First of all, I would like to clarify a concept. Different technology is not necessary replace each other scenario. For every technology who will adopt in a system, first of all, you need to check how does the system has been designed, and what's the critical design factor for that product. First of all, AVC, we do have our microchannel lid, this product. We make POC, we test pass, and we have capability to mass production for this product. This is first. Secondly, similar with this one, we also have other product. Take some example, VC lid. We also have capacity, capability to make it go mass production. Microchannel lid is actually a product like combination of VC lid and the microchannel cold plate. All above those solutions, we have our capability, our technology to make them mass production. Go back to this question itself. The key when we develop a product, we develop a production, is when and how serious our key client they want to consider, they want to make a technology go mass production. As long as our client show their determination, show us the direction that they want to make some product go mass production seriously, then we will work with them along to make it happen. So far, what we see, what we are working with client, microchannel cold plate is still the main product client want. This is the direction we are working with client. Again, at the same time, when we see if client someday in the future, they want to do MCL, we will be very comfortable. We have high confidence we can make their product for mass production with them together. Okay. Thank you, Eric. The next two questions are still pertaining to microchannel lid, and Eric will still be answering these. I will still ask these questions to Eric, to see if he has any further comments to add. The next question is from Ahan Capital and KCP. How do we view the threat from microchannel lid, Eric? Yeah. Same like what I just shared. We do not see it a threat now, and we are also comfortable that once this solution become a choice, a potential option for client, then we will be ready by that time. For so far, the current stage and the near coming future, what we see, microchannel cold plate is still the key product client would AVC to work with them to make it go mass production. Thanks, Eric. The next question is from LMR and SinoPac. Please, can Eric comment on microchannel lid, MCL, versus microchannel cold plate, MCCP? Okay. microchannel cold plate. It's a cold plate. It make the inner channel thinner and quantity wise more, so it can dissipate, can bring the heat out of the system several times much better than the regular cold plate. The MCL is a combination of cold plate and the VC lid. They are like similar technology, but using in different situation. The key of MCL being adopt into mass production will be the upstream tester and the upstream chip maker. If they can make this solution go mass production, high quality, better sustainability, it will be key in that function. For us, we are still working on MCL at all as a major PO product for client now. Okay. Thank you, Eric. Please may I remind our investors if you have any questions to put it in the chat as we are approaching our final question right now. Our final question that has been pre-submitted is by Robeco, and it is for us to comment on the progress of ASIC projects and revenue contribution size. I'll be answering this question. As a portion of our current AI server business, our ASIC projects contribute approximately 20%-30% of our AI business. We are of the opinion that ASIC will outpace NVIDIA in the coming years, and our leadership team is of the opinion that our revenue generated by ASIC projects will match that of NVIDIA within the next two years. In terms of progress with the current ASIC projects, we are currently already the leading designer for three CSP ASIC projects, and we are working very hard to become the leading designer for all ASIC projects. We believe that with our size, expertise, resource, and capacity, we have a great advantage over our competitors in this space. Thank you. Now we will be moving to answering questions from the chat area. Please may I encourage investors to submit your questions. The first question, should microchannel lid be adopted, the reference design will be shared with AVC, and AVC will be able to quickly ramp yields to meet demand? Thank you for this question. We have actually already been in in-depth conversations with our customers, with NVIDIA regarding this. As we've seen in the past, our reference design has been shared with our competitors so that NVIDIA can build a rigid supply chain. Should MCL be adopted, we will definitely be participating actively in that solution. We, as Eric has mentioned earlier, Microchannel Lid is a combination of a cold plate and a lid, both of which are solutions that we already have mass production experience. We have the confidence to be able to deliver, to ramp yields, and to mass produce in a short period of time. Next question by George. Can you comment on the dollar content per rack for future generations of NVIDIA, for example, Rubin Ultra? Is it reasonable to assume a similar jump from Hopper to GB? Thank you for that question. We are unable to comment quantitatively, just to comment a little bit more on the question. The jump from Hopper to GB was the transition from air to liquid. In that transition, not only is the solution on the chip different, also the entire system that comes with it is very different. When we quote that three to five times jump in content, we're talking about that content on a systematic level, divided by the number of chips. When we're looking into future content per rack, broken down into price and quantity, we have heard and we have looked at different designs that are laid out by NVIDIA. As we've seen in GB300, they have attempted to roll out a fanless version. In that case, our quantity of cold plates will increase. On the price side, although content does not scale linearly with the amount of heat, it does correlate vaguely. As chips get hotter, the design becomes more difficult. That gives us the opportunity to generate more value for our customers, and therefore charge a higher content going into future generations. We continue to see growth, in both P and Q, going into the future, while we cannot comment numerically. Okay, the next question from Yun Fan, thank you for the question. The question is: outlook for 3C revenue in 2026, given iPhone also applies VC. Eric will take this question. We continue to see the phone industry have higher adoption rate of vapor chamber product. This is a trend, we are already in this game. We expect we have even better performance next year versus this year, because our long-term business target is always get into a sizable business, get more and more market share, we can have a deeper cooperation with client, including current product and including product they are going to plan in the future. We are still very optimistic about our vapor chamber growth in next year. Okay. Thank you, Eric. We are not seeing any more questions in the chat section. We're going to give everyone another minute, and if we still do not receive any questions, we will finish the meeting here. Thank you. Okay, we have a question from Winnie. In the past, we have some collaborations with an EV customer. Do we have project with their robot? Thanks. We are not going to be able to comment on individual customer and customer projects at this time. What we can say is we are working on EV projects, and we are already working on robotic projects with leading players in these areas. Thank you. Next question is: will other current competitors also be able to rank MCL quickly, or will the market share be split among the incumbent and yourselves only? From what we see right now, we are very confident in our own ability to deliver because of our expertise in both the cold plate and lid solutions. Whether or not our other competitors will be able to ramp MCL2 as fast as we can will be dependent on whether they have had relevant experience in the past, which we will not be able to comment on. Okay. There's another question from Jonathan. Does AVC have any upgraded cold plate solution to bridge TDP requirements from now till Rubin Ultra/Kyber? Yes, we do. We are working with client closely, and it's similar to current MCCP. That's why we continuously emphasize that technology itself, in most of the time, it's not a replacement, A replace B or B replace C. Lots of time, it's a continued enhancement, continued upgrade itself. Currently, what we see, the MCCP is still major product we work with client, and the client is satisfied with the performance. I'm going to add to that answer to say, when we're looking at Rubin Ultra specifically, we know that the current architecture means they want to change from a two die chip to a four die chip, therefore increasing the surface area of the chip. Even if TDP increases, if surface area also increases, if TDP doubles and then surface area doubles, that doesn't actually increase the thermal difficulty because the heat is spread over a larger area. When we're designing for thermal solutions, what we actually look at is heat density instead. That would have to be considered before. We would have to look at heat density as a key requirement rather than absolute TDP. Just adding to what Eric has described earlier, we have already been in discussion to extend cold plates to the Kyber rack. Next question from Yong Ri. Can we understand what would be the content of your thermal products in EV/robotics, say, compared to server? This, we will not be able to comment currently, but what we can say is that we will still be supplying similar solutions, heat sinks, fans, VCs, heat pipes, cold plates. These are all things that we are exploring with our customers, but we do not have a numerical value to share at this moment. Next question from Donna. Could you please comment on the near-term seasonality outlook into Q4 2025 and Q1 2026? We see a positive quarter in first quarter this year, and also first quarter next year. We see these two quarter, we still have a positive view. For the number, sorry, we are not able to share at this moment. Yeah. Just to add to that answer, we are continuing to see very good strength in our NVIDIA solutions, in our ASIC solutions, as well as in our consumer-grade solution, the ultra-thin vapor chamber. These are the key factors that have been driving our growth in the past few quarters. They are also still continuing to ramp up into the next two quarters. As Eric has described, we are expecting continued strength. Next question by Alex. Any comment on the competitive intensity from potential Chinese local players into NVIDIA/ASIC supply chain to take market share? What we see right now is you need to treat China and the U.S.A. as two totally separate market right now. If we are talking about China market, we do see large China local vendor. They are working very hard to try to penetrate into this industry. We are competing with them, but we feel comfortable because we have been working with all the major China CSP client for a long while, even more than 10 years. For U.S.A., it's another story. We see the majority competition is still coming from Taiwan-based company or major management team are relatively higher. Their major management team are coming from either Taiwan, either China, but, I mean, Mandarin. Yeah. Still, in U.S.A. market, we see majority competition is between Taiwan vendors. Thank you, Eric. All right. Going to give everyone another few seconds to post in any last questions, and then we will conclude this meeting. Okay. As we're seeing that no further questions have come through the chat room, we'd like to thank everyone for participating in our Q3 earnings call. Thank you, everyone, for attending, and we will speak to you in our next earnings call. Thank you.