Now, we will start the 2026 Q3 earnings call of Samsung Electro-Mechanics. Good afternoon. This is EVP Dong Woo Lee, head of IR and planning team at Samsung Electro-Mechanics. Thank you for joining our 2026 Q3 earnings conference call. On today's call, I am joined by CFO Sung Jin Kim, EVP Taegon Lee, head of strategic marketing, VP Kyu-Taeck Park, head of support team, component division, EVP Pan Bae Kim, head of support team, package solution division, and Hyun Jung Kim, head of support team, optic solution division. We will start with a presentation of our Q3 company level and divisional business results, followed by market trends and outlook by product before taking your questions. First, our 2026 Q3 results. Our Q3 revenue was KRW 3,763.3 billion , which is approximately 9% increase QoQ and 30% increase YoY.
The details regarding revenue increase/decrease factors by division will be explained later during the divisional results. Q3 operating profit was KRW 691 billion , which is approximately 57% increase QoQ and around 165% increase YoY. Pre-tax profit in Q3 was KRW 609.7 billion, net profit, KRW 468.7 billion . Next, in terms of our financials, as of end of Q3, total assets was KRW 17,061.2 billion , a roughly 3% QoQ increase. For major financial indicators, liability to equity increased 6 percentage points to 63%. Debt to equity was 33%, an increase of 5 percentage points from Q2. Equity ratio was around 62%, which is a 2 percentage point drop. Next are the divisional results and future outlook. First is a look at our component division. Component division's Q3 revenue was KRW 1,883 billion , a roughly 14% increase QoQ and 36% increase YoY.
Revenue growth continued across all applications once again in Q3, and our AI-related revenue jumped as high-end product mix increased around AI data center servers, network switches, and power modules. Q4 MLCC market is expected to see continued strong demand for AI data center-related high-end products. Accordingly, we plan to increase supply of AI-related MLCC, such as our small size, ultra-high capacitance MLCCs, and focus on early development of new cutting-edge products to maintain market leadership. We will continue to respond to major big tech customers' long-term supply agreement demands. Automotive MLCC demand is expected to remain solid, driven by xEV and ADAS adoption, and we will continue to drive up penetration in ultra-high capacitance and high voltage products. Next is the package solution division. Q3 revenue was KRW 939 billion , a roughly 22% increase QoQ and 58% increase YoY.
For the FCBGA supply of high-end products for data center-related server CPU, AI accelerators, and network modules increased, and supply of high performance automotive substrates tied to ADAS and autonomous driving also increased in Q3, resulting in revenue growth across all applications. BGA revenue also increased with growing supply of substrates for memory and ARM CPUs. In Q4, customer demand for high-end FCBGA for AI data centers is likely to continue to increase. Accordingly, we will focus on timely supply of our new high-end FCBGAs for server CPUs and AI accelerators, and increase capacity across our production hubs on schedule to actively capture customer needs and maintain our strong growth trajectory. Lastly, the optic solution division. The division's Q3 revenue was KRW 941.3 billion , a roughly 9% decrease QoQ, but a 3% increase on a YoY basis.
IT camera module revenue decreased QoQ due to soft demand from overseas customers, but we started mass production of the new 200-MP folded zoom module and increased our supply for foldable phones. Automotive camera module revenue increased, thanks to increased supply for the global EV customer's next platform, increased supply of high pixel all-weather camera modules for EV and robotaxis, and starting mass production of high pixel sensing modules for humanoids. In Q4, for IT camera modules, we will focus on on-time mass production of the high image quality slim camera modules for the strategic customer's next flagship, and for automotive, we will focus on timely supply of specialty products for the global EV customer's next generation platform and increasing supply to the Korean auto OEM. Also, to capture humanoid robotic-related demand, we will increase new design in activity by leveraging our automotive camera technology and developing integrated sensing modules.
That completes the presentation on our Q3 results, and now Head of Strategic Marketing, Taegon Lee, will talk about market trends and outlook by key product.
Good afternoon. This is Taegon Lee, Head of Strategic Marketing. I will share market updates and outlook for MLCC, substrate, and camera modules respectively. First, MLCC. MLCC Q3 demand increased QoQ across all applications, including IT, industrial, and automotive. For servers, demand increased along the entire AI data center-related infrastructure, including GPU, CPU, and network. Demand for high reliability MLCC, such as the high voltage products required for stable power supply, also jumped significantly. By application, data center MLCC recorded unprecedented growth, thanks to the strong AI server and infrastructure demand. Rapid increase in compute capacity has been driving rapid increase in MLCC content per AI chipset, leading to a significant boost in demand.
Also, the next generation AI super chips developed by global AI big tech companies went into mass production from Q3, which has significantly increased share of our high-end MLCC, and this trend is likely to continue in Q4. Automotive MLCC market also continued growth with increased penetration of xEVs and advances in autonomous driving, leading to adoption of more sensors and computing systems. Even in IT applications, MLCC demand grew, especially around high capacitance, high-end products, thanks to effects of new flagship phone launches, wider adoption of silicon capacitors in response to AI server power efficiency needs, and rapid increase in DDR5 modules. In Q4, MLCC demand is expected to grow across, once again, all applications, unlike the usual Q4 patterns. For data centers, number of dies for AI GPUs or ASICs continues to increase from two in 2026 to four dies in 2027, and eight dies after 2028.
Power consumption increase is in proportion, which means that MLCC capacitance requirements for GPUs or ASIC core power will also double each year. We have been referring to this as the AI MLCC's law and have been actively capturing growing demand for smaller size, ultra-high capacitance MLCCs. Our key product, the 1005 47 μF, is currently being supplied to all big tech new platforms and expected to see a roughly 8 x demand growth in 2027. We are closely talking with global big tech customers about their mid, long-term roadmaps and demand forecasts, and we're preparing to launch next generation high capacitance products according to the AI MLCC's law. A large part of our mid to long-term capacity has already been occupied by long-term supply agreements.
MLCCs for optical modules for high speed GPU links is expected to be SEMCO's new growth engine next year, with 3x revenue growth expected on a YoY basis. The 1.6 Tbps optical module requires MLCC capacitance per module similar to that of smartphones, and we are planning lineup expansions to target the increasing optical module MLCC content. More data centers are adopting the 800 V power systems, and 800 V DC-DC converters are likely to adopt the 1 kV plus high voltage MLCCs. This is the same specification as the MLCCs used already for automotive onboard chargers, and we can leverage our existing automotive product lineup to quickly capture this growing market. Increasing power consumption by AI chips is quickly generating growing demand for silicon capacitors to remove noise and stabilize power supply right next to the chips.
We are currently producing an embedded type, which are inserted in the interposer or main board, as well as a land side type, which are mounted on top or bottom of the substrate. As already announced, we have signed two-year supply agreements with global big tech accounts and plan to start mass production in 2027. We are currently engaged in supply talks with several other big tech customers right now, and expect to further increase our supply. In automotive MLCCs, the share of high capacitance products is increasing in order to suppress power noise and maintain stable current supply, as high performance SoCs and ADAS modules now use more power for compute. We have increased our design in activities with key SoC companies to engage them from the early platform development phase and ensure that our products are designed in.
Preempting the shift from the existing 12 V to 48 V power systems, we were the first in the world to develop the 100 V class, 100 μF high capacitance product, and has completed key customer's qualification process, securing new volume. The 48 V power system is being quickly adopted across not only automotive applications, but also new applications such as humanoid robots. We are focused on preparing the right product and technology to capture the early growth opportunities in not only automotive, but Physical AI markets. For IT applications, we are focusing on new qualification activities of the ultra small size and ultra-high capacitance MLCCs for the LSC of key SoC customers in anticipation of further advances in chip packaging technology.
Our plan is to secure volume early on through long-term supply agreements with key customers and to develop the new low ESL four-terminal MLCC products on time to maintain market leadership and revenue growth. On the memory side, with more detailed DDR6 standards becoming available and better visibility of SOCAMM2 specifications, we will preemptively develop and design in high temperature, low profile MLCC products for the next generation memory chips to lock in a strong position from standard development phase. Lastly, regarding our long-term agreements and pricing, we have signed or in the process of signing LTAs with multiple customers around high-end products, including high capacitance and high voltage MLCCs. This is expected to account for more than 60% of our next year MLCC revenue.
Regarding MLCC pricing, even though we have been proactively increasing capacity to meet AI demand growth, customer demand is growing faster than expected, and the tight supply situation is further aggravated by the capacity erosion caused by high-end MLCC production. With supply expected to get tighter, we will continue to strategically respond to pricing. Next is substrate. Q3 FCBGA market demand increased significantly QoQ, with demand for high spec FCBGAs from data center applications triggered by high performance AI model and frontier AI competition among global AI players. Our revenue also increased significantly around data center related applications such as server CPU, AI accelerators, and network switches. We have signed LTAs with a leading North America big tech company, and currently we are supplying various high-end products for data center applications, which we plan to further expand and increase.
Q3 BGA market demand increased significantly, driven by strong AI data center memory demand and pricing adjustments to reflect the supply/demand situation and higher raw material prices. Our BGA revenue also increased, thanks to increased supply of new products for the ARM CPU and increased supply for DRAM and data center SSD controllers. Q4 FCBGA market supply is expected to remain tight due to strong data center demand. Adoption of advanced packaging that mounts multiple chips on a single board continues, driving larger area and higher multilayer trends in FCBGA, leading to greater supply shortage of not only high-end substrates, but across all products in the substrate group. Considering the shortage in production capacity versus market demand and higher raw material prices, we are in price talks to improve our profitability and will aim to continue above-market revenue growth.
We have been leading this ultra-large size substrate market by combining our differentiating, embedding, and multi-core technology with our warpage control know-how. We are already mass producing server CPU substrates with embedding and multi-core technology and plan to double our supply volume in Q4. In 2027, next year, we plan to supply equivalent products for AI accelerators, network switches, and automotive applications in addition to the server CPUs, to significantly increase the revenue share of embedding and multi-core products. Also, the Sejong and Vietnam plants where new investments are taking place according to the mid, long-term product and technology roadmap aligned with global big tech customers, will be developed as the hub for next generation ultra-high specification products. Q4 BGA demand is expected to be similar to Q3 due to seasonality. That said, demand related with AEI servers and data centers is expected to continue growth even in Q4.
We will continue to actively respond to the technology inflection point of new APs for flagship smartphones to capture this new market and maintain our focus on SSD controller, cable, SOCAMM related demand for AI data centers where growth is expected to continue. We will also continue new designing activity in high growth potential sectors such as satellite antennas and industrial power module substrates. Lastly, our camera modules. Q3 IT camera module demand decreased QoQ due to sluggish smartphone sales and higher smartphone prices tied to increased memory prices, despite the positive effect from the strategic customer's strong foldable phone sales. But automotive camera demand increased QoQ with strong EV sales continuing. In Q4, year-end inventory adjustments would have some effect, but we are planning to supply camera modules for the strategic customers' new model and actively pursue revenue growth around high-end products, including aperture, large angle OIS, and folded.
Also, we will focus on early mass production of parts for the new North American customer onboarding next year and follow up with new applications and customers. Q4 automotive camera demand relevant to SEMCO is expected to be similar to Q3. We will focus on further expanding high pixel product lineup for next generation autonomous driving platforms and active design-in on new autonomous driving projects of key customers to drive continued revenue growth. From Q3, we started supply of camera modules for humanoid applications to a North American customer, leveraging our differentiating technology, including high pixel ultra-precision proximity sensing. We are currently increasing our supply capacity in North America in anticipation of future humanoid market growth and aim to lead the market and develop new business opportunities by leveraging our IT and automotive camera technology. Thank you.
[Non-English content]
Now Q&A session will begin. Please press star one, that is star and one, if you have any questions. Questions will be taken according to the order you have pressed the number star one. For cancellation, please press star two, that is star and two on your phone.
[Non-English content]
The first question will be provided by Hyunji Cho from DB Securities. Please go ahead with your question.
[Non-English content]
I have two questions regarding the MLCC business. First question is about the supply agreements. Since June, you have been announcing various long-term supply agreements with your customers on MLCC. Can you give us a bit more detail of the supply agreements and also your future plans regarding supply agreements? Second question is data points on your Q3 MLCC shipments inventory ASP and your fourth quarter guidance.
[Non-English content]
To answer your first question about our MLCC long-term supply contracts. We have continued in Q3 to sign long-term agreements with key accounts, especially for the 1005 47 μF and 1608 100 μF. These cutting-edge MLCC products are mainly used in data centers. Until now, we have either signed or are in the process of signing long-term agreements with about 50 customers for high capacitance, high voltage, these high-end MLCC products, as well as cutting-edge MLCC products that are in very tight supply currently. At this point, we expect more than half of our next year MLCC revenue to be generated from long-term agreements. Long-term agreements have many benefits. It lets us secure customer demand early on, but it also allows us to efficiently plan out our production for maximum productivity.
It also naturally leads to adoption of our new products in customers' next-gen projects. Therefore, it continues to contribute to boosting our mid to long-term growth.
[Non-English content]
[Non-English content]
To answer your second question about Q3 MLCC shipments, inventory, ASP, and Q4 guidance. In Q3, revenue and shipment increased across all applications of industrial, IT, and automotive with a significant jump in the high capacitance, high reliability, cutting-edge products for data center related AI accelerators and power infrastructure, combined with the launch of the new smartphone by global OEM customers, as well as continued xEV penetration and ADAS upgrades. Shipments have increased, and as a result, overall inventory has decreased. Blended ASP in Q3 increased double digit on a quarter-on-quarter basis, driven by increased share of high-end applications such as industrial and automotive, combined with increased supply of the 1005 size 47 μF for AI server applications, and increased sales of ultra-high capacitance 0603 10 μF and the 1608 100 μF, driving productivity mix improvements.
Q4 usually is a slow quarter due to IT customer inventory adjustments and seasonality, but this year we are expecting automotive MLCC demand growth to continue in Q4 and this to be supported by strong growth in cutting-edge MLCC demand for AI servers. Therefore, with increasing share of high-end products combined with higher ASP, we are expecting another significant revenue increase in Q4.
[Non-English content]
[Non-English content]
The following question will be presented by Kyung Hoon Park from SK Securities. Please go ahead with your question.
[Non-English content]
I have two questions. The first question is regarding your package substrate investment plan that you announced last September. You announced that you will be investing KRW 6.8 trillion in the Sejong site and Vietnam. Can you give us a bit more details about the scale of mass production, mass production schedules, as well as how much upside revenue increase you expect from this new capacity? The second question is about the AI-related MLCC demand, which is very strong right now. Given that strong demand, can you give us some updates on your line operation and capacity increase plans?
[Non-English content]
[Non-English content]
To answer your first question about our capacity investment plans. The larger size and higher layering of packaged substrates for AI data center applications has aggravated capacity consumption. On top of that, only a few companies have the technology to supply these special substrates. Also, customers seem to have realized that packaged substrates are a strategic component, and they want to build a resilient supply chain and long-term partnerships with suppliers for co-development. Amid this trend, SEMCO has been steadily increasing its AI data center revenue share and building a larger presence in the market. Recognizing the technology expertise that we offer, also our stable supply capabilities, key big tech customers have signed large-scale investments as well as long-term supply agreements with SEMCO for expanding flip chip BGA substrate capacity in Sejong and Busan in Korea, as well as Vietnam.
The project is targeting rapidly growing AI data center-related applications, and it plans to start mass production in stages starting from 2027 and start meaningful revenue contribution from 2028. The new capacity is expected to drive significant revenue growth for us and further strengthen our position. Meanwhile, even this large-scale capacity expansion won't be enough to fully satisfy the customer's demand, and we are continuing to receive requests for further supply volume. So we will carefully study the market situation and demand supply situation to prudently determine the timing and size of further investments.
[Non-English content]
To answer your second question about our MLCC line operation. As you mentioned, AI-related MLCC demand remains very strong, and the key AI server customers are signing long-term supply agreements. We have been making continuous investments preemptively to capture this AI data center demand. This year, we have continued to increase our supply capacity versus last year. The new plant under construction in the Philippines is scheduled to start mass production from next year, and it will be focusing on cutting edge high-end MLCCs. The Philippine plant will also leverage the latest AI and autonomous robot technology to achieve a next level of manufacturing competitiveness. The cutting edge MLCCs that are used for AI servers require advanced technology as well as stable supply know-how. Only a few companies are currently able to supply this, and capacity expansion does take a long time.
Considering all of this, we expect the supply situation to remain tight for the time being.
[Non-English content]
[Non-English content]
The following question will be presented by Junseo Park from Mirae Asset Securities. Please go ahead with your question.
[Non-English content]
I have two questions. First question is about your camera module business. With the increase in autonomous driving and humanoids, so-called Physical AI, the importance of camera modules has been increasing. Can you give us an update on SEMCO's current camera module business and how much growth it expects going forward? Second question is about the substrate business. SEMCO is differentiated by the fact that it has the ability to supply not only the packaged substrates, but also passive components together. This can be used for embedding substrates. Can you give us your current supply status and how much expansion do you expect from that embedding substrate business?
[Non-English content]
To answer your question about the camera module business, in the xEV market, the xEV market demand has been increasing with upgrades in autonomous driving technology, as well as wider adoption of driverless autonomous driving services, such as robotaxis.
We have been continuing to increase high pixel products for the next platform of the global xEV customer, and we plan to further strengthen our leadership in the autonomous driving market by winning design-ins in the new autonomous driving platforms of key customers. Robotics, which is also a growing sector. In robotics, we started mass production in Q3 of camera modules for humanoids for a global customer, and we are at the same time developing new products for the next model with further improved sensing precision. Robotics requires specialized technology depending on the use case and environment. Therefore, we are collaborating with several top-tier companies on next generation camera module development, such as integrated sensing modules using our lens and actuator technology to claim a first-mover advantage in the growing robotics sector.
[Non-English content]
To answer your second question about embedding substrates, power consumption and heating of AI chips is increasing. Therefore, stable power supply is becoming more and more critical to chip performance. Embedding substrates embed passive components inside the substrate, which minimizes power supply pathway for better power efficiency and stability. Also, the shorter signaling distance between components and chip also improves high-speed signal quality. This is why embedding substrates have emerged as a core solution for high-performance AI chips, and the market is rapidly growing, driven by customer demand. SEMCO is the only company in the world to supply passive components together, as well as embedding flip chip BGAs, which is a clear differentiating feature of our company. With the industry's top embedding technology, we are already supplying embedding substrates for the key big tech accounts and plan to significantly increase supply after 2027.
The high spec embedding substrates for data centers are expected to evolve by not only increasing the number of embedded passive components, but also starting to embed active components. We will increase our collaboration with key customers on developing these new products and maintain the leading position in this new market.
[Non-English content]
[Non-English content]
The following question will be presented by Woon-ho Kim from IBK Investment and Securities. Please go ahead with your question.
[Non-English content]
I have two questions. First question is about your CapEx. With the strong AI demand continuing, I am expecting your CapEx to actually increase more than what you had originally announced. Especially there was that KRW 6.8 trillion investment plan announced in September. Including all of that, can you give us an update on your CapEx plans this year? Second question is about your Q4 and 2027 guidance. Despite the FX not being positive to your performance in Q3, you have delivered very, very strong performance. With that in mind, can you give us your guidance for Q4 and next year?
[Non-English content]
Regarding the first question about our CapEx, we have been executing an active capacity expansion and supplementary investment program this year to capture the explosive increase in AI core component demand, including high capacitance, high voltage, high temperature MLCCs, and the high-end FCBGAs for AI accelerators and network. Also, at the same time, we are running an active investment program for new business opportunities, including the silicon capacitors, which already received the KRW 1.5 trillion supply contract last May, as well as glass substrates, which are aiming for a full operation in 2028. At the same time, we are collaborating with customers in investing in new capacity to meet the AI data center related MLCC and FCBGA demand. Considering the upfront investments that are necessary for these projects, including land purchases and infrastructure, our CapEx is expected to increase than originally expected.
Going forward, our approach to CapEx will be to continue to invest in alignment with customer demand and market situation, not only in the high growth AI server related areas, but also in areas that are necessary to guarantee our future growth.
[Non-English content]
[Non-English content]
To answer your question about our Q4 and next year guidance, looking back at Q3, as you mentioned, despite the negative impact from FX movement, AI data center component demand growth was strong. ADAS-related demand continued growth, driving significant sales growth in both MLCC and FCBGA, leading to a new record quarterly business performance by SEMCO. In Q4, the tight supply situation for both MLCC and FCBGA is expected to deepen versus third quarter, with big tech customers adding new next-generation AI server infrastructure, and this raising the bar for component requirements and specifications. Our plan for MLCC is to actively increase supply of high capacitance, high temperature, high voltage MLCCs used in these AI server CPUs, optical modules and converters, and to continue to sign long-term agreements with global big tech customers, including hyperscalers and semiconductor companies.
We will also focus on strategic capacity operation and pricing in line with the tight supply situation to continue our MLCC business performance growth. For the FCBGAs, we plan to actively expand supply of new high-end products, such as the embedding substrates targeting multi-core server CPUs and exercise strategic pricing, reflecting the tight supply situation to maintain top-line growth momentum driven by high-end product sales. Q4 is expected to deliver another out-of-the-ordinary QoQ top-line growth with product mix improvements, combined with higher ASP and AI data center related revenue mix further increasing for both MLCC and FCBGA businesses.
[Non-English content]
[Non-English content]
This momentum that we have been seeing until Q3 is expected to continue until next year. Tight supply of MLCC and package substrate is likely to continue next year with demand for high-end products, that is high capacitance and high voltage MLCCs and large area FCBGAs to continue grow and even grow faster, keeping ASP on an upward trend continuing for some time. For MLCC, we will focus on strengthening our market dominance by expanding new product development, such as high capacitance embedded related new products for data center applications, and focus on increasing the share of large size, ultra high capacitance, high voltage, high temperature MLCC for xEV and ADAS. We will continue to leverage LTA and market sensitive pricing. Also for silicon capacitors, we will enhance our revenue and profitability by adding new applications and new customers.
For package substrates, we will focus on increasing penetration of our new high-end substrates, such as the embedding substrates and substrates for AI accelerators among the big tech customer base. Also we will focus on maximizing revenue by actively increasing capacities supported by tangible customer demand. Pricing for packaging will also be strategically managed to further enhance our bottom line. Lastly, capacity expansion for mid to long-term growth will be executed as planned to secure a foundation for our future growth. Overall, considering all of this, 2027, next year, is expected to be a record-setting year for SEMCO, with significant growth achieved each quarter- after- quarter, starting from Q1. Thank you.
[Non-English content]
This completes our earnings conference call. If you have any further questions, please reach out to our IR team. Thank you very much.