United Microelectronics Corporation (2303) Earnings Call Transcript & Summary
October 28, 2026
Earnings Call Speaker Segments
Operator
operatorGood day, everyone, and welcome to UMC's earnings conference call for the third quarter of 2026. Every line is currently muted so that background noise doesn't interfere. Following the presentation, we'll hold a question-and-answer session. If you'd like to ask something, please follow the instructions that will be provided at that point. For your information, this call is being streamed live over the internet right now. A webcast replay will be ready within two hours once the conference concludes. You can find it on our website, www.umc.com, by going to the Investor Relations, Investors, Events section. At this time, I'd like to hand things over to Mr. Michael Lin, who leads Investor Relations at UMC. Mr. Lin, please go ahead.
Michael Lin
executiveThanks, and welcome everyone to UMC's conference call covering the third quarter of 2026. With me today are Mr. Jason Wang, who serves as CEO of UMC, and Mr. Chitung Liu, our CFO of UMC. Shortly, our CFO will walk us through the third quarter financial results, and after that our CEO will share key messages covering UMC's focus as well as guidance for the fourth quarter of 2026. When both our CEO and CFO have finished their remarks, we'll open things up for a Q&A section. UMC's quarterly financial reports can be found on our website, www.umc.com, in the Investors, Financials section. Over the course of this conference, we may make forward-looking statements that reflect management's current expectations and beliefs. Such forward-looking statements carry a number of risks and uncertainties that might cause actual results to differ materially, including risks that may lie beyond the company's control. If you'd like a more detailed description of these risks and uncertainties, please refer to our recent and subsequent filings with the SEC and the ROC security authorities. During this conference, you may view our financial presentation material, which is being broadcast live through the internet. At this point, I'd like to introduce UMC's CFO, Mr. Chitung Liu, to discuss UMC's third quarter 2026 financial results.
Chitung Liu
executiveThank you, Michael. I want to walk everyone through the third quarter 2026 investor conference presentation material, which you can download or view live on our website. Beginning on page 4, consolidated revenue for the third quarter of 2026 came in at NT$74.24 billion, with a gross margin rate of 34%. Net income attributable to the shareholder of the parent was NT$16.43 billion, and earnings per ordinary share were NT$1.32. Capacity utilization rate during the third quarter hit 92%, versus 85% in the prior quarter. Overall EPS for the quarter was NT$1.32. Turning to page 5, revenue rose 8% sequentially, reaching NT$74.2 billion. As we mentioned earlier, gross margin rate likewise improved, from 32.5% in the previous quarter to 34% this quarter, or NT$25.2 billion. EPS per US ADS is $0.206. For the year-over-year comparison on page 6, revenue was up 16.1% to NT$204 billion, and gross profit also increased 30.8% year-over-year to 65.3 billion, or 32% of the gross margin rate. Helped by the non-operating income, overall net income attributable to the shareholder of the parent rose 136.5% year-over-year, reaching NT$74.86 billion for the first nine months of the year. EPS for the first three quarters came to $6 in NT dollar terms, or $0.945 per ADS in US dollar terms.
Chitung Liu
executiveTurning to page 7, at the close of the third quarter our balance sheet showed total equity of NT$470 billion, and cash made up roughly 142 billion of that as of September 30th. Blended ASP in the third quarter was fairly steady and held firm versus the prior quarter, as you'll see on page 8. Looking at the revenue breakdown on page 9, the split by geography showed very little movement between quarter two and quarter three of 2026. On page 10, our breakdown of revenue by customer type was likewise essentially flat, with IDM at approximately 14% of total revenue and fabless making up the remainder. As for the breakdown by application, consumer held steady at about 32%, while communication and computer were also unchanged quarter-over-quarter. Page 12 shows that our 22- and 28-nanometer revenue continues to make up over one-third, hitting roughly 35% of the total pie chart, with 40-nanometer at around 16% and 65-nanometer at about 19%. Quarterly capacity kept rising modestly, driven mainly by our 12i facility in Singapore. For quarter four of 2026, we anticipate overall UMC capacity reaching approximately 1.33 12-inch equivalent — 1.33 million in 12-inch equivalent capacity by Q4 2026. CapEx remains unchanged from last quarter's estimate at around $2 billion.
Chitung Liu
executiveThat covers the highlights of UMC's results for the third quarter of 2026. You can find further details in the report, and that report is now up on our website. At this point, I'll hand the call over to UMC's CEO, Mr. Jason Wang.
Jason Wang
executiveThank you, Chitung. Good evening, everyone. What I'd like to do now is walk you through UMC's results for the third quarter. During the third quarter, demand was healthy across every market segment, which drove wafer shipments up 7.3%, to over 1.2 million 12-inch equivalent wafers — a record high. Our overall fab utilization rate came in at 92%, and several of our 12-inch fabs ran at full capacity, while 8-inch operations likewise experienced a strong recovery. Turning to technology nodes, our 22/28-nanometer business delivered another record-breaking quarter, with 22-nanometer revenue now making up 18.3% of wafer sales, as customers continued to tape out new products, among them display drivers, automotive radar, and Wi-Fi ICs. In our new business area, active customer engagement for advanced packaging and silicon photonics stayed robust. We expect that engagement to turn into accretive growth over the quarters ahead as we ramp existing projects and support new product tape-outs. Looking toward the fourth quarter, we expect this positive demand momentum to be sustained, supporting stable shipments and utilization rate. Revenue from our specialty technology portfolio should keep growing sequentially, helped in part by the ramp-up of our advanced packaging and silicon photonics businesses. Demand for these new business areas has risen significantly, and we are currently converting some existing capacity in order to address near-term unfulfilled demand.
Jason Wang
executiveLooking further out, the two-site build-out we have underway in Singapore and Tainan is going to be essential if UMC is to grow capacity and stay in step with what our customers require. We're happy to report that our board of directors gave the go-ahead today for an extra capital expenditure budget of roughly US$3.2 billion, and that lets us speed up the pace of that expansion as we keep matching our customers' product roadmaps and specifications. Turning now to guidance for the fourth quarter of 2026: wafer shipments should rise by a low single-digit percentage. ASP denominated in US dollars should hold steady. Gross margin should come in around the same level as the quarter before. The capacity utilization rate should land in the low 90% range. For 2026, our cash-based CapEx budget is set at US$2 billion. That wraps up what I wanted to say. Thanks, everyone, for listening. We're now open to your questions.
Operator
operatorYes, thank you. Ladies and gentlemen, the question-and-answer session is starting now. Should you wish to ask something of any speaker with us today, hit the star key plus the number one on your phone's keypad, and that puts you in the queue. Once your name is announced, go ahead and pose your question. Should it turn out that someone has already covered what you wanted to ask before your turn arrives, press star key and number two, and the question will be cancelled. Thank you. The first question comes from Sunny Lin at UBS. Please go ahead.
Sunny Lin
analystGood afternoon, and thanks so much for taking my questions. I have a few questions from my side. First, now that UMC is set to close out 2026 on a strong note, do you have any early thoughts on 2027 given what you can see today?
Jason Wang
executiveYes. For our early view of 2027, we expect the growth momentum established in 2026 will continue into the next year, supported by the healthy demand across our core business and growing opportunity in differentiated technology and high-value applications. If we look at this, in 2026, our growth has been driven primarily by high wafer shipments, improving utilization, and customer share gain. Looking ahead to 2027, we expect the further growth. Although with utilization already above 90% and several technologies operating at our near-full capacity, our shipment growth will increasingly be constrained by available manufacturing capacity. AI remains to a key growth driver, creating demand across power management, FPGA, and connectivity application. Meanwhile, the silicon photonics and advanced packaging are emerging as a new growth engine for UMC, addressing the next-generation optics connectivity, power delivery, and memory bandwidth requirement for AI infrastructure. Beyond AI, we continue to see opportunity from customer market share gain and increasing foundry share across premier smartphone, automotive, industrial, and other specialty applications, despite an uneven recovery in some end markets. With customer demand in selected technology already exceeding our available capacity, we see opportunity to capture greater value from our existing manufacturing footprint through improved pricing and richer product or technology mix in 2027, alongside continued shipment growth.
Jason Wang
executiveAlthough volume and utilization were the main drivers of 2026 growth, we anticipate that 2027 will increasingly be driven by value, with higher value per wafer and differentiated technology contributing more to UMC's revenue growth and long-term earning potential.
Sunny Lin
analystThanks a lot. That’s really useful and encouraging. On pricing, how should we think about the ASP environment for UMC, particularly as we head into 2027? And how are negotiations going with key clients for the next few quarters?
Jason Wang
executiveRight now we're in discussions with customers regarding pricing for 2027, and we're observing a notable change in the pricing landscape as we head into 2027, backed by stronger customer demand and tightening capacity in certain technologies. We see a clear chance to improve pricing in both our 8-inch and 12-inch operations, while keeping our disciplined, value-based approach to pricing. Importantly, we expect 2027 to be a turning point in our blended ASP trajectory, shifting away from an environment of ASP erosion toward a meaningful, positive uplift starting in the first quarter. While we anticipate a 10% increase in blended ASP year-over-year, this represents more than just an incremental pricing improvement. It reflects both a more constructive pricing environment and our growing ability to capture value through a richer product and technology mix. Our record-high 22-nanometer revenue contribution and continued momentum in specialty technology also give us a strong foundation, while silicon photonics and advanced packaging create additional higher-value growth opportunities as they scale.
Sunny Lin
analystThank you very much. And maybe one quick follow-up. You brought up a 10% year-over-year improvement. Is that first outlook for your blended ASP improvement heading into Q1?
Jason Wang
executiveNo — that figure will be measured year over year.
Sunny Lin
analystSo through 2027, 10% is what you're aiming for — that's not only for Q1?
Jason Wang
executiveTaking it all together, looking at 2026 against 2027, on a year-over-year basis — yes.
Chitung Liu
executiveGuidance on Q1 ASP will be shared when we hold the earnings call for the next quarter.
Sunny Lin
analystUnderstood. So regarding CapEx, you said there’s another $3 billion budget in addition to the $5 billion budget that was announced one quarter earlier. Could you give us more detail on what that means for capacity expansion over the next few years? And on silicon photonics and advanced packaging, what’s your view of overall market demand in the years ahead?
Jason Wang
executiveWe're speeding up certain investments because a number of our new growth businesses are transitioning out of technology development and into commercial execution, and that's happening faster than the capacity roadmap we originally laid out anticipated. We're watching three things occur simultaneously: the market is growing as AI infrastructure pushes up requirements for bandwidth, power delivery, memory, performance, and system integration. Our customer engagements are likewise advancing out of qualification and into production ramps and capacity planning. Meanwhile, our solution roadmap keeps moving ahead, with one product already in mass production and additional product completing tape-outs, customer validation, and a new generation of product already under development. What we're focused on now is no longer whether this opportunity will materialize but rather how quickly we can extend capacity to meet the customer demands. Under the expansion roadmap we have today, we're targeting monthly capacity of approximately 18k wafer for silicon photonics, 22k for DTC and silicon capacitor, and 20k for TFLN modulators. This next phase of capacity will begin its ramp in late 2028 and reach its planned scale by 2030. This isn't a broad-based capacity expansion. We're directing our capital toward customer-backed growth opportunity that will fuel UMC's next phase of growth. That should give you a flavor of why we're accelerating and a milestone view of how we plan for those CapEx.
Jason Wang
executiveOn your question about silicon photonics and advanced packaging: we think UMC is positioned well to seize the next wave of AI infrastructure opportunity, thanks to our competitive silicon photonics performance along with proven manufacturing execution and advanced packaging capability. Where silicon photonics is concerned, our data show that our PIC technology beats peers on key parameters, including propagation loss, and that is backed now by high production yield in our manufacturing. Our 12-inch PIC platform already supports 1.6T optical applications, and our 8-inch TFLN platform has finished 1.6T system validation with industry-leading yield. Building on that, our roadmap stretches toward 200G — 200G per lane and in the end a 400G per lane solution for next-generation 3.2T applications, enabled by hybrid bonding, TSV, and chip-to-wafer integrations. On the advanced packaging side, we are advancing in two key solution areas: a DTC for power delivery and customized memory stacking. Our discrete DTC solution for mobile entered mass production this quarter. For AI and server applications, our substrate-embedded DTC solution has completed tape-out and been functionally verified. And our customized memory stacking solution has passed initial customer verification, addressing the growing bandwidth requirement of AI inference. All of these achievements show UMC's ability to turn technology innovation into commercial execution now.
Jason Wang
executiveMerging device performance with manufacturing know-how and tight collaboration with customers lets us broaden what we do in next-generation AI infrastructure, assisting the customer in tackling three essential requirements: quicker optical connectivity, power delivery that is more efficient, and high memory bandwidth. That is the reason we think we sit in a strong spot to seize those opportunities.
Sunny Lin
analystThanks. It certainly seems as though your capabilities are quite competitive. Then, when it comes to both silicon photonics and advanced packaging — for EDTC or silicon capacitors, for instance — a number of your peers are likewise adding more supply. If AI demand were to slow at some point, should we be worried about possible overbuild? How ought we to think about that overbuild risk?
Jason Wang
executiveWe tend to view this in two ways, correct? The first is demand. To start, we're witnessing extremely robust demand fueled by applications tied to AI for silicon photonics as well as advanced packaging. On silicon photonics, we've been collaborating with a top customer. Recently we hit important milestones as multiple hyperscalers qualified our 1.6T optical transceiver platform. Our 12-inch PIC platform supported that 1.6T application, while the 8-inch TFLN finished 1.6T validation along with the manufacturing yield already demonstrated. With the roadmap moving toward the next generation, and with the conclusion of advanced packaging, our discrete solution—as we mentioned—and our first industry discrete DTC solution for AI and server applications has finished tape-out with our key customer, and the customer's focus in discussions has already shifted from validation toward planning customer demand. Our present expansion plans are linked to those customer commitments where demand is concerned. What about the overbuild? Judging by what we see today, we don't think we're building ahead of the market. Demand for silicon photonics and related AI infrastructure is rising faster than qualified capacity gets added. We also keep watching our peers' expansion and the timing of new industry supply closely.
Jason Wang
executiveOversupply and overcapacity are really an industry-wide issue, not something that applies only to UMC. We examine how fast end demand is rising compared with the capacity the industry as a whole is adding. Given the customer demand and the industry capacity plan that we see today, we think the market can absorb the expansion we have in mind. In the meantime, UMC is adding capacity in phases to reduce risk. At the same time, we stay cautious and disciplined about the way we bring on capacity. Above all, our expansion is supported by customer commitment, as I mentioned earlier, and it is tied to customer product roadmaps, production ramps, and capacity requirements. Despite all of this expansion, we expect customer demand to outstrip our planned capacity, according to the customer, which highlights how strong the commercial opportunity ahead is and why we decided to speed up the selected investments.
Sunny Lin
analystUnderstood. That works. Apologies — one final question, if I may. You also mentioned a quarter back that AI-related sales would come to roughly $300 million for 2026, and your goal is to triple that, or grow it more than threefold, to beyond a billion by 2029. So given the faster build-out in both silicon photonics and advanced packaging, what's the right way for us to view your AI revenue by 2029?
Jason Wang
executiveRight. You recall that. Our AI-linked operations now cover a wide array of specialty semiconductor offerings — power management support, connectivity, FPGA, plus our expanding advanced packaging and silicon photonics. Every one of those applications is turning into a more significant growth contributor for us down the road. As we reported, we're adding capacity and putting money into technology and manufacturing infrastructure so we can meet all of that rising demand. AI-related revenue for us comes in at nearly $300 million in 2026. If we look three years ahead, we anticipate that AI will likely account for more than a quarter of our revenue at that point.
Sunny Lin
analystMaybe 25% of UMC sales by 2029?
Jason Wang
executiveYes.
Sunny Lin
analystUnderstood — many thanks. Apologies, allow me to return to the queue.
Jason Wang
executiveThanks.
Operator
operatorThanks. Moving on to Laura Chen from Citi. Please go ahead.
Laura Chen
analystYes. Hello there. Thanks for fielding my questions. Good afternoon, gentlemen. I just have a brief follow-up regarding that 10% pricing increase. Is that more of a blended ASP uplift, or is it a like-to-like rise going into next year?
Jason Wang
executiveYeah, it's a blended one.
Laura Chen
analystIt's
Jason Wang
executiveThis is more than just a small pricing gain. It captures both the better pricing climate and how we’ve increased capture on the other products, a richer product, and a different product mix, yes.
Laura Chen
analystUnderstood—that’s very clear. We also believe we have a range of different plans for applications tied to AI. Could you share more detail on that—specifically, how much capacity we’re planning for silicon photonics, including in Taiwan and also a lot in Singapore? And regarding TFLN, since we’re ahead of our peers, I’m wondering whether you can also update us on the capacity build-up plan, and perhaps in 2 years, how large a capacity are we aiming for? Thank you.
Jason Wang
executiveWith the current milestone plus the facility and tool schedules in mind, we are carrying out this expansion in phases, and we'll keep updating you on how it progresses. In the meantime, we continue to work closely with our suppliers. Our hope is that we can speed up and expedite those installations. Also, as I said earlier, when I look at the overall CapEx plan that our board approved, the capacity build-up comes to roughly 18k for silicon photonics, 22k for DTC, which is silicon capacitor, and 20k for TFLN modulator. At this stage, we expect the initial install to take place in late 2028, continuing into 2029, and to reach the scale by 2030. As for location, our focus is on silicon photonics at the Singapore facility, silicon capacitor at our Tainan facility, and TFLN in our 8-inch operation.
Laura Chen
analystAbsolutely, thanks a lot. I realize we'll likely discuss this further next year, but considering that utilization is already above 90% and pricing for next year looks more favorable, based on what we can see now, are we noticing any structural upside to gross margin?
Chitung Liu
executiveSure, that price increase will probably come in the first quarter, and we're also facing some headwinds from higher raw material costs, which we have to work hard to offset the impact of. It's our view that structural profitability remains intact. On EBITDA margin, certainly we'll see it in good shape if we compare to the previous cycle. Because of stronger-than-expected CapEx driven by stronger customer demand, depreciation will be higher, and as a result gross margin is unlikely to see near the previous high level. We'll maintain disciplined spending and try to capture the megatrend of the AI growth opportunity.
Laura Chen
analystAll right, thanks — that’s quite useful.
Operator
operatorThanks. We’ll take the next question from Bruce Lu at Bank of America. Please go ahead.
Haas Liu
analystYeah, thanks to the management team for taking my question. I think, just to start with the 10% year-on-year price increase, you clarified that it's going to be on a blended basis. Could you share, on a like-for-like basis, how much the increase is going to be? How much of that price hike is tied to passing through costs? And how much of it comes from the tighter supply-demand landscape?
Jason Wang
executiveAt a very high level, as I mentioned, we anticipate that 2027 is going to be a turning point. The ASP trajectory is going to shift from what has historically been erosion to an uplift. We're in discussions with our customers across the broad range of applications. Now, the conversation is actually more complicated, because, first, there are current contractual relationships, a portion of that, and second, there's a focus on different nodes and applications and technology. They vary. There's a mix of pricing adjustments, you know. So it's much easier to look at this on a blended basis rather than going into every specific area, since they are quite complex in terms of the mix, yeah. Okay.
Haas Liu
analystThe reason I'm posing this question is that, according to your guidance, utilization has improved 16% from the fourth quarter of last year to the fourth quarter of this year, while your pricing has stayed fairly stable. I'm just wondering why your customers all of a sudden chose to accept a 10% price hike in 2027, given that over the past couple of quarters you've been working to raise utilization by 16%, yet blended pricing has been comparatively more stable — and now you're seeing an inflection in that? Is it due to product mix, or cost pass-through, that your customers decided to share the burden with you?
Jason Wang
executiveRight, both. The product mix plays into this, and that partly reflects the inflationary cost increase. As Chitung mentioned earlier, we're seeing some rising cost headwinds as well as additional depreciation tied to our investment. We anticipate that, while overall customer demand keeps growing and capacity tightens on selected technology, the combination of all those leads to the conversation with the customer. I think customers grasp the industry-wide capacity tightness and also grasp the rising cost. In a variable space. That's why I think this is naturally a conversation we'll have with the customer at this point. We typically hold this on a yearly basis. Again, we began this conversation in Q3, and with alignment carrying into Q4, we're at a very late stage of those conversations. We have fairly high confidence in the guidance we just provided, yeah. Got it.
Haas Liu
analystThat's quite useful. My second question concerns what you currently anticipate for consumer and market demand in 2027, since the supply chain appears to indicate that unit demand for smartphones, TVs, and PCs might decline from a unit demand standpoint. Should that hurt your utilization and keep it around the low 90 percentage levels rather than rising further toward nearly full levels, could your customers ask for price discounts or push back against the price increase through 2027, and how would you handle that?
Jason Wang
executiveLet me sort of walk you through how we're thinking about utilization. Capacity utilization, we expect, will stay at a really healthy level. During this year, we already went up from 85% in Q2 to above 90% in the second half of 2026, and at this point several of our 8-inch and 12-inch fabs are running at full loading. We're seeing upside demand driven by AI, things like power management and FPGA, and today we can't meet that demand, with backlog PO piling up. We'll keep optimizing production efficiency by debottlenecking the key steps and transferring technology between different fabs so output is maximized. On the operational side, there are minor mismatches between our demand and capacity configuration. When you see less than 100%, that doesn't mean the fab isn't fully loaded. It's really more a matter of mismatch between demand and capacity configuration. We're working diligently to address that capacity configuration mismatch. We're also upgrading our manufacturing capability to support long-term product mix optimization. For instance, we're converting one of our 8-inch fabs to expand TFLN capacity. While we're doing that conversion, the loading will be lower, but that doesn't mean the facility isn't occupied and being worked on.
Jason Wang
executiveAs we reinforce the silicon photonics roadmap and go after UMC's growth down the road, every one of those matters for our future expansion. We're handling that by means of this effort. That being said, the 90-some percent doesn't mean there's vacant capacity; it's really more a case of a mismatch, yeah.
Haas Liu
analystYeah. Just a quick question before I go back to the queue — it's really encouraging to see that you approved CapEx, additional CapEx, to back your new businesses, including silicon photonics, TFLN, and silicon capacitor. Could you give us the timeframe for that spending over the next few years? Because last time you guided that 2027 CapEx could be up to like $3 billion, but now, with the incremental CapEx, could you just give an update on that? And relatively, how should we think about the economies of scale for your new businesses, including TFLN, silicon capacitor, as well as silicon photonics — that by 2029, when this capacity starts to ramp to a certain scale, would those businesses, would the gross margins for those businesses be comparable to the corporate average, or are they still going to be below because of the higher depreciation burden? Thanks.
Jason Wang
executiveWe can give you that update once things become clearer on our end. For instance, we have board approval, and at this point we're working closely with the supplier to ensure we can speed things up and move the schedule along faster. When that information is finalized, we'll be in a position to offer you a more concrete outlook regarding the milestones. In the meantime, we're bringing in these different capacities across various timeframes. Because negotiations are still ongoing, it's probably not the right moment to share that with you. At a high level, we feel quite comfortable about all of that demand. The only thing we're concerned about right now is whether we can get it to our facility quickly enough. Let us work on that, and as we head into Q1 of next year, we should be able to give you a clearer picture along with the next phase. Now, regarding the economics of each, I mean the scale and the benefits for our business, silicon photonics will ramp in 2027, with advanced packaging following in 2028 due to facility reasons. And I think all of these will be accretive to our current models, because right now they sit above our current corporate average, yeah, higher than our corporate average when compared to existing solutions.
Haas Liu
analystAll right. When this new business begins ramping up in 2028 and 2029, those operations ought to be delivering margins above the corporate average, even accounting for the higher depreciation.
Jason Wang
executiveYeah, that's the figure we've got in the model as of right now.
Chitung Liu
executiveIn the same situation, or for that same ramp, if you compare it with, say, the earlier 22/28, then naturally that curve appears better.
Haas Liu
analystUnderstood — that’s really useful. Thanks a lot.
Operator
operatorThank you. Just a quick reminder — if you'd like to ask a question, press the star key plus the number one on your keypad. Thanks. Our next question comes from Charlie Chan at Morgan Stanley. Please go ahead.
Charlie Chan
analystThanks. Hi, Jason. Hi, Chitung. Congrats again on those very strong execution results. My first question is probably for Jason — I know in the past I've raised some concerns about the non-AI portion of demand offsetting the AI upside, right? Realistically, are you seeing any kind of cuts from non-AI segments recently? Do you have a preliminary assumption for next year's PC and smartphone segment demand, and what's your observation on inventory in this end market? Thank you.
Jason Wang
executiveAt present, we anticipate that positive demand momentum will carry through from 2026 into 2027, Q4 and 2027. At a high level, the momentum we've observed is going to keep holding up. Clearly, AI is still the main growth engine. On the AI front, there is no concern. Past AI, we keep seeing opportunities from gaining customer market share as well as raising foundry share. That applies to UMC, since we mainly serve the premier smartphone, automotive, and industrial segments, and at this point we see a healthy recovery in automotive and industrial. I believe we have fewer worries about automotive and industrial. On the smartphone side, though, given that our exposure to mid and low-end smartphones is smaller, we think our exposure there is lower, even as we project that the market outlook in that segment is weaker than the others. In the near term, meanwhile, we also see some decline in computers, because demand is soft in the computer AI, LCD controller, DDI, and Wi-Fi areas. Communication, automotive, industrial, and AI remain — non-AI, communication, automotive, and industrial remain robust, but we do see some weakness appearing in the PC space, yes.
Charlie Chan
analystUnderstood. All of those soft spots are being taken into account. Thanks. My apologies, but I’d like a bit more clarity on blended ASP rising 10%, since I believe that covers both 8-inch and 12-inch blended ASP, and a bigger 12-inch mix will pull blended ASP down considerably, correct? I want to confirm — is it the 8-inch side where you have separate product lines that aren’t clear —
Jason Wang
executiveApologies — your audio cut out on my end. Could you say that again? Yes.
Charlie Chan
analystOh, okay. When you say blended ASP went up 10%, does that cover both 12-inch and 8-inch, or is it some other product mix? And you mean the 8-inch is up 10%?
Chitung Liu
executiveAs we reported, that's a blended ASP across the entire company. We didn't disclose the figure, though we do have an annual number for 2026. The 10% figure we're talking about applies to the whole company for the full year 2027. It should represent roughly a 10% year-over-year rise, covering both like-for-like price increases and a richer product mix.
Charlie Chan
analystUnderstood. All of it merged together.
Chitung Liu
executiveCombined, yeah.
Charlie Chan
analystAll right. You wouldn't want to split that out any further — say, if the 12-inch mix change were taken out, what would the like-for-like ASP increase look like then? Could you say anything about that? It's
Chitung Liu
executiveIt's quite a complicated process, and a few of the technologies are completely different, or a different product. It's really hard to give you a generalized idea. Probably 10% is the best indication for now.
Charlie Chan
analystUnderstood, thanks. The final one is likely a fairly open-ended question. Since we basically view the strength of AI demand as obvious, correct? I believe Jason mentioned that we aren't concerned about AI demand. Regarding the recent capital market view on demand for AI infrastructure, whether it's cloud financing, there's a ROI concern, correct? I'm not certain how visible this genuine AI demand is to UMC, particularly for the products you have exposure to. How can you be confident that it's what's called an undersupply? Just a somewhat open question, because the capital market's reaction seems not to believe that AI demand will be so sustainable into next year.
Jason Wang
executiveAI demand isn’t just general AI demand. When we discuss what we’re actually serving, it’s really about the selected technology and where we’re focused. Take the optical space, for example—I’d say the ramp is only just beginning, so it’s more of a possible replacement mode. From our perspective, silicon photonics is genuinely a new area that we’re entering, and the industry is entering it as well. The PIC platform we offer, the TFLN, is all tied to that. As the industry shifts from 1.6T into 1.6T and toward 3.2T in the future, we actually believe this demand will stay robust. Advanced packaging is really a solution where we talk about the DTC, a deep trench capacity solution, a capacitor solution, right? Those are more like devices that can be implemented across multiple different areas. The power integration—all this power will keep driving the industry forward. With all those solutions, I think the market will continue, and demand will remain robust, yeah.
Charlie Chan
analystUnderstood. Even if AI infrastructure slows down somewhat, that's essentially a technology migration—like moving from 1.6T to 3.2T, and so on. Those things will keep going. So you're not especially concerned about the overall pace at which AI demand expands.
Jason Wang
executiveSee, when you break it down, compute is really what fuels that demand, along with power and speed, correct? Actually, the optical speed requirement and the power integration requirement lean more toward forward-looking hybrid applications. This isn't about AI demand broadly. It's about technology migration that supports future growth and satisfies the system requirement.
Charlie Chan
analystGot it—that's very clear, thanks. I'm also on the same page that DTC, the deep trench capacitor, is a really important next-generation piece. Thank you, Jason, and thank you, Chitung. That's really helpful. Thanks.
Jason Wang
executiveI appreciate that.
Operator
operatorOver to Gokul Hariharan at J.P. Morgan. The line is yours.
Gokul Hariharan
analystHey there. Appreciate it, Jason and Chitung, for fielding my questions. I just want to get some clarity, Jason — you said long-term AI revenues could reach 25%. Did you mean 25% of the revenue we have now, which is roughly $9 billion, or do you expect that by 2030 it grows to 25% of revenues, which would likely be a far larger figure? I just wanted to clarify that.
Jason Wang
executiveI mean the revenue over the coming three years — the three years starting now, yes.
Gokul Hariharan
analystAlright, three years from now, that could represent 25% of overall revenue.
Jason Wang
executiveYes, exactly.
Gokul Hariharan
analystAll right, got it. Thanks so much. Would you be able to share any indication regarding how much revenue silicon photonics and the embedded deep trench capacitor might contribute next year, since those appear to be ramping earlier? And on silicon photonics, what aspirations do you hold for your market positioning, given that you're starting later than the competition but also have some unique advantages here?
Jason Wang
executiveRight now we're operating at full capacity, and as we've said, this year should land somewhere around the $300 million mark. As for the new addition, the vast bulk of that capacity won't begin to come online until late 2028. That means 2027 will still be constrained. While we attempt to optimize — as I mentioned, we're trying to reconfigure some of our facility — hopefully we can make some short-term gains, but the constraint remains in place for 2027. We should be able to shed a bit more light on this in the next quarter. Again, it won't be significant, since that capacity isn't available yet. That covers revenue for 2027. What was your second question? My apologies.
Gokul Hariharan
analystSecond, what’s your ambition when it comes to share of the silicon photonics market, considering that you’re entering somewhat behind your rivals, yet you’re also essentially moving directly to 12-inch, and you’re already beginning to see at least some test results that are comparable to or better than what your competition has? I just wanted to get a sense of where we might ultimately land in terms of silicon photonics market share.
Jason Wang
executiveSure, absolutely. The goals here are extremely ambitious, and our confidence is very strong regarding the choice we made to go straight to 12-inch rather than first working through 8-inch. At this point, I'd say that decision has genuinely paid off. Through our 12-inch PIC technology, we've already shown superior results across many of the critical parameters — both with our customer and with their end customer. As things stand, the product is undergoing validation and qualification by the end customer too. Once production yields reach a manufacturable level, we believe our 12-inch PIC platform is positioned to serve that customer's existing 1.6T optical application and to carry them into the next generation as well. Now, TFLN presents a somewhat different picture. In TFLN development, I'd say we're considerably ahead of just about everyone else. The customer, I believe, also recognizes the advantage of integrating a TFLN modulator with our 12-inch PIC solution. On top of that, our advanced packaging work is now bearing fruit as well, since with hybrid bonding and chip-to-wafer capability, those elements should help make us even more competitive — the combination of the silicon photonics solution together with the packaging capability places us in a unique position.
Jason Wang
executiveYeah, in that space we feel very competitive.
Gokul Hariharan
analystAll right, a quick follow-up regarding the TFLN engagement, Jason. Will the support mostly go to HyperLight, or do you already have a wider group of customers past HyperLight — which, as I recall, you've already put money into and announced publicly too?
Jason Wang
executiveWe made the HyperLight partnership announcement, and since HyperLight became an extremely early adopter of the TFLN, I think we feel quite good about that program. Because we have already shown results together with HyperLight, and we have witnessed very strong interest in the TFLN solution at this point—not only from HyperLight, but from numerous other customers in this space, yeah.
Gokul Hariharan
analystAll right. Turning to 2027 — with the capacity limit you have, do you have any sense of how much more capacity you can still extract overall, say for 12-inch? Is mid single-digit capacity growth achievable next year, or is it still potentially low single-digit next year? The
Jason Wang
executiveWell, I suppose my response is simply that there isn't enough. I mean, we'd like to boost that figure in any way possible. At present, the clearest way to put it is this: the answer is not enough.
Chitung Liu
executiveThat's correct. Yeah, very minimal, yeah. Yeah.
Gokul Hariharan
analystAlright, Chitung, one last thing I wanted to look at is the gross margin puts and takes. Clearly, the price increase — a blended ASP around 10% — is pretty positive. Carrying those utilization gains from Q4 across the whole of next year would also help. Could you give us a sense of how depreciation is going to shape up? I believe on the last quarter's call you said depreciation would be trending higher, but it seems the pace may be a bit more than that. Are we basically saying we'll stay in a similar gross margin range, or can we still squeeze out some gross margin expansion next year, even with heavier investments coming through? The
Chitung Liu
executiveDepreciation will definitely be higher, and it could be that way for a couple years in a row. At least for 2027, we're talking about low single-digit depreciation growth, and possibly more for 2028. For gross margin, that's going to be a big number to offset. Beyond that, we also have rising raw material costs, etc. As I said earlier, we feel more confident about having a healthy EBITDA margin structure. For gross margins over at least the next two years, depreciation along with the rising raw material costs is going to put pressure on our overall financial gross margin.
Jason Wang
executiveI'd also like to back up what Chitung said a moment ago. We anticipate ASP gains, we anticipate better utilization, and each of those things is going to help our profitability in 2027. At the same time, we expect significant rising cost headwinds plus extra depreciation tied to the investment decision we made. What matters most to us right now is keeping a healthy profitability profile while we keep investing in various differentiated technologies and back our long-term growth. Hopefully that gives you a bit of color on how we're looking at this gross margin question, yeah.
Gokul Hariharan
analystUnderstood. Many thanks. Thanks a lot, yes.
Operator
operatorThank you. Ladies and gentlemen, we appreciate all of your questions. That brings today's Q&A session to an end. I'll now hand things over to UMC's Head of IR for closing remarks. Thank you.
Michael Lin
executiveThanks for joining the conference today. We value the questions you’ve raised. As usual, should you have any further follow-up questions, don’t hesitate to reach out to <<email_hidden>>. Have a good day.
Operator
operatorThanks. That wraps up our call for 3Q '26, everyone. We appreciate all of you joining UMC's conference. A replay of the webcast will be available within two hours. Head to www.umc.com and look under the Investors, Events section. You can go ahead and disconnect now. Thanks once more. Goodbye.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete United Microelectronics Corporation transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to United Microelectronics Corporation earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in United Microelectronics Corporation or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.