Mediatek Aktie

Welcome to the MediaTek 2026 Second Quarter Investors Conference Call. Financial results and presentations for today's call are available on the Investors section of the company website at www.mediatek.com. And now I would like to turn the call over to Ms. Jessie Wang, Deputy Director of Investor Relations. Ms. Wang, please go ahead.

Good afternoon, everyone. Joining us today are Dr. Rick Tsai, MediaTek's CEO; and Mr. David Ku, MediaTek's CFO. Mr. Ku will report our second quarter results, and then Dr. Tsai will provide our prepared remarks. After that, we will open for Q&A. As a reminder, today's presentation will provide forward-looking statements based on our current expectations. The statements are subject to various risks and factors, which may cause actual results to differ materially from these statements. The presentation materials supplement non-TIFRS financial measures. Earnings distribution will be made in accordance with financial statements based on TIFRS.

For details, please do refer to the safe harbor statement in our presentation slides. In addition, all contents provided in this teleconference are for your reference only, not intended for investment advice. Neither MediaTek nor any of the independent providers is responsible for your any actions taken in reliance on content provided in today's call.

Now I would like to turn the call to our CFO, Mr. David Ku, for the second quarter financial results.

Okay. Thank you, Jessie. Good afternoon, everyone. Now let's start with the 2026 second quarter financial results. The currency used here is NT dollar and the average exchange rate for the second quarter was TWD 31.6 to USD 1. Revenue for the quarter was $152.2 billion, up 2% sequentially and up 1.2% year-over-year. Gross margin for the quarter was 46.2%, down 0.1 percentage point from the previous quarter and down 2.9 percentage points from the year ago quarter. The year-over-year decrease was mainly due to a one-time benefit in the year ago quarter.

Operating expense for the quarter were $47.4 billion compared with $46.2 billion in the previous quarter and $44.5 billion in the year ago quarter. Operating income for the quarter was $22.9 billion, down 0.1% sequentially and down 22.2% year-over-year. Non-TIFRS operating income for the quarter was $23.7 billion. Operating margin for the quarter was 15%, down 0.3 percentage points in the previous quarter and down 4.5 percentage points year-over-year.

Non-TIFRS operating margin for the quarter was 15.6%. Net income for the quarter was $24.6 billion, up 0.9% sequentially and down 12.3% year-over-year. Non-TIFRS net income for the quarter was $25.3 billion. Net profit margin for the quarter was 16.2%, down 0.1 percentage point from the previous quarter and down 2.5 percentage points year-over-year. Non-TIFRS net profit margin for the quarter was 16.6%. EPS for the quarter was $15.28, up from $15.17 in the previous quarter and down from $17.5 in the year ago quarter. Non-TIFRS EPS for the quarter was $15.71. A reconciliation table for our TIFRS and non-TIFRS financial measures is attached in our press release for your information. And that concludes my comments. Thank you.

Thank you, David. And now I would like to turn the call to our CEO, Dr. Rick Tsai, for prepared remarks.

Good afternoon, everyone. Thank you for joining us today. MediaTek's second quarter revenue achieved TWD 152 billion, exceeding the high end of our guidance range, primarily driven by solid business momentum across Smart Edge platform products. Second quarter gross margin was around the midpoint of our guidance range, in line with our expectations. Recently, the release of increasingly capable frontier AI models, together with the rapid transition towards agentic AI has become a key driver of compute demand across both cloud and edge AI.

Agentic AI, which executes a series of actions, including planning, reasoning, execution, and self-correction further increases workloads. This represents a compelling growth opportunity for MediaTek as we are well positioned to support both the scaling of AI data centers and the proliferation of agentic AI experiences across a broad range of edge devices. In data center, demand for customized solutions to pursue optimized performance per TCO and performance per watt at scale remains strong. Through close partnership with major U.S. CSP customers, we have successfully built our first AI accelerator ASIC with leading performance.

Production of this product is scheduled to begin in the fourth quarter of this year. We expect our data center revenue to exceed $2 billion in 2026 and will continue to scale substantially in 2027, primarily driven by increasing customer demand for our solution with superior TCO. With that, we now believe the 2027 TAM to be USD 80 billion, and we are raising our share target to 15% to 20% from 10% to 15% that we guided last quarter. Meanwhile, our second AI accelerator ASIC is progressing well with a meaningful upgrade in compute performance, which further optimized TCO. Additionally, with close collaboration with our advanced packaging partner, the yield and reliability of the second ASIC are on track for high-volume production in 2028.

And we are confident in capturing additional market share when this ASIC ramps up. Furthermore, we continue to deepen our engagement with several customers for data center ASIC opportunities, backed by our industry-leading IP, deep ecosystem partnerships and proven execution capabilities, we believe we can secure more design wins in the data center market. Building on our industry-leading IP portfolio, we offer best-in-class pre-validated subsystem solutions for memory, I/O and connectivity. This largely reduces design complexity and shortens time to market for our data center customers to scale from AI ASIC chips to full-scale system and platform deployments.

In addition, our development of 448G SerDes is progressing well, delivering best-in-class performance and powered with our co-packaged copper system solution. Looking beyond 448G, we continue to develop CPO system solutions on TSMC COUPE platform to enable next-generation system connectivity. In addition, we offer end-to-end 3.5D platform with best-in-class 3.5D IP, packaging and design flows for large data center design on 3.5D. Furthermore, we serve as a major enabler for data center customers in this increasingly complex global supply chain environment. Through deep design technology co-optimization, DTCO with TSMC and close collaboration with key advanced packaging partners, we leverage our experience in advanced node design such as 2-nanometer and strong engineering and architecture capabilities to develop high-performance ASICs across a broad range of very large chip sizes using CoWoS and EMIB-T technologies.

We also create significant value beyond semiconductor by orchestrating other key components throughout the supply chain such as memory and substrate to drive successful execution. Looking ahead, as data center architectures and broader semiconductor ecosystem continue to evolve rapidly, we are confident in our ability to support our data center customers' innovative road map as their trusted long-term partner.

With that, I will now talk about the recent business performance for our 3 revenue groups. In the second quarter, mobile phone declined 14% quarter-over-quarter and 20% year-over-year, primarily due to weak demand as a result of increasing smartphone POM costs. Mobile phone accounted for 41% of total revenue in the second quarter. Our view for global smartphone shipment remains unchanged, and we continue to expect the market to decline by about 15% in units this year. Specifically, we have observed that customers are prioritizing both flagship products with differentiated user experiences and entry-level products for price-conscious demand.

For the flagship segment, we will launch our 2-nanometer SoC in the third quarter to power customers' next-generation agentic AI models. SoC delivers strong agentic AI computation with a step-up in user experience at a more efficient cost structure. Together with ongoing engagements with global players, we will increase our presence continuously in the flagship smartphone market in 2027.

For the entry segment, we continue to support customers with our leading SoC portfolio and the memory utilization technologies to optimize their cost structures. As rising costs across the supply chain have become an industry-wide reality, we are taking pricing actions to ensure these increases are appropriately reflected in our product pricing. For the third quarter, we expect mobile phone revenue to be flat to decline mid-single-digit percentages sequentially as the ramp of our flagship SoC is expected to largely offset weakness in other segments.

Moving on to Smart Edge platforms. In the second quarter, this group grew 19% quarter-over-quarter and grew 26% year-over-year, accounting for 53% of total revenue. The sequential growth was primarily driven by our continued market share gains across connectivity, computing and automotive products as well as contribution from DRAM content in TV SoC.

Our strength in advanced computing and connectivity continue to help us gain shares in multiple areas. During Computex this year, we announced our collaboration with NVIDIA on RTX Spark, a new class of Windows PCs built for agentic AI applications and best-in-class graphics capabilities. The products will be on shelf for the holiday season this year. This represents a meaningful step in broadening our computing businesses and a strong validation of our capabilities in the high-performance CPU system integration. For the third quarter, as several new connectivity and automotive projects are scheduled to enter mass production, we expect Smart Edge platforms revenue to grow mid- to high single-digit percentages sequentially.

Moving on to Power IC. This group accounted for 6% of total revenue in the second quarter and grew 11% quarter-over-quarter and 6% year-over-year. The revenue strength in the second quarter was mainly from share gains in computing and data center businesses. For the third quarter, we expect Power IC revenue to be flattish sequentially.

Moving to the guidance. In the third quarter of 2026, we expect the revenue growth from Smart Edge platforms to offset the weakness in mobile phone and gross margin to be within the current range. We expect our third quarter revenue to be in the range of TWD 152.2 billion to TWD 159.8 billion, flat to up 5% sequentially and up 7% to 12% year-over-year at a forecasted exchange rate of TWD 32 to USD 1. Gross margin is forecasted at 46% plus or minus 1.5 percentage points and operating expense ratio is forecasted at 31%, plus or minus 2 percentage points.

With that, we now aim to achieve the high end of our full year revenue target range, which is high single-digit percentage growth in U.S. dollars. For gross margin, through our disciplined pricing strategy that reflects the rising costs throughout the supply chain, we aim to deliver our full year gross margin within the current quarter guidance range. To secure supply chain capacity and fuel our strategic expansion from AI ASIC chips to full-scale systems and platforms, the Board has approved a discretionary USD 5 billion financing budget today.

This flexible framework provides us with optionality when needed to agilely support our long-term growth and capitalize on massive data center opportunities. In summary, we will continue to focus on executing our growth strategy across both edge and cloud AI. Supported by our technology leadership, broad portfolio and strong ecosystem partnerships, we believe we are well positioned to deliver sustainable long-term value for our shareholders.

This concludes my prepared remarks. Thank you.

Thank you, Rick. Operator, we are now ready for Q&A. Can we please have the first question?

[Operator Instructions] First one to ask question, Gokul Hariharan from JPMorgan.

First of all, I think, Rick, you mentioned your second-generation ASIC is on progress for mass production in 2028. Could you talk a little bit about how does it influence your TAM estimate? I think you've already updated your TAM to $80 billion in 2027. But as this project starts to ramp up, how does the TAM start to look like? Definitely it feels like the compute component within that chip is much bigger than the first generation that you're going to ramp in the next quarter? And could you also talk a little bit about how much market share can you take within this family of projects? Given that you seem to have some confidence that the market share will continue to expand. Can you potentially be the majority source within this larger customer when it comes to 2028, 2029 kind of time frame? That's my first question.

Gokul, I understand your question. I think first, I want to point out that the second ASIC certainly is much more powerful chip from both performance per TCO and performance per watt point of view. So while we are not ready today to guide the 2028 revenue range through the TAM yet, but we are certainly confident our 2028 TAM will be quite a bit higher than 2027. And our share because of also the strong capability demonstrated by our first ASIC in 2028, we are looking forward to both the first ASIC and the second ASIC to be in production in the same year. And that we believe certainly will enhance our market share, well, I would say, in a very significant manner.

Do you see any line of sight into you becoming a majority source for your customer? Or you still think that you will be the smaller vendor when we think about the next couple of generations?

Well, Gokul, we are not going into this because we know there are a lot of, shall we say, reports in the market. What we are doing basically, as we have always been doing is to stay on the ground to solidly execute what we have committed to our customers as we have and build -- continuous building a very strong trust between us and the customer. And as such, we believe we will get our hopefully, better than fair share of the market share. So that's really what we are doing, and we will continue to execute that way. Thank you.

Got it. So second question is on your 448-gig 2-nanometer-based SerDes. Could you give us a little bit more color on where we are in that process? Like when do we have the IP ready to kind of talk to your customers, given I think the third generation of this family of ASICs is kind of right now seems to be in RFQ stage. And any other IP that you are kind of putting forth to the customer in addition to the strong supply chain support that you have when it comes to the key enablers for the next project, which is in discussion?

Well, 400G or 448G SerDes IP is really going well. We certainly expect that to be ready sometime in the next year, maybe second half of next year. We were having kind of a middle of the road SerDes now already well proven. I can say that. We have certainly very, I would say, really at least equivalent industry-leading D2D IPs in terms of performance and power consumption. But however, that's some IPs we are also providing leading-edge packaging technology capability for very large chip size applications.

In addition, our team, I think, is working with some other key partners to put all those IPs together, as I mentioned in my text, a subsystem with which our maybe some different data center customers can use to -- because those chips are all very complex, and people want to go to the system level as soon as they can, but we can provide the time to market needs to the customers, some different customers. So I think we are moving into -- compared to, say, a year ago or not to mention 2 years ago, the company is moving to not only provide very competitive IPs, but also integration of those IPs into a chassis format. So the customers or various customers can utilize for fast time to market and to achieve their complex design.

Next, I have a question, Sunny Lin from UBS.

Congrats on the very steady progress, especially for the cloud ASIC side. So my first question is I want to follow up for the second cloud ASIC project. And so one is in terms of timeline for mass production, should we be looking at early 2028? And also, how should we think about the execution? You mentioned pretty steady progress. But I guess the market has some questions around like tape-out schedule, whether it's on track for Q4 this year? And also, how should we think about Intel EMIB-T?

Yes, you can assume early 2028 for early production for second chip. And the progress both in design, tape-out date, and are on track. We are working really closely with our customers, especially with the experiences we both have learned during the first chip, we really know even more how to leverage each other's capability. in getting the design out on time. So I'm quite comfortable with that.

The back-end technology, the packaging technology, of course, is another key component of the whole complex ASIC chip. We work really, again, also well probably more than closely with our supplier in -- mainly in ensuring the yields from the substrate vendors continue to improve very well. The supply of the capacity, the cycle time from the -- by our supplier, but we're working on all those down to the operation details. So while those technologies are very challenging, we all understand that. And that's also -- those challenging technologies are also the ones that will enable the performance, the great performance per watt and per TCO for that chip. And we are, I think, on the way to make that happen. Thank you.

So sorry, maybe just to follow up on a bit more color regarding EMIB-T. So would it be fair to say I definitely feel that today, you are sounding a bit more constructive even more than like a quarter ago. And recently, like earlier this week, Unimicron in Taiwan, the substrate makers, they are also sounding more committed regarding EMIB-T. So can we conclude that overall EMIB-T technology is making good progress in terms of the yield rate improvement and also the technology maturity for mass production for 2028?

Yes, in short, yes, you can. Sunny, I think you're saying that probably better than I am. Thank you.

No problem. And my second question will be to follow up on your announcement regarding this USD 5 billion for financing budget regarding the supply chain. And so would you be able to provide a bit more color on what you are going to do? Are you going to support some of the expansions from your key suppliers and some of the key areas that you're looking to put into the investment? How should we think about that support that you just came out?

Okay. Sunny, it's David here. I think, first of all, I need to start with, we have a very strong balance sheet. I mean, currently, we actually have ample cash, $7 billion plus on our balance sheet. As you guys can see right now, the industry is moving very fast, especially for 2 elements. One is the whole supply chain situation and also the possibility for different AI ASIC business model. So what we get from our Board today is really just optionality, a flexing plan, a financing budget plan when we need it. So we can actually just power it up when we need it on top of our strong balance sheet. I think that's the general idea.

So should we take that as a flexibility that you can further strengthen relationship with key suppliers by maybe supporting their expansion?

Yes, yes.

Next, we have Haas Liu from Bank of America.

Congrats on the very great results. So first question is just on ASIC. You updated your expectation for data center sales contribution to be more than USD 2 billion this year and raised the related TAM to USD 80 billion in 2027. Would you be able to share more detail if it is mainly driven by single project or includes the other parts of the NRE contribution from other projects? And if you are going to see further upside on this market expansion, if you can secure more capacity support from your supply chain partners?

I think that's our first project. And currently, based on the supply chain situation, we feel comfortable to provide that capacity for the whole supply chain, which including the foundry, the substrate and also our customer side, the memory. I think supply chain for the revenue target both for this year and also for next year, I think we feel very comfortable.

Okay. Yes. And I think just a quick follow-up to this one is that if you can define the boundary of the TAM, is it just customer AI accelerator only? Or does it include other content like CPU or network switch?

Right now, it's accelerator only for now. Yes, our TAM estimate.

Okay. So $80 billion only includes AI accelerator?

And no HBM.

Okay. Got it. And second question is just regarding your first -- on your ASIC pipeline. You just mentioned in your prepared remarks that you are engaged with several customers. I think this is probably the first time you mentioned that you are engaged with several customers. Would you be able to share more detail on this, whether this just is more focused on Tier-1 hyperscalers or you are diversifying into Tier-2 hyperscalers and also enterprises as well? And when can we expect more design or manufacturing contribution from the other parts of the project?

There are several initiatives and projects going on, but unfortunately, we won't be able to provide detail right now. I think we will provide detail when it becomes -- when we're allowed to talk about that.

And now the line is open to Laura Chen from Citi.

So I'm just wondering that for these new high technologies, including like the process memory or CPO, et cetera, that's going to be the second project of your ASIC or other potential, like engagement with other customers? That's my first question.

I'm afraid we cannot disclose those details. What I can say is the technology, be they different form of memory or memory silicon or CPC, CPO, we firmly believe these are the technologies that we need and also the whole industry needs to move into the next-generation XPUs, AI accelerators. It's a matter of time. So what the company is doing is to invest heavily in building these capabilities, while we also continue to hire many key talent in order to make that happen.

And also just wondering, I understand that for the first time, you had generators. I'm just wondering that because of now, certainly you have much better visibility in terms of the component, the supply chain part. Would we see that there will be more market share potential, assuming a first generation product for the next year. And also I recall last time you talked about potential assuming that MediaTek can more of the dollar for AI [ accelerator ]. So are we, do we have any expectation of our potential market share or potential revenue contribution for the next year?

We certainly believe there is demand both in 2027 and 2028, I mean, a strong demand for the chips that we're building. And -- but we are also -- as I said just earlier, we always wanted -- are doing a very solid execution job. So what we know right now, for sure, what we have said earlier, but you can be assured that we are working with our supply chain partners and our customers and, of course, our own people to fulfill any and all the demand from customers in both years ahead.

Next one, we have Felix Pan from KGI.

First of all, congrats for a very solid and strong outlook for the ASIC and also more content value for the next generation. My first question regarding to the AI ASIC. I think in pre-Computex analyst event, MediaTek showcased -- you guys have the rack design solution as well. I just wonder, is there any time soon we can see any contribution or any time line we can see the contribution that you have more meaningful revenue contribution from your currently rack design service? That's my first question.

Yes. We are, I think, in further events now in building what we said pre-validated subsystem or I would say, similarly a rack design. Actually, we are having our own IP, be the D2D or with 3.5D packaging. Together also, we are incorporating NVLink Fusion connectivity onto the subsystem. We believe this is very -- can be very attractive for various kind of needs from different CSPs or enterprise customers potentially. So these are things that are ongoing. We are talking to potential customers with that in offer. We will provide more details when time is right.

Okay. My second question regarding to the price action. You mentioned you guys are going to take the price action to reflect the value and also inflate cost. But in terms of the gross margin guidance, it's pretty much flattish or maintained at current level. So is that just reflect passing through the cost? Or do we expect more positive trend in the next few quarters to reflect on the margin side based on your price action? And is that possible to rank by your 3 different segments, which one will have the largest -- the price hike by category?

I think for passing through the supply chain cost increase to our customers, I think the objective is really just to get the whole industry to share that the whole supply chain ecosystem situation. We are not trying to actually increase the price to increase the gross margin. So the goal is really to pass through the cost pressure and shares with our customers. For gross margin, the goal is actually just trying to maintain the gross margin rather than increase the gross margin. I think that's not our objective.

Okay. So is that possible to rank by the 3 business units, like which one probably we will see the higher price hike?

For that one, we won't be able to comment, okay? But in general, it's going to be pretty fair to all different business units because, again, the supply chain cost is actually pretty universal across all segments and those in general.

Next one to ask question is Charlie Chan from Morgan Stanley.

Congrats for a very good execution and outlook. So Rick and David, maybe first question is a little bit follow-up on this kind of foundry supply chain. So do you think you need TSMC CoWoS to take up your second-generation projects? Or it seems like the current EMIB-T is working pretty well, right? Do you still need that backup and need to book capacity at TSMC earlier? And for the third generation, I believe you also have some engagement, right? So do you think at that generation, you will consider to move back to TSMCs, for example, CoWoS larger and equal size?

First thing first, I think for the third generation industry, we won't be able to comment. I would say right now, probably less focus on the next 2 years, 2027 and also 2028 because we do believe the current tool project will provide pretty strong growth and also will keep us busy for the next few years. In terms of the packaging technology backup, I think we're always working on different packaging technology. Sometimes switching the second [ technology ] will take time. So right now, given all the positive reading and also the results from the second project, we are confident actually we're going to deliver this on time and start to have the volume production in 2028. I think that's the current plan.

Got it, David. So my second topic is always on this AI smartphone side, right? So again, Rick or David, do you think going forward, there will be kind of significant change of the chip design -- the way of the chip design or the smartphone business model given the AI. For example, we learned that your smartphone SoC competitor is introducing very, very aggressive wafer-on-wafer sort of packaging, so they can include more memory for the smartphone SoC -- is that example one.

And secondly, regarding business model, do you think in the future, you need to do some custom chip for LLM customer. And at some point, actually, the hardware could be subsidized and the volume for the smartphone -- AI smartphone will ramp up? Because over the past, 3 years, we have been trying to upgrade the hardware, but AI smartphone doesn't really take off, right? So I'm wondering whether there could be any business model change to stimulate the AI smartphone demand.

I think AI smartphone business, yes, you're right to talk about the last 2, 3 years. What we believe is now, at least in China market, you can see there's already a pretty major shift. For instance, toward -- I think a good example is the ByteDance's Douyin model. I think it's running very, very popularly in China market. And that, I think the example like that will drive, I think, in the coming years of chip design.

Saying all that, however, we have to also bear in mind the cost of the chip or we have to balance the needs of the compute capability and the cost of the chip. So we look at that architecture, that is architecture question. We look at the various architecture and try to decide from this generation next to the further one, of whether we can build a chip with very, very, very good computing capability to meet most -- the many, if not most, of the agentic AI requirements for the model makers, but also still at a cost structure that our OEM customers can still have their good business. So this is a very dynamic environment. We work on those technologies and -- but we want to also make sure our customers can still do very well with our chips financially. Thank you.

So Rick, are you comfortable to talk about future opportunity outside of China with LLM provider? Because as you know, one of the leading LLM vendor, they hire a lot of ex-Apple designers, technology leaders. But I feel like this kind of AI smartphone is kind of not just about China, it's kind of a global phenomenon. Are you okay to talk about the business opportunity there?

Yes, I can certainly comment in a more generic manner. We are certainly seeing various -- and I do mean various potential customers who are working on, I would say, probably different formats of the edge devices to meet or enjoy this agentic AI era. But I think the jury definitely is out, still out as to which format will win or maybe formats, more than one format. We do not know for sure. But what we do know is we have this capability in building the most power efficient and high computing capability, SoC chips to serve different needs. At the end of the day, we have those fundamental architecture capability and computing system capability and of course, the leading-edge process capabilities. Those -- the combination of which will satisfy, I would say, all of those demands from various formats and customers.

Next one to ask question is Robert Sanders from Deutsche Bank.

One of your competitors announced something called HMP, which is basically using LPDDR instead of HBM. Are you seeing any customers pursuing that kind of line? And similarly, on the tech road map, beyond 448-gig SerDes, do you see any road map? It doesn't seem like there is any sort of copper-driven road map out there from your main competitor. Do you see a potential to go up to higher speeds in SerDes? Or is that kind of the end of the road?

I think for the data center, I understand actually there are several different approach trying to replace HBM due to the supply issue. But so far, based on what we see for the hyperscalers who actually require absolutely high performance, I think HBM is still the most. So for the segment we're aiming on right now, we still believe HBM will be the mainstream, at least for the near term.

And on the SerDes?

Okay. I think SerDes right now actually for the 448G is on schedule. I think for the next generation for beyond that, most likely we need to go on the optical. So the optical side, we are working on that as well, and we have a product road map, which is actually in line and on schedule. So which is actually all we updated earlier. Yes.

Next one to ask question, Evelyn Yu from Goldman.

So my first question is actually also around the AI ASIC. So you mentioned in the last call that the next generation chip actually carries more value and higher pricing. But does that necessarily translate into a higher gross margin than the first project? And how wide is that gap of gross margin versus corporate average for the second-generation project?

I think in general, the gross margin generation after generation will be similar, okay? It will not be higher, but the dollar value and the scale will be much bigger. And compared to the corporate average, that's the same guidance we provided earlier, will be slightly dilutive compared to the current corporate gross margin. But on the operating margin side, once we start to scale, I think it will be accretive meaningfully and sizeably.

Yes. Actually, my second question is a follow-up on that. I think since day 1, you mentioned that this is operating margin accretive. So how should we think about the magnitude of that accretion? And on OpEx specifically, should we expect absolute OpEx dollar growth into 2027 and into even in 2028 to run at a similar pace as we've seen versus the past few years?

I think for the OpEx ratio due to the revenue growth strongly, I think the OpEx ratio will definitely coming down substantially. But in terms of the absolute dollar because we are still investing on so many new technology, I think it will actually increase slightly. It will increase slightly. But the ratio will come down substantially in terms of ratio. In terms of operating margin accretive, we will probably show that until the third quarter. I think once we see that more clear for the overall operation plan, we'll provide some guidance. But right now, actually, we probably only provide the directional guidance rather than the number guidance.

Got it. So is it fair that we assume the operating dollar-wise OpEx dollar will grow higher versus...

Yes. It will grow higher.

But our ratio will come down. Yes.

Ladies and gentlemen, that concludes our Q&A session, and thank you for all your questions. I'll hand it over to Ms. Jessie Wang for closing comments. Ms. Wang, please proceed.

Ladies and gentlemen, this concludes MediaTek's 2026 Second Quarter Conference Call, and an audio replay will be available in 1 hour after the call at the Investors section of MediaTek's website. We would like to thank you for your participation, and you may now disconnect.

Yes. Thank you again for your participation in today's conference. You may disconnect now. Thank you, and goodbye.

Welcome to the MediaTek 2026 First Quarter Investors Conference Call. Financial results and presentations for today's call are available on Investors section of the company website at www.mediatek.com.

Now I would like to turn the call over to Ms. Jessie Wang, Deputy Director of Investor Relations. Mr. Wang, please go ahead.

Good afternoon, everyone. Joining us today are Dr. Rick Tsai, MediaTek's CEO; and Mr. David Ku, MediaTek's CFO. Mr. Ku will report our first quarter results, and then Dr. Tsai will provide our prepared remarks. After that, we will open for Q&A.

As a reminder, today's presentation will provide forward-looking statements based on our current expectations. The statements are subject to various risks and factors, which may cause actual results to differ materially from the statements. The presentation material supplement, non-TIFRS financial measures earnings distribution will be made in accordance with financial statements based TIFRS. For details, please refer to the safe harbor statement in our presentation slides.

In addition, all contents provided in this teleconference are for your reference only, not intended for investment. Neither MediaTek nor any of the independent providers is responsible for any actions taken in reliance on content provided in today's call.

Now I would like to turn the call to our CFO, Mr. David Ku, for the first quarter financial results.

Thank you, Jessie. Good afternoon, everyone. Now let's start with the 2026 1st quarter financial results. The currency used here is NT dollar and the average exchange rate for the first quarter was TWD 31.62 to USD 1.

Revenue for the quarter was TWD 149.2 billion, down 0.7% sequentially and down 2.7% year-over-year. Gross margin for the quarter was 46.3%, up 0.2 percentage points from the previous quarter and down 1.8 percentage points from the year ago quarter. Operating expense for the quarter were TWD 46.2 billion compared with TWD 47.4 billion in the previous quarter and TWD 43.8 billion in the year ago quarter.

Operating income for the quarter was TWD 22.9 billion, up 4.8% sequentially and down 23.8% year-over-year. Non-TIFRS operating income for the quarter was TWD 26.6 billion. Operating margin for the quarter was 15.3%, up 0.8 percentage points in the previous quarter and down 4.3 percentage points year-over-year. Non-TIFRS operating margin for the quarter was 15.8%.

Net income for the quarter was TWD 24.4 billion, up 5.6% sequentially and down 14.4% year-over-year. Non-TIFRS net income for the quarter was TWD 45 billion. Net profit for the quarter was 16.3%, up 0.9 percentage points from the previous quarter and down 3 percentage points year-over-year. Non-TIFRS net profit margin for the quarter was 16.7%.

EPS for the quarter was TWD 15.17, up from TWD 14.39 in the previous quarter and down from TWD 18.43 in the year ago quarter. Non-TIFRS EPS for the quarter was TWD 15.52. A reconciliation table for our TIFRS and non-TIFRS financial measure is attached in our press release for your information.

And that concludes my comments. Thank you.

Thank you, David. And now I would like to turn the call to our CEO, Dr. Rick Tsai, for prepared remarks.

Thank you. Good afternoon, everyone. Thank you for joining us today. MediaTek's first quarter revenue came in at the high end of our guidance range, thanks to the steady demand from our diversified platforms and a more favorable growing exchange rate. First quarter gross margin was around the middle point of our guidance range.

During the quarter, the rapid adoption of agentic AI use cases such as [ open claw ] AI agents and other compute-intensive workloads have marked the inflection point for the industry. The acceleration of computing requirements under the ubiquitous AI megatrend are not only broadening the addressable market for edge devices, but further expanding AI infrastructure investment.

Our differentiated combination of technologies and product propels us to meaningfully benefit from structural growth in both the edge and the cloud. In edge devices, building our strong positions across numerous markets, we continue to invest in and deliver advanced SoC and faster connectivity solutions spanning mobile, computing, automotive, IoT and beyond. I will elaborate more in the business discussion sections.

In data center, demand momentum is particularly strong with our comprehensive and solid capabilities across design, integration and supply chain execution. Our first AI accelerator ASIC project for a U.S. hyperscale customer is progressing fairly well. We are on schedule for production and now expect AI ASIC business to contribute around USD 2 billion in revenue in the fourth quarter of this year.

For 2027, based on the capacity we have now secured, we are very confident that this project will scale to multiple billion U.S. dollars. Meanwhile, design for the other AI-accelerated ASIC project is currently underway in close collaboration with our customers and supply chain partners with mass production targeted to start by the end of 2027. Furthermore, we are actively engaged in several new data center ASIC opportunities. some of which are already in the final stages of discussion. We are confident that our proven capabilities can address the rapidly growing AI demand of the market and deliver strategic value to customers.

Meanwhile, we continue to invest significantly and have made tangible progress in building best-in-class, system-level technology road map for future data center architecture. On the optical front, to address the industry trend towards leveraging silicon photonics to achieve higher bandwidth density, we made a USD 90 million investment this quarter in [ Air Labs ], a leader in optical engines for CPO as part of our co-packaged optics partnership.

We also announced a successful design with Microsoft Research for the next-generation active optical cable powered by micro LED light sources, which can significantly improve power efficiency in data centers. Furthermore, we are making solid progress in the development of enabling technologies for future generations of data centers with successful test vehicles.

We are very proud of our status in building next-generation 400G high-speed service, 64G die-to-die interconnect and advanced 3.5G platform, which covers IP design enablement and advanced packaging technology to realize design with much larger [ graphical ] sizes.

Additionally, our development in custom HBM solutions and integrated voltage regulators, IVR, are also advancing as planned. These ongoing investments will further improve performance per watt and power efficiency of multiple kilowatt FPU in future data centers and further strengthen our long-term technology road map, positioning us to support the evolving requirements of next-generation data centers and AI workloads.

With that, now let me talk about the recent business performance for our 3 revenue groups. In the first quarter, our Mobile Phone revenue declined 17% quarter-over-quarter and 15% year-over-year, accounting for 49% of total revenue. The current concentration of industry resources on data centers has led to elevated costs for smartphones. Our customers have raised retail prices and shifted their product mix toward higher-end models to manage that.

Given such adjustments will inevitably lower the market demand for smartphones, we expect global smartphone shipments to decline by about 15% this year. For the second quarter, as our customers stay cautious while pending further visibility into end market demand, we expect Mobile Phone revenue to decline sequentially, and we will continue to work with our customers and navigate through this period together.

Nevertheless, the indispensable role the smartphone play in our daily life, coupled with accelerating agentic AI innovation, supports our view that the smartphone replacement demand will recover when the supply conditions normalize. Therefore, we continue to invest in and empower more advanced smartphone.

We have already secured several design wins for our next-generation flagship SoC, which is our first 2-nanometer product, featuring a meaningful step-up in user experience, enabled by stronger computing and AI capabilities. Smartphones powered by this 2-nanometer flagship SoC are scheduled to hit the market by the end of the third quarter, and we anticipate our Mobile Phone revenue to improve in the second half of the year.

Moving on to Smart Edge Platforms. In the first quarter, this group grew 23% quarter-over-quarter and grew 13% year-over-year, accounting for 46% of total revenue. Our global market share gain were a key driver of both sequential and year-over-year growth. We also saw a sequential demand recovery as demand rebounded from a seasonally weak fourth quarter.

We are pleased that several new projects that we have cultivated over the past few years in collaboration with top-tier consumer brands, notebook makers, telecom operators, car makers and CSPs have begun or are expected to start volume production in 2023. We expect Smart Edge Platforms revenue excluding the contribution from the new data center ASIC project, to grow by double-digit percentage this year. This is mainly driven by our global share gain across connectivity, computing and automotive products and, to a lesser extent, higher blended ASP for certain TV SoCs to reflect the increase in DRAM costs.

As these drivers continue, we expect Smart Edge Platforms revenue to grow sequentially in the second quarter. In the future, given our leading technology and product road map, together with agentic AI trend discussed earlier, we are seeing structural growth opportunities for edge devices. The strong lineup of our advanced computing solutions allows us to expand in multiple areas, such as high-end computing devices and automotive.

Specifically last week at Beijing Auto Show, we showcased our 3-nanometer Dimensity Auto agentic AI cockpit solution which, enables active and intuitive in-vehicle agent features for both drivers and passengers. We have received very positive feedback from customers. Moreover, we will be leading the automotive industry to migrate to 2-nanometer process technology to enable more innovation.

In addition, building on our strong connectivity foundation, including Wi-Fi 7, 5G modem and 10G PON, we continue to lead the market with next-generation Wi-Fi 8 and 5G NGN satellite solutions to achieve faster and more seamless connectivity. We believe the combination of strong computing and connectivity capability will make MediaTek a strong partner for agentic AI device makers.

Moving on to Power IC. This group accounted for 5% of total revenue in the first quarter, increasing 14% quarter-over-quarter and 11% year-over-year. In the second quarter, we expect Power IC revenue to be flattish.

Moving to the guidance. In the second quarter of 2026, we expect the revenue growth from Smart Edge Platforms to partially offset the weakness in our smartphone business.

Quarterly gross margin is expected to be stable within the current range. We expect our second quarter revenue to be in the range of TWD 140.2 billion to TWD 149.2 billion, flat to decline 6% sequentially and down 1% to 7% year-over-year at a forecasted exchange rate of TWD 31.5 to USD 1. Gross margin is forecasted at 43% plus or minus 1.5 percentage points, and operating expense ratio is forecasted at 31% plus or minus 2 percentage points.

For the full year, we expect revenue to increase by mid- to high single-digit percentage in U.S. dollar year-over-year. Meanwhile, through disciplined pricing strategy, we aim to sustain our full year gross margin within the current quarter guidance range. We will continue to build long-term growth in the cloud. We are highly focused on disciplined execution across our growth opportunities and expect these efforts to drive significant operating leverage over time.

This concludes my prepared remarks. Thank you.

Thank you, Rick. Operator, we are now ready for Q&A. May we please have the first question?

[Operator Instructions] And our first question will be Bruce Lu, Goldman Sachs.

Congratulations for the recent progress of the ASIC project. But I want to ask more about your next project. So after you successfully demonstrate your IO technology, which is already proven at the Tier 1, can you do more? I mean, can you do compute die or other functional die for your customers for the next project?

Also for the ASP, I see you provide a lot more value than comparing to the first year. Recent research suggests that the pricing is meaningfully higher than the first one. It seems to be that it's too meaningful to be true, but I want to get some color in case I made a mistake in my assumption.

Also lastly is for the gross margin and operating margin. As you provide a lot more value, can we assume like higher gross margin, operating margin compared to the first project? So in summary, my question is like, can you do more than IO die for your next project? And can you update the ASIC TAM in 2028?

Again, first, we cannot comment our customers' chip architecture, clearly. However, what we can comment is we are adding more value in various ways, including silicon and packaging, both. And of course, together with our IP, die-to-die, et cetera. As you probably can imagine, the second chip is -- the second chip in the next generation and certainly will -- and if you look at the track record of the previous generations, again, it is fair to view the second-generation chip to a much stronger, also bigger chip.

So the value covers from both silicon and packaging. We are gaining values in both areas. And with the value going up, we certainly, I think, will have a higher pricing of such much more powerful chip. So I think this is -- while, as I stressed before, we continue to work with our customers to also enhance their own capability. That's always at the heart of our business model with our customers.

It's a win-win situation. We, together, add a lot of value to their chip, and the customers can utilize those powerful yet our efficient chips to implement in their data centers and provide a lot more value to their customers. So I think that's basically how I would address your question. Thank you.

But do you have the updated number for the addressable market for ASIC? Because I think last time you were talking about like, I think it's $60 billion, $70 billion, you were talking in 2028. Do you have the updated numbers?

Yes. Yes, I mean, we have done kind of a first order estimate. The market is changing so fast as you can attest from just less than 24 hours ago, or the announcement from the 4 major CSPs. Everybody can see that the demand for the data center infrastructure, data center computing power just continues to grow and, if anything, continue to accelerate.

So our estimates, in my mind, is first order in nature, and I don't know if it's a conservative start, but it's not really aggressive. We now view the market size to be about $70 billion to $80 billion in 2027. That's our view now. Thank you.

Okay. My next question is for the packaging solutions for the ASIC business. It seems to me that MediaTek choose a different path for your next chip, which is not the mainstream solution, at least not for now. So can you provide some color that how you manage the execution risk? Do you have a backup plan if things are not moving in the right direction? Or you are comfortable with the recent technology progress at your parter? And I think this one is definitely one of the biggest concern from the investors. Any color would be appreciated.

Again, we cannot comment specific technologies that we are doing together with our customers. But I think suffice to say that we are investing into packaging solutions, packaging technology solutions, both because in the industry, the demand in the AI infrastructure being so demanding. Packaging is now becoming a really, really critical part of the overall total solution. We need to invest in various technologies so that we can better prepare ourselves for various demand from various customers.

And we understand your question. We believe a second approach of the packaging solution has merit, very good merit. Technology-wise, execution is getting quite good. Again, we are working closely with our partners for both packaging solutions, and we will provide a good high yield solutions to our customers. I have no doubt about that in my mind. Thank you.

And the next question, Sunny Lin, UBS.

Congrats on the very encouraging progress made for the cloud ASIC project. And so my first question is double check. Rick, did you just mention now you're tracking the addressable market for cloud ASIC to be $70 billion to $80 billion by 2027, not by 2028?

Correct. You're correct. Yes.

And you are still expecting share target to be 10% to 15%?

We keep that for now, yes.

Right. Good to know. So my first question is, given what the phased client released on the product last week, it seems like the project that you're working on should be for training. And I think that's a bit surprising is that should be a good setup in terms of the volume potential. And so now if your project is indeed for training, then how should we think about the [ fast spot ] of your project in terms of the overall volume within the client portfolio in 2027 and 2028?

Sunny, number one is the architecture of the customer's data center, of course, it's their decision based on, as I mean to say, there are two chips. Both chip roughly can do the job. It's more well rack topology architecture decision. But we saw that. Again, the market, if you look at the total market size, I just said, we know in our last meeting, we said $70 billion market size for 2028. And today, we said $70 billion to $80 billion in 2027.

It's much more -- the first order effect is the total growth, the rapid growth of the market, of the demand. But we have a tremendous chip together with our customers, and it will be used extensively. I have no worry about it. I wouldn't say what we worry, what we really focus on is execute to deliver the very strong demand from our customers, be they training or inferencing. So I think whether it's a T or I, the truth is the second quarter in nature.

No problem. Very clear. And my second question, if we think a bit beyond, and so you just guided up the expectation for 2027, and you just also mentioned the next project which should be run by late 2027, I believe. And so meaningful contribution going to 2028. So how should we think about the trajectory of the growth beyond 2027? And so either you are able to provide more value and then the product itself, the value is also a lot higher, then should we expect the growth into 2028 to also be quite significant? Like maybe you can double or even more than double? .

Sunny, I think in general, based on the 2 projects we are executing on, I think for the growth profile from this year to '27, '28, that's clear. But in terms of magnitude, especially we're asking for '28, I think right now, maybe a little too early because that's a combination of what's the availability for the overall capacity and also what's the customers actually demand profile.

But overall, we think about the year-over-year growth that is certain. But in terms of magnitude, I guess, actually afterwards -- right now, it's only April '26, I think it will be a little bit too early to talk about that. But if you're talking from perspective, I know we are positive about that.

Next one, Laura Chen, Citi.

Also congrats on the great outlook. My question is also similar to how should we think about the company's product mix profile in like a 2, 3-years perspective? Like Rick just mentioned, if we're assuming like USD 2 billion contribution for this year and also assuming that the same like a run rate quarterly into next year or even based on like 10% to 15% market share in 2027, that means that the accelerator revenue contribution could be 50% higher.

So I'm just wondering that in the longer term, how should we think about our product mix profile? The smartphone, of course, Rick also mentioned that we'll resume the growth momentum. But just for maybe a longer-term perspective, how should we think about the priorities and also business outlook for MediaTek? That's my first question.

Okay. Again, the macro environment, everyone, I think, has a pretty clear picture. I don't need to preach to the choir. There's no question that the "AI megatrend" continues now that not only from the infrastructure computing power point of view, but also from agentic AI point of view. So if anything, this thing continues to accelerate.

But it's probably also right to say that for MediaTek, the data center AI business and the revenue will grow quite a bit faster than our more mature SoC business. so for the mobile business, of course, this year is facing a pretty strong headwind because of the DRAM supply and pricing. That's being said, we view mobile to resume its replacement cycle in a year or a little more than a year. But the growth momentum for the mobile business, I think, overall, it's going to be lower than the data center AI business.

We expect, however, the agentic AI disruption will bring in new revenue for the edge devices, including mobile phones. So on that front, we are quite optimistic, and we will be aggressive in pursuing that new revenue. Everything being said, I still feel data center AI revenue growth will have a higher rate compared to our other SoC business. Thank you.

My second question is also about the business model. You work with your CSP customers. We know that it seems like the trend of the COT customer-owned tools is also kind of emerging. We understand that MediaTek can provide more value, maybe just not the service going forward and also more of the technology integrations, back-end design, et cetera. But just wondering your view that how the COT business model could impact the MediaTek profit margin or the value of your future design?

Truthfully, I think this market is growing fast with size, with scale. So yes, the first quarter issue for me or for us is to expand our presence, our penetration in this continuing accelerating large market. And we will -- I'm very confident, we will also be getting more design wins in the market. And overall, in different cases, I believe we will provide, how should I say, various degree of value for different customers depending on customers also.

And so in general, it's quite a -- overall, if we look at our revenue and margin, both growth and operating, I think we will do very well. I don't really worry about that. Thank you.

Next one, Charlie Chan, Morgan Stanley.

Rick, David and Jessie, Just wanted to follow up on the ASIC really question. So I believe MediaTek can really kind of add value for next-generation 2-nanometer project. But just in terms of financial assumption, given very large scale of revenue, are we going to see gross margin decline? Whereas the operating margin, net margin probably can keep at the same level? Just want to get a sense about the margin assumption.

Charlie, as we explained actually quite a while ago, overall, all the data center projects are really operating margin accretive in a very positive way mainly due to the large scale, large revenue size. For the gross margin, again, it depends on case by case based on a different business model.

So overall, if you think about it, two ways, the EPS earnings accretion I think that's certain, sizable; and also operating margin accretion, that's also certain as well. Gross margin will depend because each project, different business model may be varied.

I see. Yes. So can I follow up a little bit on ASIC and move to my second question? So because your industry peers and also kind of end customers also talk about the full system support, right? So I'm not sure if MediaTek also want to develop a similar kind of customer support for the full racks. And if that's the case, whether MediaTek's market share assumption, you just mentioned 10% to 15%, can further go up?

Well, I think for that -- I'm assuming you're talking about the rack, selling a rack. I think that's actually part of the business scope as well. That's something we are discussing with our customers as well.

Really, okay. So yes, because I was a little bit concerned. Some end customers like Anthropic specifically require this kind of support. So I was really concerned if you can only supply the chip. They will constrain your market share upside. But just glad to hear that you're also working with customers for more value add for system level.

So switching gears to my second question. I also asked the same question last quarter to Rick, right? So last quarter I asked about agentic AI for China, and it was specific on [ Doubao ]. But recently, you also picked up a trend that Google is trying to offer AI agents on the phone and to link all the different kind of Android apps.

So I'm not sure if, Rick, you would become more excited about the agentic AI becomes the killer apps or trigger for the smartphone replacement cycle. I want to get your thoughts about these Android agentic AI developments.

Thank you, Charlie. I mean, I strongly believe that agentic AI is and will be creating a lot more new value and new revenue. It is now, I think, in my mind, quite clear. It's a matter of -- and the different ecosystem company, be they in China, the [ Baidu ], Alibaba or, in the U.S., Google, Meta, Open AI, et cetera, will come out with their solutions for their own models and their own business. Their models and their business models.

Our job is to -- I mean, you use a smartphone asking your question. So let me give an example. For smartphone, I mean, we are collaborating closely with Android to ensure the agentic AI applications will be incorporated in our SoC. And even beyond the smartphones. I think there is a spectrum of devices from the wearables all the way to automotive that will add a lot more value with agentic AIs. So I'm very, very positive, very excited about that.

Right. But to realize that, do you need [indiscernible] like your NPU capacity to become bigger? Or do you think your approach is to enable or empower your mainstream phone, it will have these kind of AI compute power? What would be your strategy? And when these concepts can become kind of concrete projects for MediaTek?

We don't view this as a mobile-only play. We do not. We view it as a multiple platform play, from all the way from wearables such as AI glasses to mobile, to IoT, to PC, to automotive, for instance, And of course, each -- automotive, obviously, shall we say, the power budget is much, much higher than the mobile. Mobile is, what, 8 watts? And theirs is about 40, 50, 80 watts.

So in different type of "devices", the computing power implement will be very different. As agentic comes up, CPUs is having a resurgence. So the architecture of the computing system for this agentic AI era within the multiple platforms is a challenge but also a major opportunity. Thank you.

Next one, Arthur Lai, Macquarie.

Rick, David and Jessie, congrats to the strong guidance by ASIC. I have two questions. First one actually is segue to what [ Laura ] just asked. When we think about the smartphone revenue make up 49% of the total. What's the implication to MediaTek's resource allocation, i.e., the 3-nanometer silicon location and also your IP acquisition strategy and your team? This is my first question.

I think, overall, when we look at the AI, we really think about from cloud to edge, they all have opportunity. Because when you think about the application side, agentic AI will happen every, again, from cloud to edge. Currently, due to the overall supply chain situation, the smartphone, obviously this year is having some hiccup. But that doesn't mean that's due to the demand issue. It's mainly due to the supply issue or the memory supply issue to be precise.

So when we think about a 2- to 3-year time frame, we still believe the whole AI demand, if you like, will be across the board, both from cloud to edge, okay? But by saying that, when you look at the addressable market and also from the growth rate perspective, the data center obviously will play a very important role, but it doesn't mean that actually the edge device, including but not limited to smartphone. Don't forget that when we talk about the edge device, which also including the computing device, the automotive, even for the IoT.

So that's how we see from a market perspective. With that, internally, how do we develop the technology, also develop the resources? The computing technology is the foundation. The process technology is also the foundation and also the package as well. So we will invest all those foundation technology and maybe at different times, we have different applications on different segments, okay? So it's not exclusive. It's not exclusive. But obviously, from last year to this year and also for next year, the investment for data center actually almost doubled up.

And the second question actually, also recall last time you guided and told that you will disclose the ASIC revenue when visibility go higher, right? So I think that this time actually, we are prudent to revise the guidance. And so can you share with us what's the timing you decide to spread the ASIC?

I think we will consider probably starting from next year, yes.

Next one, Robert Sanders, Deutsche Bank.

I just had a question around ARM server CPU. I think you've licensed CSS. I was just wondering if there's an opportunity to not just sell ASICs for TPUs but also to do CPUs for the head mode, maybe do LPU equivalents. And I have a follow-up.

I think for the CPU and also for the edge server, I think that's definitely with the possibility. Now, currently, I think our focus will start to ramp our ASIC solution first, at the first stage. But internally, we also have a team to actively monitor and working on that. So far, at least for the next 2 to 3 years, the focus will be the ASIC opportunity. After all, the addressable market is huge, and that's something we would like to put as a top priority.

Got it. And just a quick follow-up on packaged optics. How would you compare your capability today to sort of best-in-class? I mean, would you say you're 6 months behind, 12 months behind or on parity with best-in-class? Just so we understand a bit more about your capabilities in-house versus what's out there in the market.

I would say we are a latecomer. I think there's no denial of that. However, I think, again, the technology now is a lot more mature compared to even a year ago or 18 months ago, and we now have a very strong partner in [ Air Lab ], that I think we can catch up and accelerate our progress in the coming couple of years. And if you look at the real needs, more depending on who you talk to, of course, but I think it's probably more in the 2029 or after 2028, 2029 time frame that the CPO will become a must. So we feel confident and comfortable that we will be there when the need is there. Thank you.

Next one, Haas Liu, Bank of America.

I jumped in the call a little bit late. So apologies if there is any repetitive question. So I think my first question is just regarding your hyperscaler customers on the potential solution. Does that mean you are actually work on the collaboration on the hyperscaler

[Technical Difficulty]

Excuse me, we cannot really hear you. Your background noise actually is very strong. We can't really hear you.

Okay. Do you hear me better now?

Not really okay. Not really. For some reason, we hear a lot of noise. I'm not sure if you're outside or the operator hear the same things. Operator, do you hear the same thing?

Actually, yes. Should be coming from...

Hello. Can you hear me better this way?

Yes, yes, this is much better. Okay.

Okay. Yes. That sounds good. Sorry. I just wanted to ask about your rack. You discussed that you are working with your customer on the potential rack, solutions system-level solutions. Does that suggest that in addition to the core ASIC project you are working with your customers, you are also engaged with your customers on the potential networking switch and also the related peripheral chipsets and the system level solutions? And if so, how should we size up that part of the opportunity? Is it part of the ASIC opportunity you mentioned earlier on this call?

Well, I think it's too early to comment all the details because that's something we're still developing and working with our customers because, after all, we need to start to ship our ASIC solution first, and the customer will turn on the ASIC into the rack. If we get into there, at least based on what the customer is providing, actually for the whole rack solution, technology-wise, we'll still -- even though it becomes the customers' product, so it will be provided by the customer. So we don't need to worry about all those details, but it's more the business model, okay? But again, it's too early to provide all the details. We will provide more detail when it becomes material.

Got it. And just a very quick follow-up on this. If there is a business like this, should we consider more margin dilution from this kind of project, or not really?

Again, it will depend on the final business model, and it's truly coming right -- yes. But for this year and next year, when we talk about the growth profile, we are not factoring in any rack business, just for the record. It's purely ASIC.

Sounds good. Very clear. And second question is just regarding the scarcity of the industry supply across memory and also larger wafers, and I would like to specifically focus on larger wafers. This year, because of the mobile demand is probably weaker than you could reallocate the capacity you got from your foundry partners converted more easily to your other ASIC projects. But in 2027 since the scarcity of the supply across the industry on the 2-nanometer, 3-nanometer wafers is still going to continue, I was just wondering how you are going to allocate the wafer supply you could get from your foundry partners between your ASIC project versus your own core merchant business.

Well, of course, this question is not just for us. It's an industry-wide issue. I think, certainly, we're trying to advance process wafers or TSMC. TSMC's investment is increasing significantly, I forgot the number, $60 billion or -- and they are certainly accelerating their capacity for the advanced technologies, 3-nanometer, 2-nanometer, And we work closely with our partner.

And I cannot speak for TSMC, but I do believe when a customer like MediaTek provides tremendous value for the industry, for our customers, and our customers are really doing very well in their market, there will be supply from that point of view. So I cannot say exactly how much percentage-wise, but I think, I believe as a long-term strategic partner with our supplier, we will get our fair share of the pie.

Got it. And I think just a very quick follow-up before I jump back to the queue is that in the extreme scenario, for example, you only get 80% of the allocation versus what you're demanding for those supply on a wafer level, I was just wondering how you are going to prioritize for your customers' ASIC projects versus your core business because -- how are you going to manage it or balance it?

Well, again, just like our CEO talked about, that's a common question all industry players are facing. So we will probably not be able to disclose internally what's our rules and policy or strategy. But overall, I guess, we need to consider all different factors. But we will not be able to comment this publicly.

Next one, Gokul Hariharan, JPMorgan.

So first of all, on the data center ASIC business, Rick, I think your bigger competitor has signed this long-term 5-year deal with the customer. Now that you have kind of executed your milestone on taping out and potentially starting to ramp up this business with the customer, should we expect that MediaTek also kind of gets elevated to that kind of strategic partner status, potentially look to sign some multiyear deals given that there's a lot of noise in this business, right? So if you can give some longer-term clarity to your investors, this probably helps you and also helps you to plan your business on a multiyear basis. Any thoughts around that, Rick, in terms of how you're thinking about it?

Gokul, Dave here. I think, first of all, we don't really have the detail of our so-called all those partnership announcement agreement. Sometimes actually, GDRs, actually, nobody knows the detail. I think probably the best way to judge is what's our guidance for the revenue guide. And this is what matter. We can sign, you can sign, anyone can sign whatever agreement you like. But by the end of the day, actually whether or not we can deliver the target revenue or higher, like we upsize, I think this is what matter.

Got it. Understood. And secondly, for the second or the next generation project that you're working with this customer, I think Rick mentioned that you are working on 2 packaging solutions, competing packaging solutions. So just wanted to understand, like how do you kind of manage that? Like any risk of execution on this kind of project? I think there is a view of that in the market that, I think, the packaging solution becomes a critical factor in terms of the success of the project. Could you comment about your perspective on that? Or is it like the packaging solution is not really the critical part. It's more on the other areas of execution that basically drives the success of this second project?

Yes, Gokul. I think, first of all, we will not be able to comment on specific project and what technology because there's the confidentiality. But to answer your question, maybe let me try to answer from a different perspective. Like our CEO said, we actually invest 2 different packaging technology. And probably it's a better way to say, actually, we are the frontrunner on both technologies, okay?

So we're trying to do is actually, we have one customer all the industry need for digital technology. We want to make sure we are fully ready and we're the frontrunner for that. That's point number one.

Point number two, going back to TSMC specifically, I think we've been working with TSMC very closely on all the leading-edge packaging. For example, for N2, N1 SoC, I think we probably will be the first one to have a test out in September this year. So again, I understand there's a lot of worry or concerns out there. But the better way to think about that is actually, we actually are the leading player for all different packaging technology. We are not override or underride any one of that. So it's a really balanced strategy and to provide the best solution to the customer.

Got it. Just one follow-up on CPO. I think based on your conversation with the customer, what is the timeline for your ASIC customers to potentially need CPO-related solutions? Is it in this generation, it's the second generation that you're working on itself? Or is it probably a follow-on of that?

I'm sorry, Gokul, you said TPU? Or...

Packaged optics, sorry.

The current project we're working on is actually a TPU. So I'm not quite following.

I was saying CPO, co-packaged optics.

Oh, CPU, okay, Okay. I think currently, I think like our CEO say, at least right now for all those projects we are working on, and some of them, we are building out, all the way to '28 and '29, so far, the CPO is actually not a hard requirement. So that's why we think most likely will be a '28 and '29 beyond. And that's why actually said to have a jump start to prepare that.

I think overall, we don't think we are behind especially if you're judging from when the industry requires CPO because, therefore, you're not going to see anything come out until late '28, probably early '29. So so far, it looks actually we should be okay.

Thank for all your questions. I'll hand it over to Ms. Jessie Wang for closing comments. Ms. Wang, please go ahead.

Ladies and gentlemen, this concludes MediaTek's 2025 First Quarter Conference Call, and an audio replay will be available in 1 hour after the call at the Investor section of MediaTek website. We would like to thank you for your participation, and you may now disconnect.

Thank you, Ms. Wang. And ladies and gentlemen, we thank you for your participation in today's conference. You may disconnect now. Thank you again. Goodbye.

Welcome to the MediaTek 2025 Fourth Quarter Investors Conference Call. Financial results and presentations for today's call are available on the Investors section of the company website at www.mediatek.com.

Now I would like to turn the call over to Ms. Jessie Wang, Deputy Director of Investor Relations. Ms. Wang, please go ahead.

Good afternoon, everyone. Joining us today are Dr. Rick Tsai, MediaTek's CEO; and Mr. David Ku, MediaTek's CFO. Mr. Ku will report our fourth quarter results, and then Dr. Tsai will provide our prepared remarks. After that, we will open for Q&A.

As a reminder, today's presentation will provide forward-looking statements based on our current expectations. The statements are subject to various risks and factors, which may cause actual results to differ materially from the statements. The presentation materials supplement non-TIFRS financial measures. Earnings distribution will be made in accordance with financial statements based on TIFRS. For details, please refer to the safe harbor statement in our presentation slides.

In addition, all contents provided in this teleconference are for your reference only, not intended for investment advice. Neither MediaTek nor any of independent providers is responsible for any actions taken in reliance on contents provided in today's call.

Now I would like to turn the call to our CFO, Mr. David Ku, for the fourth quarter financial results.

Thank you, Jessie. Good afternoon, everyone. Now let's start with the 2025 fourth quarter financial results. The currency used here is NT dollar. The foreign exchange rate applied to the quarter was TWD 31.1 to USD 1 compared with our fourth quarter guidance of TWD 30.6 and the third quarter foreign exchange rate of TWD 30. With that, revenue for the quarter was TWD 150.2 billion, up 5.7% sequentially and up 8.8% year-over-year. Full year 2025 revenue totaled TWD 596 billion, up 12.3% from the previous year.

Gross margin for the quarter was 46.1%, down 0.4 percentage points from the previous quarter and down 2.4 percentage points from the year ago quarter. Gross margin for 2025 was 47.5%, down 2.1 percentage points year-over-year.

Operating expense for the quarter were TWD 47.4 billion compared with TWD 43.9 billion in the previous quarter and TWD 45.6 billion in the year ago quarter. Full year 2025 operating expense was TWD 179.6 billion compared with TWD 161 billion in 2024.

Operating income for the quarter was TWD 21.9 billion, down 1.5% sequentially and up 2% year-over-year.

Non-TIFRS operating income for the quarter was TWD 22.5 billion. Full year 2025 operating income was and TWD 103.5 billion, up 1% year-over-year. Non-TIFRS operating income for the year was TWD 106 billion.

Operating margin for the quarter was 14.5%, down 1.1 percentage points in the previous quarter and down 1 percentage point year-over-year. Non-TIFRS operating margin for the quarter was 15%.

Operating margin for 2025 was 17.4%, down 1.9 percentage points year-over-year. Non-TIFRS operating margin for the year was 17.8%.

Net income for the quarter was TWD 23.1 billion, down 9.3% sequentially and down 3.6% year-over-year. Non-TIFRS net income for the quarter was TWD 23.6 billion.

Full year 2025 net income was TWD 106.1 billion, down 1% year-over-year. Non-TIFRS net income for the year was TWD 108.2 billion. Net profit margin for the quarter was 15.4%, down 2.5 percentage points from the previous quarter and down 1.9 percentage points year-over-year.

Non-TIFRS net profit margin for the quarter was 15.7%. Net profit margin for 2025 was 17.8%, down 2.4 percentage points year-over-year. Non-TIFRS net profit margin for the year was 18.2%.

EPS for the quarter was TWD 14.39, down from TWD 15.84 in the previous quarter and down from TWD 14.95 in the previous quarter. Non-TIFRS EPS for the quarter was TWD 14.71. Full year 2025 EPS was TWD 66.16 compared with TWD 66.92 in 2024.

Non-TIFRS EPS for this year was TWD 67.46. A reconciliation table for our TIFRS and non-TIFRS financial measurement is attached in our press release for your reference, and that concludes my comments. Thank you.

Thank you, David. And now I would like to turn the call to CEO, Dr. Rick Tsai, for our prepared remarks.

Good afternoon, everyone. Thank you for joining us. I'm pleased to report that our fourth quarter revenue reached TWD 150.2 billion, which came in at the high end of our guidance range, thanks to a more favorable foreign exchange rate and better demand for smart edge products. For full year 2025, MediaTek revenue was TWD 596 billion, up 12.3% year-over-year or USD 19.1 billion, up 15.6% year-over-year, registering another year of record revenue. I believe this milestone underscores MediaTek's resilience in navigating a dynamic global economic environment. The strong support and confidence from our customers and partners and the strength of our leading technology and IP road map that has laid the foundation for our growth.

MediaTek's 2025 revenue growth was mainly driven by global share gains of our flagship and premium edge AI Dimensity chips as well as our industry-leading connectivity solutions, including WiFi 7, 5G modem and 10G PON.

We stand firmly as a market share leader across many categories, such as smartphone, tablets, Chromebook, TV and broadband connectivity. Meanwhile, we continued to make very solid progress in technology development and business expansion into new growth areas. For example, we were one of few companies completing 2-nanometer tape-out at TSMC. We worked with partners to enable the world's first Release 19 5G advanced NR-NTN connection over OneWeb LEO satellite.

We successfully expanded our project pipeline for data center ASIC and automotive platforms. We also started to register revenue for high-performance computing products such as GB10 that we co-developed with NVIDIA for its DGX Spark AI supercomputers.

Looking ahead to 2026, as ubiquitous AI continues to push the boundary of more efficient computation and AI applications, we believe MediaTek will continue to thrive on those growth opportunities.

In data center, demand has been increasing rapidly. We are very confident in achieving data center ASIC revenue of more than USD 1 billion in 2026 and multiple billion U.S. dollars in 2027. Furthermore, we are fully committed to executing a follow-on project in our pipeline with revenue starting in 2028.

For MediaTek, to strengthen our position as a trusted partner for data center customers, we are accelerating investments on several key fronts. Firstly, we have consolidated internal R&D resources and increased external hires for data center, AI system architecture, critical IP development and advanced technologies.

With these efforts, we have demonstrated strong capabilities in architecting high-performance heterogeneous compute systems on advanced process nodes as well as implementing advanced packaging technologies that deliver meaningful TPA improvements and enhance power density for custom data center solutions.

In addition, to support our long-term growth and capture emerging market opportunities, we continue to invest in several key technologies, including high-speed 400G SerDes, co-packaged optics solutions, 3.5D packaging, custom HBM and integrated voltage regulators, IVR. We believe these strategic investments will further enhance our technology road map to support customers' evolving requirements.

In automotive, our scalable Dimensity auto cockpit platforms, telematics solutions and power IC continue to gain share globally. In addition, we are making an inroad into ADAS by partnering with DENSO, one of world's leading automotive technology providers to co-develop custom ADAS chips. The solution will combine DENSO's automotive-grade safety expertise and deep vehicle integration with MediaTek's power efficient high-performance SoCs and AI capabilities to offer a scalable platform for next-generation driver assistance. We expect significant revenue growth from our automotive solutions in 2026.

Our computing solutions, which includes tablet, Chromebook and the GB10 project for DGX Spark that we collaborate with NVIDIA grew more than 80% in 2025 to USD 1 billion.

Driven by strong AI adoption and share gains, we have observed very positive feedback for NVIDIA DGX Spark, and we expect its revenue growth to accelerate into 2026.

With AI serving as a catalyst for industry expansion and driving the surge in demand, the global supply chain is facing challenges in fully meeting the increasing needs in 2026, resulting in higher costs across the supply chain.

We have secured the capacity required to support MediaTek's growth in data center ASICs and other areas. We will also adjust our pricing to reflect the rising supply chain cost and allocate our supply across products based on the overall profitability.

We believe 2026 will continue to be a year where MediaTek's business growth is supported by the AI megatrend. Building on this solid foundation, we will continue to invest in critical technologies, including 5G satellite, 6G, advanced process nodes and advanced packaging for both edge AI and cloud AI solutions to capture growth opportunities ahead.

With that, now let me talk about the recent business performance for our 3 revenue groups. Our fourth quarter mobile phone revenue grew 18% quarter-over-quarter and 8% year-over-year in NT dollars, achieving record high quarterly revenue and accounting for 59% of total revenue. This was driven by the successful ramps of our flagship SoC, Dimensity 9500 as well as our premium SoC, Dimensity 8500.

For the year 2025, MediaTek's mobile phone business grew 8% year-over-year in U.S. dollar, Surpassing USD 10 billion in revenue and achieving a record high. Within this, flagship SoC exhibited robust growth and contributed USD 3 million in revenue. These results strongly demonstrate the widespread recognition and market position that MediaTek Dimensity platform has successfully established.

For 2026, under the pressure of increasing memory and BOM cost, we expect overall smartphone end demand to be negatively impacted. We will work closely with our customers to strategically adjust the product portfolio in order to mitigate the impact. For the first quarter of 2026, we expect mobile phone revenue to decline significantly quarter-over-quarter.

Moving on to Smart Edge platforms. In the fourth quarter, this group declined 8% quarter-over-quarter and grew 13% year-over-year in NT dollars, accounting for 37% of total revenue. The sequential decline was due to consumer electronics seasonality, whereas the year-over-year growth was mainly driven by share gains in high-end tablets powered by our Dimensity 8000 and 9000 series.

In 2025, Smart Edge platforms grew 21% year-over-year in U.S. dollars, benefiting from market's continuous adoption of AI and faster connectivity as well as our global share gains. For example, our connectivity revenue grew strongly and exceeded USD 3 billion in 2025, with WiFi 7 revenue tripling and 5G modem doubling. We anticipate these positive trends to extend into 2026.

Even without factoring in the expected data center ASIC revenue, our Smart Edge platforms revenue is still expected to grow healthily this year, demonstrating the structural strength across our business portfolio. In 2026, in addition to the exciting growth opportunities from data center, auto and computing that I mentioned earlier, we also expect higher adoption of AI and faster connectivity to continue across applications. We are once again at the forefront of technology upgrades with our industry-leading WiFi 8 solutions unveiled at CES this year, and we expect early customer adoptions towards the end of the year.

For the first quarter of 2026, we expect our smart edge products to grow sequentially and year-over-year as several products such as connectivity, TV and computing devices recover from the low seasonality.

Moving on to Power IC. This group accounted for 5% of total revenue in the fourth quarter and was down 8% quarter-over-quarter and 11% year-over-year. The sequential decline is due to consumer electronics seasonality, while automotive and industrial grew in the quarter.

In the first quarter of 2026, we expect our IC revenue to be flattish.

Moving to the guidance. In the first quarter of 2026, we expect the recovery of smart edge platforms will partially offset the sequential decline of mobile phones. We further expect our first quarter revenue to be in the range of TWD 141.2 billion to TWD 150.2 billion, flat to decline 6% sequentially and down 2% to 8% year-over-year at a forecasted exchange rate of TWD 31.2 to USD 1.

Gross margin is forecasted at 46%, plus or minus 1.5 percentage points. Quarterly operating expense ratio to be at 31%, plus or minus 2 percentage points.

In closing, we believe 2026 will be a year where MediaTek's business growth is supported by the AI megatrend. Meanwhile, we aim to protect our profitability through disciplined pricing and the strategic capacity allocation. We will also remain fully focused on execution and collaborate closely with our global supply chain partners. With the vast opportunities brought by ubiquitous AI ahead, we firmly believe we are on the right track of our mid- to long-term growth journey. This concludes my prepared remarks. Thank you.

Thank you, Rick. Operator, we are now ready for the Q&A session. May we please have the first question?

[Operator Instructions] The first one to ask question, Sunny Lin from UBS.

So to kick off, first question on cloud ASIC. In recent few months, cloud ASIC demand appear to continue to strengthen. So with that, should we expect upside for your total addressable market estimate of USD 50 billion by 2028 and your market share target of 10% to 15%?

I believe we shared about USD 50 billion last time and 10% to 15% market share. We do not have a really formal estimate -- reestimate of this market size, but I certainly believe it's safe to go up by $20 billion, for instance, about $50 billion to $70 billion. I think that's without really a detailed analysis, but that's my estimate. Thank you.

Right. Well -- sorry, just to clarify, given people really pay attention to your guidance. And so let's say, if addressable market is up to TWD 70 billion in 2028, and you still aim for a 10% to 15% type of market share or you think could be higher given the progress made in recent few months?

I think 2028 is a bit further. We have ideas, but I think let me say, 10% to 15% is rather safe estimate. We certainly strive to go higher.

Got it. And then my second question is to follow up on the business model for a cloud ASIC. And so maybe one is, David, if you could help us understand, going to 2027, when you scale the ASIC shipment, how should we think about the incremental OpEx increase. And given now we are into the ramp, could you also give us an update on gross margin, also OP margin profile for the project?

I'll probably start with the OpEx. I think OpEx ratio, which is scaled into the revenue so far for the full year, also for next year, I think we are looking for the flattish, okay, the ratio should be similar. But given the fact that revenue will grow, so the OpEx dollar will continue to increase, we will basically invest more, like I think, our CEO explained, we're going to invest more for all key technologies.

Profitability-wise, I think that's the same approach. I think overall due to the mix of different business model and also accounting principle, the better way to think about that, especially for 2027, when we start to ramp is you're going to see operating margin accretive. I think that's probably the best way to think about that.

Got it. Sorry, maybe one last follow-up. Since you mentioned you have already secured the next-generation core ASIC project, how should we think about your value add, your content and then also competition evolving into the next generation.

Well, the next-generation products certainly is quite a bit more complex in just about every aspect. And we have demonstrated our capability to elevate our technology and IP capabilities convincingly to our customers. And also on top of that, our execution capability, which we are already demonstrating right now in our first chip situation. So I really believe together with our -- what we've been discussing before, our hybrid model with flexibility with a tremendous service to our customers. I think all these add up to a tremendous value add to our customers for their buildup of their very massive compute power development. Thank you.

Now next one to ask question, Laura Chen, Citi.

My question is also about the ASIC. We're already working on the next-generation project. Can you share with us, is there any change in terms of business model. Like we now mainly doing service and also the turnkey service for our clients. Since Rick, you just mentioned that the design is getting more complex. Will the business model change involve more IPs that MediaTek will be able to provide. And also as we are targeting like a 10%, 15% of the market share. So can we kind of assume that this business will be able to achieve like 20% or higher of our total revenues in next year?

All right. The business model, I think we remain, as I said, our hybrid business model. The value add, of course, will vary from generation to generation. But I'd like to also say, given the high degree of complexity, which is added to the next-generation chip we are providing, I think, actually more value if you look at the total, while our customers is also adding more value in their aspect. So that's why we have been always said this is a win-win model. We get more value, they get more value. And together, they will certainly gain more market share in their end of the business. So I'm very pleased with that development. As to the -- what was question?

20% or higher revenue contribution.

I -- of course, I think you're asking for really advanced or indication, I guess, for 2027 revenue. I think it's possible, certainly. I think 20% is something we believe we have reasonable chance to get to, and we'll do our very best to achieve that.

Yes, sure. Maybe just a quick follow-up. I recall that we are also working with NVIDIA on NVLink Fusion. So are we also seeing that any like progress on the potential engagement with our ASIC business?

The model -- the NVLink Fusion model is being developed. This is -- again, this model by itself is a complex model from a technical point of view and from a business development point of view. I would say we continue to have a very strong and close relationship with NVIDIA. And we both are building the model, the build model and the targeted potential customers, and I cannot really disclose too much more, but it's ongoing.

Yes. Okay. Sure. No problem. My next question is about the gross margin outlook. I recall last time, David, you mentioned that the gross margin for flagship smartphone SoC seems to be lower than corporate average. So I'm just wondering that actually, we see that the component price hike has more impact on the entry level or low-end segmentation. And the portion of our flagship -- on the good side is that probably the revenue will be able to kind of support it to compensate the weakness in the low-end space. But how would the impact on the gross margin side? And how should we think about the gross margin trends in this year?

I think for full year 2026, our goal is actually to aim to sustain our gross margin at around 46% level, which was similar to the guidance we gave starting from last quarter. When you look at our gross margin in fourth quarter last year and also the guidance this year, I think you can consider that will be our goal to maintain at full year. And how do we achieve that? I think like what we explained during the opening remarks, given the overall supply chain tightness right now, I think we've been going through a very rigid about the capacity allocation among the business growth rate and also the profitability. And that's point number one.

Point number two, given the overall -- the cost increase in pressure, we are also in discussion with our customers for certain products, I guess, we have to pass on some of the cost increase. So combined of that, I think will be reasonable, but the goal will be maintain the gross margin around 46% for the full year.

Next one to your question, Gokul Hariharan, JPMorgan.

First off, on the data center ASIC project, could you talk a little bit about your understanding of what you're facing from a competitive landscape perspective because this has been largely the domain of one vendor until very recently for this project.

And also, Rick, could you kind of help us understand a little bit on the technology preparation that you've done on high-speed SerDes, including 400 gig plus, 3.5D and CPO. Could you give us a little bit more details on how we can look at MediaTek's capabilities because that seems to be quite critical for the next generation, the follow-on project that you mentioned?

Yes, Gokul, I guess your first part of the question is if you're asking whether -- how we view other suppliers, our policy is not to comment on competitors' situation. But on the other hand, for our own investment, the multiple technologies and IP areas, I mentioned -- I discussed during the remarks. I think we have staffed through internal resource consolidation. We have many very capable engineers. While we hire high-caliber technologies, including architects, including the power experts, including optics experts. Yes, those are very difficult technologies. We certainly know that because we are doing them. But we also -- by the way, we are also collaborating with some third-party, major third-party players.

So while they remain challenged, very challenging for us, but if you look at the past 2-plus years, I think our track record has shown that we have been able to not only catch up but also to provide level playing field capabilities for our next generation, and we will continue to compete for future and broader range of business opportunities.

And just to follow up. So this current project that is going into production pretty soon this year, do you think that the deployment model is going to be different compared to previously that you were working on some different workloads compared to what your competitor is working on? I'm just asking because so far, if you look at data center AI, most of the demand has been largely at the very high end of the spectrum, whether it is for NVIDIA GPUs or AI ASICs. So I just wanted to understand, is there like a very big division of workload that is happening at the customer, which gives you this opportunity to kind of break in and kind of for the customer also to segment the demand more clearly?

Well, basically, Gokul, this question, of course, needs to be answered by our customers. While -- yes -- but you -- I'm sure you understand the pace of the ASIC chips of such complexity, you have to go back certain amount of years to start when we started the design and then you look at the progress or the status of the AI or generative AI at that time. And then during -- through that design process until the production is, again, we're measuring in -- I wouldn't say -- I cannot say exactly how many years, but you know well.

Things change. But we believe the design of our customers' ASIC chips, they have taken into consideration of their own internal workloads needs plus their external requirement workloads. And how they -- but I cannot say how they balance those workloads. But we have -- as far as I understand, they really are -- they are very high on this chip.

Just one last question on flagship smartphone. Given we are expecting a tough market for smartphones, are we still expecting the flagship segment to grow this year? Or we think we might have to give up some opportunities to protect profitability?

Gokul, I think from a market share perspective, flagship market share perspective, given our new product coming out every year and also all new products are very competitive. I think we still feel comfortable the market share will continue to gain. And likewise, actually, because right now, it's more silicon content in there, we're also seeing generation after generation, the blended ASP generation after generation is higher. But the last variable, if you like, really the overall shipment, I have to say, given the overall -- the memory situation. Right now, it's still evolving or developing. Based on what we see, probably that will become somewhat year-over-year having some pressure on the flagship overall, the market shipment, not the market share. But at least we can somehow balance that with our continued gain on market share and also continued accretion on our ASP.

Next question as Haas Liu from Bank of America Merrill Lynch.

Congrats on the solid results and guidance. So I wanted to follow up on the smartphone business. Would you provide more detail about the smartphone linearity for this year factoring your guidance for a significant decline in first quarter? And from the full year perspective, it is going to see negative impacts from the other component supply shortage. Just wondering if you have any quantitative outlook for the overall market versus your own shipments. And then your overall situation across the mid, high, low-end smartphones as you are reallocating or optimizing your product mix, could you also just share your view on the market share gain or loss as the previous questions?

Okay. Some of your questions involve a detailed resolution. Probably I won't be able to comment, but let me probably start with our full year view for smartphone. Very straightforward, I think given the overall memory situation, and we believe actually the overall smartphone this year year-over-year, maybe we're seeing a decline. I think that's on the top line.

In terms of the detailed resolution for flagship, mainstream, entry level, even including 4G, we probably won't be able to provide the detail. But the key point is when you look at our overall business portfolio, our non-smartphone business, like we explained during the opening remarks, will offset a big part of that and will grow nicely. And plus our data center new revenue coming in, which right now we guide is more than $1 billion. I think overall, we're still looking for the top line year-over-year growth. I think that's probably just our feedback. But for the detailed segmentation resolution, we -- in general, we don't comment on that.

Okay. Okay. Yes, that's clear. And then I think just on ASIC beyond hyperscalers accelerators, should we expect your ASIC business to grow beyond this part of the market, for example, like servers, CPUs and also networking chipsets? And beyond cloud AI computing, could you also discuss more in detail about your potential in automotive ADAS partnership with DENSO because that is going to be a huge chunk of opportunity with higher level of ADAS being adopted across the vehicles.

And then I think as part of my second question is that you also mentioned about you are going to put more effort in 5G satellite. Could you discuss more in detail on what kind of chipsets you are working on with your customers? Or is it going to be part of your effort in low earth orbit?

Okay. About data center, AI, of course, now and our first priority and focus remain at the -- what we're going to ship this year, next year and also the next project following. But in the meantime, we are diligently -- we're working on the market -- the AI market is just changing very, very fast every month, every quarter for sure. So we are rapidly adapting that dynamic. We also -- that's why we are investing in technology that is not just for one customer. We are developing technology, which are generic, more generic also for other applications, while like inferencing is getting a lot more important as the GenAI development has gone. So the answer is we are, but I do not like -- I don't think we can share a lot more than that.

On other new business satellite, you have -- you asked automotive and satellite. Automotive, certainly, we are very excited to work with DENSO. I mean we all know is one of the top automotive Tier 1 player, very, very deep technology capability and very, very strong business relations with multiple major car OEMs. And so the combination of between the 2 companies is just perfect. They provide a deep vehicle integration, understanding all the safety requirements. We provide a low-power SoC, very high performance plus a very high AI capability that the ADAS will need for the next at least 4, 5 years. So we're excited, and I believe they are too. We look forward to major revenues for both.

Satellite, we actually we have we both work on the NTN. From modem point of view, we also actually have WiFi solution for some of their station on earth for the satellite business. So this is one area that we feel very good. We cannot say we believe the fast growing, but we don't -- we cannot say how fast so far. But from whatever we are seeing, our connectivity solutions from both modem and WiFi point of view, we have very strong position already, and we expect to grow on that basis.

Okay. Yes. That's very clear. Just one small follow-up, and I will be back in the queue is that for that NTN projects or the satellite chipsets that you have, is it going to be more on the edge side or more going to be on the infrastructure or base station side? And then I think a couple of years ago, you provided a few -- the CAGR for the next few years on the sales and also margins. I'm not sure if you are able to provide such guidance at this stage for 2026 and also 2027?

I think for the NTN it's more on the edge side, not on the satellite side, I think just to be clear. In terms of the guidance, we probably won't be able to provide guidance so far because there will still be a few years to come. It's growing, but still be a few years to come.

Next one to ask question and Charlie Chan, Morgan Stanley.

Congratulations for a very strong outlook, especially on the ASIC side. So I do have 2 part of questions. One is also same on ASIC. And second question will be on smartphone. For the ASIC, I wanted to ask management whether you see strength over the past months, meaning there's upside to your forecast, especially next year. And if that's the case, you just mentioned you already assured the key components, but are you sure that your TClass HBM can really support that so-called upside order.

And lastly, gross margin trend. I'm wondering for those kind of upside revenue, do you think the gross margin -- incremental gross margin would decline? And lastly, given the revenue size is so big, do you think the customer want to change to the COT model in the current generation or next generation? So that's my questions about the ASIC business.

We are -- as we said in our remarks, we are very confident for our fourth quarter revenue of more than $1 billion. We said for 2027, multiple billion dollars. That's the range certainly.

I would say we believe if and where I believe we will, when we deliver, when we execute, we expect to see some upside. Then the supply -- from a supply point of view, and I also said earlier that we have secured supplies for data center ASIC and other areas. We have a good scale to work with the major suppliers. And we also have certain flexibility among our own different product portfolio. So we also have high confidence that we can manage -- we can provide the supplies for the demand coming next year. We are -- we have been preparing for that for some time.

Lastly, again, that question is not really up to us. But what I -- again, as I said earlier, I really believe -- I think all depends on overall objectives. The overall objectives, I believe our customers or actually, for that matter, all the CSPs is really to provide timely and high-quality low-power computing capability at a very good TCO to grow their business. And the market is changing so fast, the technology pace is so fast.

In the -- if you look at the history of our industry, usually, it would take partners to work together to achieve that required growth target for everyone. And unless you are in a very mature industry, you probably can go just by yourself. But we are not in that phase at all. There is a huge amount of growth ahead of us.

So I fully believe the value add we provide and we are also -- as I said earlier, we are also working hard with our customers to enhance their value add. It's not a zero-sum game, period, and we believe this will continue.

Yes, that makes a lot of sense. So just a small follow-up on the ASIC side before I move to smartphone. So you mentioned about your -- you feel like your 10%, 15% market share is pretty safe to achieve, right? And my understanding is that your customers' ASICs probably account for half or even more of the kind of total ASIC TAM, right? So is that the right way to think about your allocation from the key customers would be like 20% to 30% or even higher -- based on those data points that I just described?

Charlie, I think, again, the earlier question talking about the target market shares or the market share goal is 2028. This is still a few years away. So there are so many moving targets. I think our focus right now is we ask them well about the 2 projects on hand and market share will become hopefully a happy consequence, okay? So that's why actually we don't want to give out the detail. Actually, we will not be able to give out the detailed guidance for the market share for 2028.

Got you. So my second question on smartphone will be relatively brief. I wanted to get management's thoughts about the agentic AI in China smartphone markets. What do you think about the top -- so the kind of feedback and our MediaTek customers want to do similar kind of application? When do you think that will become a killer app?

I think we've been dealing with this question for at least 2 years, if not longer. And I will not give you kind of a boiler plate answer. It's not obvious this "killer app", but I don't think -- I'm not sure that's the right thinking. I think the agentic AI will just blend into your life with the smartphone, and that smartphone can be a very smart smartphone and it may be just a mid-end because you can -- depending on what you need, you can access very sophisticated GenAI models on your phone. But for even more bigger applications, you need to go to the cloud.

So if we provide a solution that will have more than sufficient compute power, but that compute power will be limited by the form factor of the phone itself. That's physics. So the phone will provide ample AI or GenAI or agentic AI capability by itself. But and I believe the demand of the application will increase so much and that phone certainly will play the other role of being just connecting to the cloud very effectively and efficiently. And that will fill both ends. It's -- I think it's part of life. It's a -- and this -- and there may be other form of devices that will enhance people's life using agentic AI. And we certainly -- if there is opportunity or outlook, we are pursuing them also.

Ladies and gentlemen, with the interest of time, we are going to take the last one to ask questions, Arthur Lai from Macquarie.

So my first question is still on the cloud AI. Just want to follow up the Charlie's question. When you comment the scale and also flexibility, can you help investors understand, are you trying to allocate the scarcity resources to incubate your new cloud business? And what's the resources you referred to? Is the wafer location or is the substrate or your experienced engineer? That's my first question.

So David here. I think long story short, we will allocate among all what do you say, internal R&D resource like during the opening remarks, I think Rick is talking about, we've been investing sizable R&D on and plus we've been hiring aggressively external resource. So internal resource, I think we will definitely allocate to this fast-growing area, especially with a huge addressable market. For the overall manufacturing capacity, I think we will put a priority on that as well. So all of that, yes.

And the second question is more on the cycle and also history. This is actually not the first time we engage envision softness of the smartphone. Compared to the last 2 cycles, can management or Rick comment on what the MediaTek's strategy right now compared to the previous second down cycle?

Okay. Certainly, 2026 is a tough year for smartphone. I mean we -- I'm sure you guys have read about that. No need to further elaborate. But I also believe this is a cyclical year. The demand will remain solid next year or maybe in 1.5 years. I cannot say for sure. But I am quite sure that the demand will come back. And when it come back -- when it comes back, it will come back in force. And we need to be prepared for that, and we will be.

MediaTek during the past few years, we have, I think, developed and equipped ourselves to I think to live through the cycle is too weak actually to actually -- I believe we will be stronger after the cycle. The capability we have built the market position, I feel very confident this -- the mobile business -- well, at -- it's now in the sixth year of the 5G generation. Well, certainly, the growth -- overall growth is saturating. We all know that. But MediaTek's position, the capability is much, much stronger compared to our last generation of modem technology. So we feel -- I don't like this down cycle, but that's life, we will be better after that. Thank you.

Okay. Ladies and gentlemen, we thank you for all your questions. I'll hand it over now to Mr. Jessie Wang for closing comments. Ms. Wang, please go ahead.

Ladies and gentlemen, this concludes MediaTek 2025 fourth quarter conference call, and an audio replay will be available in 1 hour after the call at the Investors section of MediaTek's website. We would like to thank you for your participation, and you may now disconnect.

Yes. Once again, we thank you for your participation in today's conference. You may disconnect now. Goodbye.

Welcome to the MediaTek 2025 Third Quarter Investors' Conference Call. Financial results and presentation for today's call are available on the Investors section of the company website at www.mediatek.com.

And now I would like to turn the call over to Ms. Jessie Wang, Deputy Director of Investor Relations. Ms. Wang, please go ahead.

Good afternoon, everyone. Joining us today are Dr. Rick Tsai, MediaTek's CEO; and Mr. David Ku, MediaTek's CFO. Mr. Ku will report our third quarter results, and then Dr. Tsai will provide our prepared remarks. After that, we will open for Q&A.

As a reminder, today's presentation will provide forward-looking statements based on our current expectations. The statements are subject to various risks and factors, which may cause actual results to differ materially from the statements. The presentation material supplements non-TIFRS financial measures. Earnings distribution will be made in accordance with financial statements based on TIFRS. For details, please refer to the safe harbor statement in our presentation slides.

In addition, all contents provided in this teleconference are for your reference only, not intended for investment advice. Neither MediaTek nor any of independent providers is responsible for any actions taken in reliance on contents provided in today's call.

Now I would like to turn the call to our CFO, Mr. David Ku for the third quarter financial results.

Thank you, Jessie. Good afternoon, everyone. Now let's start with the 2025 third quarter financial results. The currency used here is NT dollar. The foreign exchange rate applied to the quarter was TWD 30 to USD 1, representing a 2.8% NT dollar appreciation compared with the foreign exchange rate of TWD 30.8 in the second quarter. Every one dollar of NT dollar appreciation against U.S. dollar will reduce our NT dollar revenue by 1%. With that, revenue for the third quarter was TWD 142.1 billion, down 5.5% sequentially and up 7.8% year-over-year. If we exclude the FX factor in U.S. dollar, third quarter revenue was down 2.6% sequentially, and up 16.2% year-over-year.

Gross margin for the quarter was 46.5%, down 2.6 percentage points from the previous quarter and down 2.3 percentage points from the year ago quarter. Please be reminded that gross margin in the second quarter benefited from a onetime item which increased gross margin -- second quarter gross margin by roughly 1.9 percentage points.

Operating expense for the quarter were TWD 43.9 billion, compared with TWD 44.5 billion in the previous quarter and TWD 40.5 billion in the year ago quarter. Operating income for the quarter was TWD 22.2 billion, down 24.5% sequentially and down 7% year-over-year. Non-TIFRS operating income for the quarter was TWD 22.8 billion. Operating margin for the quarter was 15.6%, down 3.9 percentage points in the previous quarter and down 2.5 percentage points year-over-year. Non-TIFRS operating margin for the quarter was 16.1%. Net income for the quarter was TWD 25.5 billion, down 9.3% sequentially and down 0.5% year-over-year. Non-TIFRS net income for the quarter was TWD 26 billion.

Net profit margin for the quarter was 17.9%, down 0.8 percentage points from the previous quarter and down 1.5 percentage points year-over-year. Non-TIFRS net profit margin for the quarter was 18.3%. EPS for the quarter was TWD 15.84, down from TWD 17.5 in the previous quarter and down from TWD 15.94 in the year ago quarter. Non-TIFRS EPS for the quarter was TWD 16.18. A reconciliation table for our TIFRS and non-TIFRS financial measurement is attached in our press release for your information. And that concludes my comments. Thank you.

Thank you, David. And now I would like to turn the call to CEO, Dr. Rick Tsai, for prepared remarks.

Thank you, Jessie. Good afternoon, everyone. MediaTek's third quarter revenue in U.S. dollars came in at the high end of our guidance range, thanks to better-than-expected Dimensity 9500 demand. Quarterly revenue in NTD exceeded our guidance range, mainly due to a more favorable foreign exchange rate of USD 1 to TWD 30 compared to our assumption of USD 1 to TWD 29. During the quarter, we observed exciting developments in ubiquitous AI. Ongoing innovation in Agentic AI and AI application across industries are fueling demand for AI computation. As a result, several hyperscale companies have announced plans to significantly increase capital expenditure to build substantial AI computing power, which will be deployed through multiple years. We believe this drives a virtuous cycle for long-term AI growth, and we are still at the very early stage of the AI megatrend.

Underpinned by our leading key technologies and strong execution capabilities, MediaTek stands firmly to capture growing opportunities brought by ubiquitous AI. Here are some recent progresses. In the cloud, improving data center total cost of ownership, TCO has been crucial for hyperscale companies. So customized solutions that are optimized for CSPs specific workloads are well suited to add value to data centers.

Our first AI-accelerated ASIC project is well executed. We are on track for our USD 1 billion cloud ASIC revenue in 2026. And we expect multiple billions of revenue in 2027. The achievement reflects our ability in integrating our world-class interconnect IP to custom designs and managing the advanced nodes and advanced packaging supply chain.

Our solid execution, optimizing technology and scale have gained increasing recognition in the market. Follow-on project that is more complex than previous designs is already underway with revenue expected in 2028 and beyond. Meanwhile, we continue to aggressively engage with the second hyperscale company for new data center ASIC projects with high confidence. We believe in our value proposition for cloud AI customers and expect rapid growth in the future.

At the edge, we build powerful AI chips to enable more on device AI innovations across smartphones, tablets, PC, automotive and more. In the third quarter, we launched our latest flagship mobile SoC Dimensity 9500. It features the industry-leading dual-core NPU architecture with 1 performance NPU to handle more complicated AI instructions and 1 efficient NPU to run light AI models for always on AI experiences. Customers have utilized this powerful architecture to enhance user experience in applications such as photo shooting and AI assistant.

On the other hand, GB10, the edge AI chip that we codesigned with NVIDIA has recently begun mass production. GB10 powers the world's smallest AI supercomputer DGX Spark, which can run inference on AI models with up to 200 billion parameters and fine-tune models of up to 70 billion parameters locally. Looking ahead, we believe that advanced process technologies and packaging solutions will unleash more AI innovations and usage cases across multiple industries. And MediaTek is one of few companies capable of making continuous investments. We completed our first 2-nanometer tape-out at TSMC in the third quarter and will be a frontrunner in launching 2-nanometer chips starting in 2026 for customers across industries.

With that, now let me talk about the recent business performance for our 3 revenue groups. Mobile phone accounted for 53% of total revenue in the third quarter and grew 4% year-over-year and declined 4% quarter-over-quarter. In U.S. dollars, this revenue group grew 13% year-over-year and declined 1% sequentially. Demand for our Dimensity 9500 has been stronger than expected. The first wave of customers has launched flagship smartphones, such as the Vivo X300 and OPPO X9 series in China and plans to expand into more regions such as India, Southeast Asia and Europe by the end of the year. For the fourth quarter, thanks to the strong ramp of Dimensity 9500, we expect mobile revenue to grow strongly quarter-over-quarter. The flagship smartphone revenue for the year has been tracking above our expectations. We are confident of exceeding USD 3 billion of flagship smartphone revenue in 2025, representing more than 40% of year-over-year growth.

Now let me move on to Smart Edge platforms. In the third quarter of 2025, this group grew 14% year-over-year and declined 6% sequentially, accounting for 42% of revenue. In U.S. dollars, Smart Edge platforms grew 23% year-over-year and declined 4% quarter-over-quarter. The year-over-year growth was mainly driven by a better mix in tablets with strong AI adoptions and global share gains across connectivity products. The sequential decline was mainly due to the certain demand being pulled forward to the first half of the year.

For the fourth quarter, we expect Smart Edge platforms revenue to decline seasonally. However, we expect automotive to grow strongly in the quarter. Several of our eCockpit customers in China will launch new car models across high-end and mid-range segments, driving our auto revenue to more than double year-over-year in the fourth quarter. And we expect the strength to extend into 2026. In addition, our premium cockpit solution C-X1 is also expected to begin volume production in late 2026, adding momentum for future growth.

Now moving on to Power IC, which accounted for 5% of total revenue in the third quarter declined 4% year-over-year and declined 10% quarter-over-quarter. In U.S. dollars, Power IC grew 3% year-over-year and declined 7% quarter-over-quarter. For the fourth quarter, Power IC revenue is expected to decline seasonally quarter-over-quarter. Moving to the guidance in the fourth quarter. We expect revenue from our flagship smartphone GB10 project and automotive business to grow sequentially. These revenue strengths are expected to more than offset the seasonally weaker consumer electronics demand. With that, we expect our fourth quarter revenue to be in the range of TWD 142.1 billion to TWD 150.1 billion, flat to up 6% sequentially and up 3% to 9% year-over-year at a forecasted exchange rate of TWD 30.6 to USD 1.

Gross margin is forecasted at 46%, plus or minus 1.5 percentage points. Quarterly operating expense ratio to be at 31%, plus or minus 2 percentage points. With the midpoint of our fourth quarter revenue guidance, 2025 is expected to be a record year of more than USD 19 billion of revenue. For 2026, we continue to see great growth opportunities. Amid the tight capacity environment, we will strategically adjust our pricing and allocate our capacity among different product lines to reflect the increasing manufacturing costs. For the mid- to long term, we are firmly on our growth journey. We're well positioned with our leading key technologies and strong execution capabilities to capitalize on business opportunities brought by the ubiquitous AI. We believe our major growth drivers, including flagship SoCs, automotive, computing and data center will continue to bear fruit in the future.

This concludes my prepared remarks. Thank you.

Thank you, Rick. Operator, we are now ready for Q&A session. May we please have the first question?

[Operator Instructions]

The first one to ask question, Sunny Lin from UBS.

My first question is on cloud ASIC. So obviously, last few months, stronger industry developments across the board. So what's your current expectations on the addressable market compared with your prior $40 billion expectation for accelerators by 2028? And how is your progress on expanding the client and project base? I know you mentioned you are engaging with the second project. How's the overall implication to your market share in the space?

Yes, we did talk about $40 billion TAM for the data center ASIC revenue. With the -- I think about 2 years ago, with the current -- actually starting from last year, the CapEx increase from the CSP companies, we now are looking at, I believe, at least $50 billion TAM for the data center ASIC revenue. We certainly strive to gain a lot of new businesses. And I would say, at least right now, we strive for 10% to 15% plus market share going forward in next 2 years to plus. Thank you.

So 2 years meaning by 2028. And so if we use $50 billion total addressable market value, that would imply like over $5 billion type of opportunities. So does that imply that you are on good track to expand your project base to more than 1 by 2028?

Yes. Yes. Number one, yes, we have projects at hand. We feel quite confident that we -- certainly, we need to execute and we will execute to get those products to our customers and to their data center. And in addition, we believe we will be able to gain some other projects from different CSP company. Thank you.

Sorry, maybe just a follow-up, given Rick what you just shared, seems pretty encouraging. So for the new projects, because you talked about the first project going to mass production in 2026, and that's on track to $1 billion sales. How should we think about the other projects? What's the timing for revenue contribution? Will that be in 2027 or mostly from 2028?

As I said in the remarks, we believe the first project will generate multiple billions in 2027 and our -- another project will start delivering revenue starting 2028 and beyond.

Sorry, my second question, if I may, will be on gross margin. So maybe for David, how should we think about the gross margin from here? Based on the guidance for Q4, it seems like the midpoint is down a bit versus the prior 47% to 48% type of range. So what's the decline? Why is the decline? And for 2026, how should we think about the puts and takes?

I think for the fourth quarter gross margin movement is mainly driven or impact by the product mix or revenue mix in the fourth quarter. For next year, I think we are still strategizing about our overall capacity allocation and pricing strategy. I think given the overall semiconductors industry background, which I think everybody know for the leading-edge node, which is, a, is actually somewhat limited capacity and because the demand is very strong and also b, the cost of those leading nodes are increasing. So with that industry background, we need to allocate our capacity strategically to those higher value-added segment. In the meantime, we're actually working in the direction to basically passing on that increase of cost to our customers as well. So for the gross margin next year, please give us some more time. We will update later. But for fourth quarter, again, that's actually mainly due to the product mix and industry mix in the second quarter.

Next one to ask question, Laura Chen from Citi.

My first question is also on the ASIC business. I'm just wondering how would MediaTek manage your R&D resource to work on the scalar AI growth? In particularly, we have a great potential for the second customer. Will MediaTek to consider any potential joint venture or investment? Also, I'm just wondering, we see that scale up and scale out are very important for AI data center buildup. What MediaTek can provide to gain more market shares going forward?

From R&D point of view, we are definitely, not only we are -- we have been moving our R&D budget, which certainly includes the people and the financial means into data center technologies, IP and execution capabilities. This has been going on for a while, but we continue to increasing the investment. We have strengthened significantly our talent in the U.S. so that we can have also not only technical capability, but also the efficiency of the communication directly in the U.S. We are embarking on numerous IP certainly in the high-speed interconnect area, either across the chip or from chip to the rack, but also including the silicon photonics. But we also certainly are working with on the 2-nanometer process technologies, the 3.5D large, very large reticle size chips for the packaging.

The company is fully committed to building the technology and IP capability so that we can not only build the chips for our customers next year and '27, '28, but also the capabilities that we can provide for '28 and beyond time frame. And that's...

Yes, because it sounds quite encouraging for the next few years' growth. So just wondering, will those like the new technology will be developed by MediaTek internally? Or are you also seeking for other partnerships outside the company?

We do both ways. Certainly, we -- many of the IPs we develop internally, we also certainly in the photonics, for instance, area, we have strong partnership with building partnership with other major IP suppliers, not to mention TSMC on the packaging plus the technology -- process technology. We are also certainly looking at opportunity acquiring talent outside of the company. If we do something, we'll let you know later.

Sure. My second question is about the smartphone demand into next year. As we see that the resource in the tech supply chain has been moved to the AI data center, including foundry, memory, substrates, et cetera. So do you see any impact on your customers' order pool momentum into next year or product portfolio planning? So would that also impact MediaTek indirectly given that inflationary environment? So how should we look at the smartphone shipments outlook? And how would MediaTek to help your customers mitigate the risk at the same time to maintain your gross margin and also shipment outlook?

I mean, we've been maintaining our position about the smartphone shipment -- unit shipment overall about low single-digit growth, and we have not changed that view. The changes, of course, happens mostly in the distribution of those shipments among from the very high -- from the flagship premium to the mainstream entry. And as we have said before, there's no question about the -- a bit of the moving towards both ends of the spectrum. And we have been very successful in building our position in the flagship and premium, like I said. We will continue doing that.

What we have seen so far is our OEM customers, they have been able to also still moving very good quantity in their sell-out using our latest SoC, flagship SoC 9500. And they are also, I think, making adjustment in their pricing. And by that, I mean, moving up in their pricing. So we recognize also some of the challenges in the DRAM supplies and the other pricing environment. What we believe is with our value brought by the 9500 chips and our overall system advantages, we will get our value for our chips. And as David just kind of outlined in his answers a few minutes ago. Thank you.

Next one to ask question, Gokul Hariharan, JPMorgan.

My first question is on the data center ASIC as well. Recently, there have been so many announcements coming through on data center ASIC project wins. But so far, it seems like it's very much a winner take all kind of market given one vendor seems to be getting pretty much all the newly announced projects so far. What are you observing here talking to some of these -- like I think you're already talking to a couple of CSPs, if not more. What is the feedback you're getting from these companies that gives you the confidence to kind of have a more balanced kind of market share profile in this market?

Secondly, also for this first project, could you update a little bit on what is the exact status given we've had a bit of kind of delays, it looks like. What is the exact status right now? Has the chip already been kind of taped out? And when we talk about $1 billion of revenue next year, is that all turnkey-related revenue? Or it also include some of the NRE revenues that you would be booking? That's my first question.

Let me try to answer your first question. I think there's -- I mean, we all have seen the announcement either from multiple CSP company plus OpenAI, et cetera, et cetera. I certainly don't need to repeat. Simply because that scale of the CapEx, I think it is pretty, I think, natural for the CSP companies to look at their risk exposure plus the -- plus the TCO requirements for their investment. And so the ASIC design, which can optimize their own workload, plus the flexibility of the business model that we provide to enable to gain -- to gain our business and also enable their capabilities. Just I think it's also a very good business model and business decision for them. So we feel comfortable. We understand this is a very difficult and challenging task, but we are confident that we can execute well.

For the second question...

I think for the second question, like our CEO said, we feel fairly comfortable. We're passing a certain key milestone basically we been taping our portfolio ready. And the revenue we talk about next year is not including NIE. It's purely the revenue for the shipment.

Understood. That is very clear. Second question is in terms of the business model, I think previously for data center ASIC, you outlined you want to do second as well as GDS II in kind of projects. Is there any thinking about potentially trying to move down the curve and also start thinking about taking some role in some pure back-end design projects to build up scale, given the number of projects available are probably still quite few in numbers. Is there any thinking strategically to kind of go down to those kind of projects as well, even if margin threshold might be a little bit lower, but still gives you a lot of dollar profit from those kind of engagements.

We understand the different business models and I must say for now and the foreseeable future, we will focus on our current business model, which we believe provides a really good balance between a very -- sometimes very high cost or technology provided or design provided business model and very kind of a much lower value-added back-end turnkey service. I think we have the right balance that will benefit our customers and ourselves and MediaTek for quite a while. We do -- well, unless, of course, there is some really special situation, we can consider it. But right now, I must say we are pretty fully occupied to deliver what we are already awarded.

So just a follow-up on that is, do you think like, let's say, a customer goes to GDS II in kind of projects from currently, they are basically full spec into the largest ASIC vendor out there. Do you think the customer stops there? Or do they continue to in-source more and more of the IP and basically convert that into a purely back-end service kind of model like one of the CSPs already doing?

Again, this is a strong function of different companies and different companies' philosophy and their design complexity -- so I cannot give you a definitive answer as to which models each customer would like. I'm sure the customer would like to be able to contribute more value through their own means overall at the end of the day. But we understand that, and we -- but we also understand from a customer point of view, there are certainly some pretty high -- still a strong value-added area that they are -- they probably decide not to do themselves because, again, you have to weigh the investment and the timing. The schedule at the end of the day also it's not only critical, but it's probably the most important factor. I think, again, our model provides the balance of the capability, GCO and the schedule.

Now asking question is Charlie Chan from Morgan Stanley.

So my first question is about your AI smartphone. It was great to hear that the selling was very strong. Actually, I attended your product launch events in Shenzhen and your presentation and marketing was great. So I'm wondering what is the so-called sell-through so far? And you mentioned that you hope to gain some value from your customers, right? But we're also hearing some price competition from your U.S. competitor into next year in the high-end segment. Can you help us to understand, one is that whether AI smartphone really has that demand? And two is that whether they can really offset some sort of pricing pressure from your U.S. competitor in the high end?

Yes. Charlie, I think what we see the AI smartphone, especially for the flagship smartphone demand is very solid, and I would say it's good. And especially with our new product launch, I think we are still seeing we continue to gain market share. Taking the fourth quarter, for example, right now, the fourth quarter guidance we give out is actually 0% to 6% growth. I think this anti-seasonal growth, a big part of the reason is actually we see a fairly strong smartphone shipment starting in fourth quarter this year. Of course, that actually lead to some of the gross margin movement due to the product mix.

I think as far as for the competition you talk about, the competition has been there for actually forever. So we've been dealing with that kind of competition for a long time, especially given the overall supply situation, we actually think there's an opportunity for us to manage the overall capacity allocation and also balancing the pricing going forward. It's challenging. It's not easy, but I think that's the direction we are driving so far.

So just a very quick follow-up on your first question's answer. And also, I do have a second question. So you mentioned that the flagship kind of shipments are going well. So is that kind of dilutive to your corporate margin because your kind of corporate margin is like slightly down to 46%. You mentioned about the product mix. So it's kind of the question number one is a follow-up. And my second question, maybe to Rick is that you mentioned about the second big customers in ASIC. Is this project win confirmed? Or there is still some uncertainty. So the second project is more like your kind of hope or plan, but it's not like a done deal yet.

Maybe answer the first question first. Yes, actually, Charlie, I think like we explained, the gross margin movement in fourth quarter is partially due to the product mix, i.e., the smartphone.

Well, I really cannot elaborate too much. But as I said in my opening remarks, you are unique engaging with high companies. Thank you.

Yes. So Rick, so what would MediaTek's a key differentiation besides the TCO? I think following Gokul's kind of question, right, it seems like there's a dominant player kind of win essentially almost all the kind of high-end projects. So can you elaborate recent what gives you confidence? And what is MediaTek's key differentiation?

I kind of articulate our business model and our capabilities. I probably don't want to spend too much time again. But I really think the best evidence is what we are doing now, what we're executing. We are getting good projects and with really strong revenues to come. And we have done that in, I would say, a fairly short 2 years' time. And I believe our -- from that point of view, we have gained a lot of credibility in the market among different customers.

Next one to ask question, Bruce Lu from Goldman Sachs.

I'm happy to hear that you guys have pretty good progress on the GB10 projects. Can you tell us a bit more about the business model you have with NVIDIA, the user case for GB10 and what kind of impact is going to be when NVIDIA invested in Intel, which theoretically they can do more in x86 architectures in this specific domain?

Yes. Bruce, I think GB10, basically in the business model, we provide the CPU and also the overall system integration and getting the GPU chiplet. But the final product go-to-market will be managed by NVIDIA. So there's a long way to explain. This is actually the NVIDIA's product, and we are the design service ASIC partner. And in terms of segmentation, I believe GB10 right now is being considered about the supercomputer. You can run on a single GB10, you can run more than 200 billion parameters, you can even stag it together to run 400 billion parameter. Probably that's the best workstation like personal computer, personal supercomputer you can get. And also the market -- the product is actually in the market up and running already.

So I think both from the product segmentation perspective and also from time-to-market perspective, I think we actually have the advantage. And I think the second-generation product right now, the partnership is also ongoing. So we do believe both from the technology perspective and also the speed, i.e., the time-to-market perspective, we are actually very competitive as a partnership with NVIDIA.

Okay. Well, you did also ask about the NVIDIA Intel announcement. Well, number one, of course, we cannot really answer such questions. We're not the party. But on the other hand, we read -- from what we understand, our -- you just heard from David, the chip that we collaborate with NVIDIA, GP10 or there may be a different name for other different applications such as in the PC field. The segment is quite high. I mean this is a really powerful chip. So we don't really believe the chips that we build together, the segment of which that we can pursue will be impacted by the other announcement.

A quick follow-up. What is the addressable market for this product in '26 and '27?

It's really what we call that professional...

Professional...yes.

Professional to high end, very, very high end. Yes.

I see. Okay. My second question is going back to the profitability, right? If you exclude the one-off impact from third quarter, the company's gross margin has sequentially declined for like 5 quarters already. And we understand all the puts and takes for the product. But if you look at your growth driver, which is flagship models, autos, even for the ASIC, which might have $1 billion or multibillion dollar revenue in '27, which are not margin accretive. So when can we see the stabilization of gross margin given that the growth drivers are margin dilutive? Or can we -- we need to shift focus to focus on like operating margin in the future?

Bruce, I think we kind of explained that before, especially for the ASIC business model. The key for the ASIC business model, which is very operating margin accretive starting from day 1. Because this is -- the business model natural and also sometimes actually a different arrangement. So judging from the gross margin ratio probably is not the best way. It's really just the operating margin ratio or in that way ROI is actually enhancement once we start to ramp this actually especially with the meaningful revenue, okay?

On the other product line, on the other hand, there's actually -- we're just dealing with about the increasing cost for the supply chain and also the limited capacity. So we just need to strategically allocate our capacity and also trying our best to pass on that increased cost to our customers. So we are working on this direction.

So for the modeling purpose wise, given that you have like multiple billion revenue coming from ASIC in '27, we should model like a sequential margin decline for the gross margin, but should be accretive for the operating margin. Is that right understanding?

Yes. So I think -- again, this -- I think the big revenue for '26, we're talking about $1 billion from ASIC. For '27, we just give the guidance is going to be multibillion dollars. But for '27 because it still happens on time. So we will provide some guidance or heads-up probably somewhere sometime in 2026. But in general, that's the right direction.

Next one to ask question [indiscernible] from Bank of America.

My first question is regarding your profitability. For next year, with growing competition pressure in the mid- to low-end smartphones, I'm just wondering if you are able to fully pass on higher manufacturing costs to your customers? Or if you could just share your smartphone strategy? Would you prioritize sales growth or margin stability when you face more pressure from both your suppliers and competitors?

Well, I think our overall strategy is actually trying to find the best value for our limited capacity. So again, with your question, basically, we're trying to prioritize the profitability over the market share. I think that's our goal.

Okay. So then we could actually assume that your smartphone business on the profitability on the margins next year would actually be more stabilized into 2026 versus what we have been seeing in the past few quarters for 2025, right?

I think for next year's gross margin and also the pricing right now, we are still in the stage of working with the market and the customer directly. So we will provide the guidance next quarter.

Okay. And then just a quick follow-up on the margin questions. With much higher design and also manufacturing costs on the advanced nodes, so beyond 2-nanometer, will you still consider to migrate as aggressively as before? Or if you could share your node migration strategy going forward?

That's being looked at very, very closely and carefully. It is no secret that the Moore's Law at least from a 2-dimensional point of view is slowing down. Each generation still provides values, but it's not as strong as before. So yes, I think 2-nanometer is certainly, we believe, the right thing to do. The next node, we're working closely with also our partner, foundry partner to see what is the best way for both parties. It's not fully decided yet.

But should we consider if your design service products or your own products will be the one migrating the earliest to the most advanced nodes? Would you give some indicator on that?

Up to 2-nanometer, yes. Up to 2-nanometer, as I said, it's being looked at very closely and carefully. We cannot comment definitively yet.

Okay. And then my second question is just regarding your long-term strategy. The traditional market demand growth could still be pretty mild in 2026 and ASIC could be the next key growth driver for the company. And I was just wondering if you could share your target for the next 2 to 3 years. How much of your revenue is going to come from the edge applications versus the cloud business?

Yes. I think for next year, 2026, I think we already gave our guidance for the new revenue, what should I say, for the data center revenue because next year will be the first year we start to ramp. So it is -- will be $1 billion. So which means next year, a big part of the revenue on the growth side still be other revenue as well. But overall, if you only see about the gross incremental revenue, if I, a big part of that will be data center and also the rest of that will be the traditional or the existing business. I think that's the guidance. But once we get into 2027, most likely, I think the data center will start to pick up. It will become one of the major incremental revenue contributor.

Yes. And then just a quick follow-up to my second question is that if you could share your R&D intensity, which means your R&D expense as a percentage of your sales to each of the business. For cloud versus edge, I would assume probably in your cloud business, the R&D intensity will be higher, right? So in the mid- to longer term, 2 to 3 years view, do you think your R&D expense as a percentage of your sales will actually be higher than what we are seeing now?

No, actually, I think R&D spending in terms of the revenue ratio, I think, will stay probably even coming down a little bit, mainly due to the revenue growth, sales growth.

Next one to ask question, Brett Simpson from Arete.

I have a 2-part question on MediaTek's kind of structural profitability. First of all, in the near term, -- if I just look at the guidance for Q4, you have flagship smartphones rising in the mix. You have a solid premium automotive ramp-up. I mean it feels like the mix within the business is getting richer in Q4, and you have an FX tailwind sequentially, yet you're guiding down gross margin. So I just want to understand the dynamics as to why this is happening. Are you driving pricing aggressively in flagship? Is that why we're seeing profitability sort of decline sequentially? Any help there would be useful.

And then the second part, kind of looking more longer term at MediaTek structural profitability. If I just sort of step back, you have a relatively low gross margin and a high OpEx to sales versus your large fabless peers. If I look ahead, we have a structural wafer price hike coming from TSMC. You're moving flagship smartphones to 2-nanometer. You have rising Arm royalty rates. And it feels like we're heading into a period of high inflation in a price-sensitive smartphone market. So are you able to pass on these cost hikes? And how should we think about structural profitability given all these dynamics that we see long term?

Brett, I think for fourth quarter, like you say, both for smartphone and automotive, we see a pretty strong growth. But just in terms of magnitude of contribution, the big part of it still belong to the smartphone. So the gross margin movement is, like I explained earlier, for fourth quarter specifically, it's mainly due to the mix, i.e., actually the smartphones bring down the gross margin a little bit. I think that's actually for the fourth quarter. For next year and ongoing, like you say, we see the inflation on the semiconductor supply chain. And on the side of the inflation, we also see some supply constraint actually very tight supply along the whole supply chain. So we are in the process to decide about the capacity allocation and also passing on the inflation to our customers. We're in the process of doing this right now.

And sorry, David, just to clarify, when you say smartphones is lower gross margin than corporate average, is flagship lower than corporate average gross margins?

Yes, for the fourth quarter, yes.

Okay. And I had a second question for Rick. I think there's some confusion in the market about MediaTek's with the D9500 specifically, whether or not MediaTek is a CSS licensee with Arm. And I ask because Arm is certainly signaling that you are a CSS licensee, and that would imply a 10% royalty rate is being paid or something in that general vicinity, which is structured a lot higher than we've seen historically. So I'm just wondering if you can clarify this position because we're a little bit confused here.

The confusion is not from us, Brett. It's from our supplier. As far as we understand, we are not a CSS licensee. We are not. I just want to be very clear.

Ladies and gentlemen, with the interest of the time, we are going to take the last one to ask question and the last one will be Arthur Lai, Macquarie.

Can you hear me?

Yes.

First, congrats on the big win in the ASIC tape-out. I think that the investor actually underappreciate your company has a very unique position. Number one, the strong partnership with NVIDIA and also second is the SPU and scalability. So this question is to Rick. Can you share with us how you get the leverage between these 2 camp among NVIDIA and also the TPU. I hope you can give us some your strategic thinking in the next 2- to 3-year time horizon.

Of course, I'm sure you understand the nature of the relationship MediaTek has with NVIDIA, which focus on actually both on the edge side for one is the high-performance CPU, GPU chiplet design and application in the computing and automotive side. But they are most -- they are in the edge side. There is certainly now effort between the 2 companies to see whether NVIDIA's very strong networking IPs can be utilized in the ASIC business model. That work is ongoing, and we certainly cannot comment further. But again, from MediaTek point of view, the ASIC data center ASIC business and the business model are ongoing, and we are reaping fruit in the near future, as you said in your question. And that strategy was formulated in the last 2, 3 years, and that strategy continued to progress. And we feel actually really much more confident now compared to before.

Yes. Ladies and gentlemen, we thank you for all your questions. And now I will hand it over to Ms. Jessie Wang for closing comments. Ms. Wang, please go ahead.

Ladies and gentlemen, this concludes MediaTek's 2025 Third Quarter Conference Call, and an audio replay will be available in 1 hour after the call at the Investors section of MediaTek's website. We would like to thank you for your participation, and you may now disconnect.

Yes. Thanks again for your participation in today's conference. You may now disconnect. Thank you again, and goodbye.