Welcome everyone to the MediaTek 2024 Q2 Investors Conference Call. Financial results and presentations for today's call are available on the investor section of the company website at www.mediatek.com. Now, I would like to turn the call over to Miss Jessie Wang, Deputy Director of Investor Relations. Miss Wang, please go ahead.
Good afternoon, everyone. Joining us today are Dr. Rick Tsai, MediaTek CEO, and Mr. David Ku, MediaTek CFO. Mr. Ku will report our Q2 results, and then Dr. Tsai will provide opening 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 materially different from the statements. The presentation materials supplement 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 content provided in this teleconference is for your reference only, not intended for investment advice. Neither MediaTek nor any of its independent providers is responsible for any actions taken in reliance on content provided in today's call.
Now, I'd like to turn the call to our CFO, Mr. David Ku, for the Q2 financial results.
Thank you, Jesse. Good afternoon, everyone. Now let's start with the 2024 Q2 financial results. The currency used here is all in TWD. Revenue for the quarter, TWD 27.3 billion, down 4.6 sequentially, and up 29.7% year-over-year. Gross margin for the quarter was 48.8%, down 3.6 percentage points from the previous quarter and up 1.3 percentage point from the year ago quarter. The quarter -over- quarter change in gross margin was mainly attributed to a one-time item in the Q1 . Excluding the one-time item, our Q2 gross margin improved from 47.9% in the Q1 to 48.8% in the Q2.
Operating expense for the quarter were TWD 37.2 billion, compared with TWD 37.7 billion in the previous quarter and TWD 31.9 billion in the year-ago quarter. Operating income for the quarter was TWD 25 billion, down 22.4% sequentially and up 69.2% year-over-year. Non-IFRS operating income for the quarter was TWD 25.3 billion. Operating margin for the quarter was 19.6%, down 4.5 percentage point in the previous quarter and up 4.6 percentage points year-over-year. Non-IFRS operating margin for the quarter was 19.9%. Net income for the quarter was TWD 26 billion, down 18% sequentially and up 52% year-over-year. Non-IFRS net income for the quarter was TWD 26.2 billion.
Net profit margin for the quarter was 20.4%, decreased 3.3 percentage points from the previous quarter and increased 4.1 percentage points year-over-year. Non-TIFRS net profit margin for the quarter was 20.6%. EPS for the quarter was TWD 16.19, down from TWD 19.85 in the previous quarter and up from TWD 10.07 in the year ago quarter. Non-TIFRS EPS for the quarter was TWD 16.36. A reconciliation table for our TIFRS and non-TIFRS financial measurement is attached in our press release for your reference. 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. MediaTek has delivered solid Q2 results, with revenue of NT$127.3 billion, coming in at around the midpoint of the guidance range, and gross margin of 48.8%, exceeding the high end of the guidance range due to a better product mix. For the H2 of the year, our revenue registered NT$260.7 billion, which grew 35% year-over-year, driven by increasing adoptions of 5G, Wi-Fi 7, and AI across products. MediaTek's commitment to technology leadership and strong execution has strengthened our global position and facilitated our strategic expansion to capture opportunities for the mid to long-term growth. For example, numerous opportunities are fueled by the AI mega trend, which is unfolding rapidly, both at the edge and in the cloud.
For data center companies, there are strong needs to customize chips in order to minimize total cost of ownership... We are able to address their needs through our flexible ASIC business model by providing our leading 112G and 224G SerDes IP. On top of our strong capabilities in complex IC integration, advanced process nodes, and advanced packaging. Equipped with those key capabilities, we are actively in discussion for more AI accelerators and Arm-based CPU opportunities. At the edge, we also recognize the industry's needs to lower inference cost and energy consumed to connect between cloud and edge, as well as users' needs for better privacy and improved latency. We believe these trends will support robust demand for a broad range of AI-capable edge computing devices, such as smartphone, laptop, tablet, automotive, robot, et cetera.
MediaTek has the world's broadest footprint in edge devices, and coupled with our leading-edge computing SoCs, we are the perfect partner for edge computing players. Today, our advanced 3 nanometer SoCs can support applications across smartphone, tablet, automotive, and more. Our SoCs all integrate MediaTek's powerful and power efficient NPUs, the processor dedicated for AI task, task at the edge. Furthermore, we closely work with all the popular large language model providers with easy-to-use toolkit, such as our NeuralPilot, to enable AI applications, including AI agents and a variety of productivity enhancement tools. The demand for more powerful edge computing SoCs and the shortened replacement cycle certainly strengthen our business outlook. We're excited about the growing business opportunities for the mid to long term, and we'll continue to invest strategically and work with global partners to bring advanced and leading products to customers.
With that, now let me talk about the recent business performance of our three revenue groups. Mobile phone accounted for 54% of total revenue in the Q2 , growing 52% year-over-year and declining 15% sequentially. In the Q3 , we see healthy demand for our 4G SoCs from emerging markets, offset by a decline in our 5G SoCs. We expect our mobile business to be sequentially flattish in the Q3 . We continuously expect full-year smartphone market to grow low single-digit % in shipments, while global 5G penetration rate is on track to reach low 60%. MediaTek continues to lead smartphone market with leading products empowering smartphones across all segments. In the flagship segment, our Dimensity 9300 enabled several 5G AI smartphones this year.
The highly anticipated Dimensity 9400, which will be our next flagship SoC, is expected to hit the market in October. Our Dimensity 9400 utilizes the industry's most advanced 3nm process and Arm's most powerful v9 core, with MediaTek's proprietary 4 big core architecture, to deliver substantial enhancement in both performance and power efficiency. Moreover, Dimensity 9400 is further optimized for Gen AI functions. We see ecosystem players aggressively developing Gen AI features, which could be catalyst to shorten smartphone replacement cycle in the future. We have received very positive customer feedback on Dimensity 9400. Compared with Dimensity 9300, there are more model adoptions in the first wave of Dimensity 9400 customers. We are confident to deliver flagship revenue growth of more than 50% in 2024.
Now, let me move on to smart edge platforms. This group grew 11% year-over-year and grew 12% sequentially in the Q2 , accounted for 40% of total revenue. During the Q2, our strategic wireless and wired connectivity solutions, including Wi-Fi 7 and 10G PON, continued to secure more project awards among operators in Europe and the US. Besides, product mix of our computing devices, such as tablet, was further enhanced due to increasing adoptions of Gen AI SoCs. For the Q3 , we expect smart edge platforms revenue to remain flattish. Connectivity is expected to be flattish quarter-over-quarter. Computing and ASIC-related applications are expected to grow moderately, offset by a decline in TV following customers pulling for global sports events in the H1 of the year.
With a broad base of product offerings in smart edge platforms, we are confident that we will continue to elevate our overall value proposition to customers via continuous technology advancement and global market expansions. Now moving on to power IC. Power IC accounted for 6% of total revenue in the Q2 , growing 9% year-over-year and growing 12% quarter-over-quarter. The growth was mainly driven by seasonal demand for most of the major platforms in the Q2 . We expect a moderate growth for the business in the Q3 . For the Q3 , we expect our revenue to be flattish across all three revenue groups to reflect a stable market demand.
With that, we expect our Q3 revenue to be in the range of TWD 123.5 billion to TWD 132.4 billion, down 3% to up 4% sequentially, and up 12%-20% year-over-year, at a forecasted exchange rate of 32.3 TWD to $1. Gross margin is forecasted at 47%, ±1.5 percentage points. Quarterly operating expense ratio to be at 30%, ±2 percentage points. For 2024 full year, our financial targets remain unchanged. Revenue in US dollar is expected to grow in mid-teens%, and gross margin, excluding one-time items, to be between 46% and 48%. With that, we iterate our confidence in our mid- to long-term strategy.
We are well positioned in the AI megatrend for both cloud and edge. Our new projects are tracking well to contribute revenue from the H2 of 2025, with 2024 being the beginning of our next growth phase. Thank you.
Thank you, Rick. Great, we are now ready for Q&A. We're pleased to have the first question?
Yes, ladies and gentlemen, we are now in Q&A session. If you would like to ask questions, please press star one on your telephone keypad, and please ask your questions after your name is announced. Please limit your questions to two at a time to allow more participants to join the discussion. After two questions, we will move to the next caller. Should you have more questions, please press star one again to come back to the queue. To cancel your question, simply press star two. Thank you. As a reminder, it is greatly appreciated that you turn off the speakerphone mode of your device to prevent possible echo effect. We thank you for your cooperation. Now, please press star one if you would like to ask questions. Thank you. The first to ask questions, Laura Chen from Citi. Go ahead, please.
Hello. Hi, good afternoon. Can you hear me clearly?
Yes, please.
Yes, thank you. Thank you for taking my question. My first question is about the demand outlook. We know that, actually, MediaTek delivered quite a decent result for the H1 . And if you look at the H1 and also considering that the Q2, even like a moderate revenue growth for the full year outlook, Rick, you just mentioned you still maintain the full year guidance maintained. Do you see that any uncertainties ahead into 2024, given the demand outlook? And also, since we launched the most high-end Dimensity 9400 series, the performance seems to be quite good, and also we know that that definitely will get a lot of attention.
So just wondering that, for the H2 , in terms of, the product mix, how would that impact our revenue outlook? And also, since we already deliver, even without the one-time effect for Q2, the gross margin is also quite decent. So just wondering, in terms of the revenue outlook and also the margin trends, how the product mix will further impact, our financial results? That's my first question. Thank you.
... Okay, thank you. Yes, I think basically, the H1 of revenue is out, and we just gave the Q3 guidance. And then for Q4 , this is the thing I want to make clear, is basically, we believe now into the H2 or maybe Q2 of this year, we're now moving into a more normal pattern in terms of the seasonality. Pandemic during, I mean, we're going back quite a bit, but basically pandemic really has caused major stir in the supply, demand, and the inventory situation, either inventory buildup, inventory depletion, then the restocking. Even after about six quarters of the end of the pandemic, we went through those fairly abnormal patterns.
We believe now, now that the, now we're in the H2 of 2024, the pattern's really, the seasonal pattern is getting more like the, well, shall we say, before the pandemic years. So Q4 , basically, as you know, represents mainly, depending on most of the products in the consumer area, a low quarter after usually a more robust Q3 . Most people have prepared for the, you know, holiday season in the US or in China, like a Double Eleven Festival, have prepared their inventory back in Q2 and Q3 . So the Q4 basically follow that pattern in general, other than the 4G...
I'm sorry, the 9400 flagship announcement, which of course will bring in uptrend from that point of view. But overall, if you look at our total revenue and the total portfolio, I think Q4 will in our current forecast represent lower quarter, Q4 , and so we are now maintaining our outlook for the whole year as a result. As to the gross margin, we remain our forecast for quite a while. We expect our way to meet between 46% and 48% going forward also. Thank you.
Okay, thank you very much. That's very clear. And also, can you give us more update about the current ASIC business progress? I recall that last time, if you mentioned, we probably will see the contribution starting from the H2 of 2025, is still the target? And other than the AI accelerators, do we also have other opportunities on, like a CPU or any other different type of ASIC business? Thank you.
Okay. ASIC business execution is going well, quite well. And our forecast remain on track for the H2 of 2025 revenue. AI, right, of course, that will be at the beginning of the that ASIC business. And as I think I stated back in early June, during the Computex talk, we are definitely moving into AI accelerators and with the CPU when needed. I mean, it's really if you look at basically, this business is very customer specific. Different customer have different kind of architecture, shall we say. So we, MediaTek, the important thing for us is to we have the capabilities both in all the IPs from computing CPU, GPU, NPU, and the interconnect like SerDes, processors and advanced packaging.
We, and most importantly, we have built strong relationship and partnership with the, ecosystem players. So we are, in addition to the current ASIC business, which is, in execution, we are, working toward, more opportunities, which will present itself, I hope in the next few years. I remain very optimistic, and we are, we are investing also, heavily and, strategically in this area of the business opportunity. Thank you.
Okay, thank you. Now back to the queue.
... Next one to ask questions, Brett Simpson, Arete. Go ahead, please.
Yeah, thanks very much. Rick, I wanted to ask, on the smartphone business, it sounds like 5G sales are going to bottom out for MediaTek in Q3, which is unusual, because normally seasonality is quite strong for MediaTek in Q3. So can you maybe just sort of frame this in terms of how much is Q3 being dampened by industry clearing inventory ahead of an AI product cycle? And how should we think about Q4? Because you have obviously a big product cycle ramp with the 9400 coming. How should we anticipate Q4? Is it going to be a period where you see a big channel fill? Any color would be very helpful. Thank you.
Okay. Brett, as I said, just now, well, Q1 , second rev, 2024, Q1 represents the final quarter for the so-called restocking. That explained quite a bit of the fairly high revenue from quarter-over-quarter and year-over-year point of view. Yeah, after that, I think the customers are consuming some of those inventory. So I don't-- Basically, the Q1 effect you see, as we stated in our opening remarks, is actually the smartphone in 2024 is doing, in a relative sense, nicely. We expect a low single digit growth, and it is happening. And in China, we are seeing also maybe about 3% growth year-over-year in 2024.
We're seeing just, I think considering, the smartphone overall market, as we really believe this is a reasonably good year for smartphone. The changes over the quarters, mostly from the inventory build inventory digestion and flagship 5G flagship, as we see, launch cycle. If you put them all together, it's really not something we worry about, to put it frankly, wrongly. We are seeing revenue being flattish in the Q3 , because, well, because the customers still are consuming some of their inventory, but their inventory level at their hand in the channel, their channel, I think remained fairly healthy. So that's what we're seeing. We feel reasonably good.
The thing for us is to really execute on our Dimensity 9400, Dimensity 9400 launch, and build a momentum to have a good growth. 2024, as I said also earlier, in the flagship SoC revenue, we, we are very confident to exceed our 50% year-over-year growth in flagship revenue. We of course forecast, forecasting higher. This momentum, we hope and expect to continue into 2025. I hope I'm answering your question, Brett. Thank you.
Yeah, that's great perspective. Thanks, Rick. And maybe just a follow on. I wanted to ask about the 9400 specifically. Can you maybe just talk about the content growth of the 9400 versus 9300? I mean, typically we would expect a 10% or 15% increase in price, you know, just given the feature sets of the 9400, there are gonna, particularly for AI, are quite a big jump on 9300. So what sort of content growth can we expect versus 9300? And I wanted to ask about the non-China customer opportunities for next year. You know, we know you've done well in China, ramping flagship. Is 2025 a year where you can start to penetrate non-China customers, particularly in Korea? Thank you.
Okay. The content growth, Brett, I assume you first ask about the capability that we can deliver in 9400 compared to the previous generation. Basically, all the computing capabilities from CPU to GPU to, of course, especially NPUs, will increase quite a lot. The CPU core, we are using Arm's v9 core, which represents, I think, at least 15% and better performance jump. NPU, I think I said that, I hope I remember correctly, back in June, the NPU by itself in 9300 is delivering 48 TOPS....
For 9400, you can expect, I don't want to steal the show in October totally, but I think you can expect, fresh, definitely, 40% something, increase in the NPU. Which is, nowadays, if you look at the trend in this AI world, NPU represents a huge element of the overall content. And MediaTek, I'm very proud and happy to say that the, I'm, I'm fairly sure we are leading in the NPU for the flagship SoC, both on the performance from TOPS and TOPS per watt point of view.
This will give us really a strong competitive edge, plus our effort to work with the ecosystem players in US. and in China to build the ecosystem so that the AI, the Gen AI developers or application developers can be much, much more easily use our capability. So, Brad, I hope I answered your question. And in addition, certainly we expect ASP to increase from our previous generation. That goes without saying. As to the non-China—
Right.
Customers, we, we're engaging, that's what I can say. We are having some good progress in certain devices, but I cannot really disclose more than that. I, we have some confidence in our capability in our flagship SoCs. By the way, the other thing which is not widely publicized is basically we are also being quite successful in expanding our flagship SoC into different segments, verticals, such as, for instance, automotive and tablets, sometimes, Chromebooks. These all do not represent as high a volume as the smartphone, but they do represent very nice incremental revenue and profits. Thank you.
Thank you.
Next one, we have Arthur Lai from Macquarie. Go ahead, please.
Hey, hi, Rick. Thanks for taking my question. Can you hear me?
Yes, please.
Hi. Arthur Lai from Macquarie Equity Research. First, I would like to congrats in Computex. You gave us a very great speech. And my question is to follow up regarding your long-term planning. It's about MediaTek's cloud market ASIC service strategy and IP. So I recall you show a slide in talking about the SerDes IP . Besides that, can you quantify how many important so-called essential IP we have on hand for the cloud CSP clients? That's my first question. Thank you.
Okay, well, we are, as I said, in the data center AI accelerator segment, the major capabilities, the first being computing IP, the second being the high-speed interconnect. The third and fourth are respectively, leading-edge process node and really advanced, very large area packaging capability. From computing, what we are seeing is the through using Arm's v9 core, and of course, their following CPU cores, our capability not only to work with them and work with us, in ourselves, to build those cores into a data center specific CPU subsystem. I think we have that capability. We have done something. We have actually through other segments such as automotive, we build a really much higher number of core CPU subsystem.
GPUs, we have both Arm and we have partners in NVIDIA. NPU, again, as I said, both in TOPS and TOPS per watt, which are the two critical factors. MediaTek has already put in seven years of investment, and each generation, we have made major progress. The key, of course, to scale it up to the level. For that, our initiative in automotive aspect will actually enable us to build really high TOPS NPU capability, because that's what ADAS will require. High-speed interconnect SerDes is all homegrown, 112G, 224G SerDes. They are, 224G is now being proven internally.
We are also investing in the next generation SerDes technology capability as well as the optical capabilities. And, you know, where we are doing pretty well as to the 3 nanometer, 2 nanometers. Of course, before we focused on mobile application, now we're moving full speed to HPC through working again, very, very closely with TSMC. Packaging also CoWoS, we add potentially other leading packaging capabilities. So I remain very confident and optimistic, because we also are very flexible. And we have advanced capabilities also with flexible business models. We want to work with our customers so they can get their more specific application-specific chips. Thank you.
Thank you. Very clear. Can I follow up one question? On the resources part, you need a good team, right? So how many R&D engineer we are going to add globally? And, roughly, how much percent team we need to- we can internal transfer from other team? And also, can we also give the global investor the guidance of the R&D spends and the OpEx trend, followed by this long-term planning? Thank you.
All right. Thank you. Good question. You sound like our board. The investment, as such, we are investing strategically, heavily. There are two facets of investment from that point of view. One being, we're putting definitely a new, not just headcounts, really key talent. We're hiring, and we have hired several key talent in the data center area, from architects all the way to packaging. We are certainly increasing our resources in the computing area. We certainly... And NVIDIA is very strong in executing our chips design and integration. We are certainly reallocating all those resources into, which means prioritizing into the AI-related chip development.
But I expect, I think our R&D now is about 22%, 22% of our revenue, about. I expect this ratio not to even as we continue to growing our revenues, this R&D ratio will not go down. It may go up somewhat. That is kind of a dynamic which as we move forward. Thank you.
Thank you. Great.
Next one, please welcome Gokul Hariharan from JP Morgan. Go ahead, please, sir.
Yeah, hi, good afternoon, Rick. Thanks for taking my questions. First of all, maybe ask a bit on the expanded partnership with NVIDIA. Could you talk a little bit about the extent of the partnership now, especially in auto? I think we already start to see some of those signs. Also, in other areas of compute that you've talked about, and there is more coming in this area. And related to that, could you also talk a little bit about the extent of design wins that you have secured in the automotive side, given that you have new Dimensity Auto SoCs and auto products that you've launched over the last two to three months?
I think about four to five quarters ago, we got an update about $200 million revenues from auto in 2023, and $1 billion of backlog. David, any update on those numbers, either on revenues for this year or for the updated backlog, given you seem to have a lot more progress on the auto side?
Okay. Boy, it's a long question. Let me go start with the partnership, NVIDIA partnership. The automotive partnership is executing, and we are executing the first chip on track, right on track. We expect chips to be out early next year. We have this is a very—it's a very, very, very good chip with all the computing power from NVIDIA's GPU and from our CPU and the ISPs. It's a, well, it's, it's a really, I, I, I would think a very competitive chip. In addition, we are also penetrating the automotive market with our, I said earlier, for instance, the flagship SoC or the premium SoC from our mobile business.
They are, and we're making good inroads also, and also in China. I cannot really, because of design wins, we, we do have good design wins, ongoing in execution. I cannot disclose them, as of now, but when customer, so automotive, they, when they launch their, cars, then we can share more, hopefully soon. And again, automotive is a long-term, business, takes time. We all know that. And we are, as, as I said many times, we are committed, and we are, we put in the, resources, and we look forward to 2027, 2028 beyond time for the major fruits. Collaboration with NVIDIA of co- goes, beyond automotive, in other, in other area, computing related.
We, in the Computex time, I think I had an exchange with Jensen on the stage, and, yeah, we also agreed that both company should also explore the opportunity in the data center area. We certainly are working toward that. The other part, David, you want to—
I can probably comment about our earlier comments, both from the addressable market, the revenue target, and also the backlog. I think first things first, I think for the revenue target for this year, I think we are fairly comfortable we can achieve that, probably even with some upside, just like Rick's talking about. There's several new design in, design win, and also overall the shipment situation. And due to the new design and design win, right now, I think the backlog, compared to the last time, actually is actually better. But we probably will not announce the backlog this quarter. What we're going to do, actually, probably by end of this year, is we're going to announce our new backlog numbers. So, please withhold that, and we're going to announce that next time.
Okay, that's very clear. So my second question is on the enterprise and data center ASIC. MediaTek has been in this business for a while. I think you had some success. It kind of did not come through in the last time, last time round. This time round, I think looks like better chances of success. Rick, could you help us understand what is MediaTek's edge right now compared to the dominant market leader? It feels like that's the market leader that you're kind of competing for many of these sockets. I just wanted to understand how is MediaTek kind of creating a niche for itself compared to this dominant player?
Also, just wanted to understand, when you talk about flexibility in terms of the data center business, what does that actually mean? Do you mean that, is it a flexibility on profitability terms, or is it a flexibility in terms of business model, that you could potentially tackle different parts of the chip manufacturing and design process, that the market leader may not be willing to tackle? Thank you.
Okay. Thank you. I think, yes, we, we've been in this market for a while, for several years. Right now, what we're seeing is a clear and large TAM going forward up to 27, 28, and beyond time. And I think everybody kind of follow the similar number, like $45 billion TAM in 2028. Sooner, maybe. And that TAM is very large. And what we are working toward with is with our... I do not plan to repeat what I just said earlier about our capabilities. To compete with our technical capabilities, as well as our flexible business model... to grab a significant share of that TAM, we understand the competition is strong.
But this business, this TAM, is big enough to have more than a couple of three major players as the customer request, basically, request. So I think opportunities certainly are there for us to grab. And the MediaTek, as far as I can see, has one of the best, is, we're one of the best equipped companies to fight for that very large TAM. And our flexibility in terms of business model, we will not insist on certain, just only one way of doing things. It is built in our DNA that we can be, we will be flexible while we deliver whatever we commit to deliver. That is how there's really no secret formula, I don't think.
It's investment in technology capability, build partnership, provide value to our customer, and execute. Thank you.
Thank you.
Next one to ask questions, Charlie Chan from Morgan Stanley.
Thank you. Good afternoon, Rick, David, and Jessie. My first question is about the AI smartphone recruiting cycle, because we all see Apple introduce Apple Intelligence, looks like a great potential. So Rick, do you think your China smartphone partners can really deliver a similar level of user experience with kind of acceptable efficiency? And when do you think they can get to that level? Because my concern is that for China domestic, they may not be able to use the Gemini Nano. But if you look at Apple, right, they have their own foundation LLM integrated with their own iOS, own processor.
I'm a little bit concerned that the integration of those stuff for China's Android smartphone could be pretty challenging. I want to get your thoughts. Thank you.
Yes, I understand your concerns. I think your concerns, at least part of which is quite valid. However, certainly, I think the China market, or our customers are not sitting idle. You talk about model building. Well, the Llama 3, open source Llama 3 is very, very good. It's as far as we can tell.
Mm-hmm.
And our customers are really adopting that, of course, very, very rapidly. And the people, I think in China, there's no lack of model development, startups and the large company. My view here is this is so new. And if you look at the offering of the applications-
Mm-hmm
... by different, including iOS. I mean, many of the applications are fairly similar between the different, say, AI smartphone suppliers. There may be some small differences, but with our customers' market share and the vast size of the China market, I remain very optimistic. I think the pie, the first thing is whether pie is big enough, and there's no question the pie is big enough. As far as we can tell, our customers and our team are working very closely to build the ecosystem. My understanding is, so they are building applications that will not only competitive on their own, but also very competitive in the China market.
Mm.
That's a very large market.
Mm. Okay.
Okay?
Yep. Yeah. Thank you. Yeah, look forward to seeing that. My second question is about your foundry costs, right? Because TSMC show their intents probably to reflect that some cost increase. And I think a couple weeks ago, I think C.C. Wei, you said the negotiation so far so good, right? I'm not sure if that's really good news for your company. But first of all, do you expect the wafer price costs to go higher for next year? And secondly, how do you justify those cost increase by your foundry partner? And lastly, how would that impact to your gross margin in 2025? Can you really pass through those additional costs, if any? Thanks.
Charlie, David here. I think TSMC made it quite public that actually they're gonna adjust their wafer pricing to everyone, to the industry. So you can consider that price hike or inflation, quote, unquote, "inflations" in the marketplace. So I think everyone in this industry all just having the same level of price ratio. So I think that's the background. But with that, I think that given this is the industry's a norm, so what we will do, we will definitely try to reflect the cost increase through our pricing, basically, especially with our new position to basically to the industry. So I think that's an ongoing process. But overall, I guess we will reflect the industry-wide cost increase to the back, actually, so to the market. I think that's the goal.
Is that 47% margin can sustain in that case?
I think for next year, again, actually, we haven't given out that 2025 guidance yet. But at least for this year, I think overall, 46-48, that was our earlier guidance we made on tech.
Understood. Great. Thanks, Rick and David. Thank you.
Next one, we have Bruce Lu from Goldman Sachs. Go ahead, please.
Hi, can you hear me?
Yeah, please.
Yes. Okay. So, Rick, when you talk about, like, you know, Edge AI, you talk about, like, MPU, a lot of IPs. But when you talk about, like, you know, current ASIC project, the key focus was our service. It seems to me that your value proposition for your Edge AI and customer ASIC are a little bit different. Does that mean that your profitability is different, or does that mean that your current stage for the customer ASIC is only temporary, you're gonna move into a more complementary or more complete solution for the customer ASIC in the future? Or, you know, you talk about like NVIDIA's collaboration.
You know, since NVIDIA doesn't want to do the customer ASIC, is it possible that you work with them for the customer ASIC, for the data centers?
Well, with that, indeed, that's why we talk not only our technical capabilities, but also our flexible business models. Again, you have to also this, of course, this data center ASIC business has been there for quite some time. But the generative AI exciting TAM is fairly new. And different customers whether they're large and small are having different needs from their own requirements or their own TCOs. We again, like, in some cases, as you said, we focus providing leading-edge interconnect capability together with customers' capability. But don't forget. We our way of working with TSMC on the leading-edge process and packaging is a very critical capability also.
But I would expect in some of the cases there will be need for our other IP functions. But that would be more kind of like a case-by-case basis.
But-
Well.
I'm sorry, but you should charge whatever you provided, right? So when you provide different service, the profitability will be different. What is your threshold for your business, though?
Well, what I can say is, right now, we're seeing very good returns for such business. Our ROI is good. I have no, really no major concern from that point of view. The improvement on our operating income will be very good, in my view. And certainly, I mean, you mentioned also NVIDIA potential. Yes, that's certainly and something we would like to explore.
Okay. Thank you. So my second question is about the dollar content increase for the edge devices moving to AI. You have exposure to, you know, PCs, smartphone, you know, a lot of consumer devices as well.... So TSMC is talking about like 5%-10%, you know, increase in terms of die size. Do you see that as a general understanding for all the, you know, dollar content increase for the edge AI? You know, when they adopt AI, we should expect to see 5%-10% ASP expansion across the board? Or is that like, you know, different from... Is that very different from different devices?
I think, I think for now, the smartphone, especially the high-end smartphone, is probably the best reference. It has high volume and certainly also a longer history. I cannot give you a specific number, obviously, but what I can say is the blended ASP is definitely improving. Actually, there's some data, third-party data in China, that the overall ASP for the smartphone is toward the high end. And we are also enjoying part of that shift into the, I would say, a significant shift toward the high end, and which also shows up in our ASP, blended ASP. Thank you.
Thank you.
Thank you, ladies and gentlemen. With the interest of time, we are going to take the last quarter for questions, and the last one will be Sunny Lin from UBS. Go ahead, please.
Thank you very much for squeezing me in. Good afternoon. Well, so my first question is on the cloud ASICs. Obviously the addressable market is large in terms of the revenue, but how should we think about the addressable market for profits? Where I'm coming from is some of the new projects from the hyperscalers appear to be low margin for CPU or for networking. And so I just wonder, for MediaTek, given you may have a higher threshold for margin, will there be a good number of new projects with good profits?
Well, I think it's like, as Rick, our CEO, explained, overall, the key was really ROI. It was actually for the ASIC project, sometimes the accounting may be different, depends on how you recognize the revenue. The key is really whether or not, because this is case by case, whether or not that project have provided positive ROI, device improving operating margin and enhance EPS. I think that's the key. So, so-
Got it.
Selectively, and we feel confident, actually, for those projects we pick, will provide very justified ROI, which improves operating margin and enhance the EPS, overall EPS.
I see. A quick follow-up is for CPU. Were the products mostly tied to ASIC efforts, or would you also look to come up with your own server CPUs that you could sell to hyperscalers?
Right now, I would say, as far as we can see the requirements from customer, large or small, fairly kind of a custom, at least semi-custom type of requirements. An ASSP type of offering is, I think... I'm not saying it's not doable, but it's not easy. So right now we are not, at least, right now, we are not planning on that business model. Thank you.
Got it. Thank you. My second question is on the foundry supply. Clearly, TSMC's leading edge, including 5- and 3-nanometer, is in tight supply through H2 and may continue to be the case through 2025. And so would that be a potential bottleneck for you to grow significantly in 2025?
I think so far, I think we've been secure for 2025 capacity what we need, so there will not be a bottleneck for us.
Got it. Thank you, David.
Okay, thank you again, ladies and gentlemen, for all your questions. Now, I'll hand it over to Miss Jesse Wang for a closing comment. Miss Wang, please go ahead.
Ladies and gentlemen, this concludes MediaTek 2024 Q2 conference call, and an audio replay will be available one hour after the call at the investor sections of MediaTek's website. We would like to thank you for your participation, and you may please come out.
Once again, we thank you for your participation in today's conference. You may disconnect now. Thank you and goodbye!