Welcome to the MediaTek 2023 second quarter investors conference call. Financial results and presentations for today's call are available on the investors section of 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 CEO, and Mr. David Ku, MediaTek 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 the actual results materially different from the statements. The presentation material supplement non-GAAP financial measures. Earnings d istribution will be made in accordance with financial statements based on GAAP. For details, please refer to the safe harbor statement in our presentation slides. In addition, all content 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 content provided in today's call.
Now, I would like to turn the call to our CFO, Mr. David Ku, for second quarter financial results.
Thank you, Jessie. Good afternoon, everyone. Now let's start with the 2023 second quarter financial results. The currency here are all in US dollar. Revenue for the quarter was $98.1 billion, up 2.6% sequentially and down 37% year-over-year. Gross margin for the quarter was 47.5%, down 0.5 percentage point from the previous quarter and 1.8 percentage point year-over-year. Operating expense for the quarter were $31.9 billion, compared with $31.5 billion in the previous quarter and $37.6 billion in the year ago quarter. Operating income for the quarter was $14.8 billion, up 2.7% sequentially and down 62.4% year-over-year. Non-GAAP operating income for the quarter was $15.4 billion.
Operating margin for the quarter was 15%, the same as the previous quarter and down 10.2 percentage point year-over-year. Non-GAAP operating margin for the quarter was 15.7%. Net income for the quarter was $16 billion, down 5.2% sequentially and 55% year-over-year. Non-GAAP net income for the quarter was $16.6 billion. Net margin for the quarter was 16.3%, decreased 1.4 percentage point from the previous quarter and 6.6 percentage point year over, over year. Non-GAAP net profit margin for the quarter was 16.9%. EPS for the quarter was $10.07, down from $10.64 in previous quarter and $22.39 in year-ago quarter. Non-GAAP EPS for the quarter was $10.42.
A reconciliation table for GAAP and non-GAAP financial information is attached in the press release for your information. That concludes my comments. Thank you.
Thank you, David. Now I would like to turn the call to our CEO, Dr. Rick Tsai, for prepared remarks.
Thank you. Good afternoon, everyone. MediaTek's second quarter revenue and gross margin both were above the midpoint of our guidance ranges, with three revenue groups growing sequentially. In the first half of the year, the semiconductor industry, including MediaTek, was affected by weak global demand, which led to a lengthened inventory digestion cycle. Recently, we observed that customer and channel inventories across major applications have gradually reduced to a relatively normal level. Recent demand from our customers has shown certain level of stabilization. However, our customers are still managing their inventory cautiously as global consumer electronics end market demand remains soft. For the near term, we expect our business to gradually improve in the second half of the year, and I will discuss each of our three revenue groups' business outlook in a moment.
Before that, I'd like to say a few words about some of our recent developments and industry trends that will fuel MediaTek's mid to long-term growth. We are very excited about the recently announced partnership between MediaTek and NVIDIA to develop a full-scale product roadmap for the automotive industry. In MediaTek Dimensity Auto platform, we will integrate NVIDIA's new GPU chiplet with AI and graphics IP into our SoC to expand to the premium smart cabin segment.
This collaboration will enable MediaTek to offer an entire spectrum of smart cabin and cockpit functions. With cutting-edge graphics, AI, safety, and security features. In addition to the full range of smart cabin solutions, our Dimensity Auto platform also include connectivity, auto drive, and components. We have recently received heightened interest from automotive customers since the partnership announcement two months ago. We believe our industry-leading low-power processors and 5G Wi-Fi connectivity solutions, combined with NVIDIA's strong capability in software and AI cloud, will help us become highly competitive in the future connected, software-defined vehicles market and shorten our time to market to accelerate our growth. Given the long design cycle of the automotive industry, we anticipate a more significant revenue contribution from 2023.
Another example of the digital transformation trend is the increasing popularity of generative AI. Today, the majority of the generative AI processing is only performed by cloud computing. We believe that they towards distributing generative AI inference workloads to edge devices like smartphones and IoT for better privacy, lower latency, and lower operating costs. By doing so, edge devices can perform generative AI on devices for seamless AI capability. They also connect larger generative AI models in the cloud by prompting more complicated AI functionalities. Edge device makers will need to adopt high computing, low power AI processors, and faster, more reliable connectivity to enhance computing capabilities and lower connection latency. As a leader in various edge devices, MediaTek powers approximately two billion connected edge devices in the market every year.
Our leading product portfolios, investment in advanced nodes, as well as next generation connectivity, position MediaTek well to capture this increasing trend as a key enabler and beneficiary of edge AI. With that, now let me talk about the recent business of our three revenue groups. Mobile phones accounted for 46% of total revenue in the second quarter and grew 3% quarter-over-quarter. The result was slightly better than our prior expectations, as customers' demand for 5G SoC improved during the quarter. With relatively normal customer and channel inventory levels, we anticipate the growth to continue in the third quarter. On the product side, we recently announced the mainstream 5G SoC Dimensity 6100+, which belongs to a new 5G segment with more affordable price point to cater to global 5G demand, transitioning from 4G models.
Smartphones incorporating Dimensity 6100+ will begin to ramp up, starting from the third quarter. For the flagship segment, we are on track to increase our shipment and revenue this year. Multiple Dimensity 9200+ smartphones have been well received by the market and contributing robust revenue in the third quarter. Our next generation flagship SoC, to be introduced in the upcoming months, will further advance the overall performance and integrate our latest APU with the capability to perform generative AI features on the device. We have been closely working with our customers for design-ins and expect smartphone, powered by our next generation flagship SoCs, to hit the market by the end of the year. In fact, not only the flagship SoC, but all the 5G smartphones SoCs across all tiers we are shipping today are equipped with MediaTek's APU to perform various AI features.
The trend of more complex AI instructions is likely to be a catalyst for smartphone replacement demand, which will enhance our product mix and support rising trends. Now let me move on to Smart Edge Platforms. This group grew 2% sequentially in the second quarter and accounted for 47% of revenue. Connectivity demand was stable in the second quarter and will improve moderately in the third quarter. Notably, MediaTek's Wi-Fi 7 solutions have been adopted across various platforms. For example, high-end retail routers utilizing our Wi-Fi 7 have been available in the market since the second quarter. Premium notebooks and broadband devices are scheduled for re-release in the third and fourth quarter, respectively, with a stronger ramp-up in 2024.
For TV, due to customers' inventory preview to take advantage of more favorable panel prices in the first half, the demand in the third quarter has slowed down. As for ASICs, we recently see growing enterprise ASIC business opportunities in AI and data center markets. With our strong IP and SoC integration capabilities, we aim to continue to grow this business in the future. Overall, as most of the consumer electronics demand remains soft, we anticipate Smart Edge Platforms business to remain flattish in the third quarter. Now, moving on to Power IC, which accounted for 7% of total revenue in the second quarter and grew 4% quarter-over-quarter. We expect Power IC demand in smartphone to improve in the third quarter. In summary, as we stated in the beginning of the call, we expect our business to gradually improve in the second half of the year.
For the third quarter, we expect revenue to improve in smartphone, connectivity, and PMICs, offsetting the decline in TV and other consumer products. During this demand cycle, we continue to execute our strategy of balancing among market share, revenue, and profitability. With that, we expect our third quarter revenue to be in the range of NT 102.1 billion-NT 108.9 billion, up 4%-11% sequentially, and down 23%-28% year-over-year, at a forecasted exchange rate of 30.7 NT dollars to 1 US dollar. Gross margin is forecasted at 47%, ±1.5 percentage points. Quarterly operating expense ratio to be at 32%, ±2 percentage points.
We are on track to reduce the full year total operating expenses by mid-single digit % year-over-year, while maintaining the investment in the key technologies and key projects for the mid-to-long-term growth. This concludes my prepared comments. Thank you.
Thank you, Rick. Operator, we are now ready for Q&A. May we please have the first question?
Thank you. Ladies and gentlemen, we are now in question and answer session. If you would like to ask the question, please press star one on your telephone keypad. Please ask your questions after your name is announced. Please limit your questions to 2 at a time to allow more participants to join the discussion. After 2 questions, we will move to the next caller. Should you have more questions, please press star one to come back to the queue. To cancel your question, please press star two. As a reminder, it is greatly appreciated that you turn off speakerphone mode of your device to prevent possible echo effect. Thank you for your cooperation. Now, please press star one to ask the question. Thank you. Our first question is coming from Randy Abrams, Credit Suisse. Go ahead, please.
Hey, yes, thank you. Yeah, I want to ask the first question, just with all the attention on AI, two parts to it. First, want to see if you can give an update, if you're seeing potential upside yet to the content per device, from that AI engine, both in smartphone and smart edge. And if, if you could also talk on the ASIC. I think in the prepared remarks, you talked about some opportunity on AI and ASIC. If, if you could give a bit more update, if you're seeing more breakthroughs that could make this a more meaningful contributor, and how you feel your IP around AI, SerDes, networking, suits you to get some of the larger hyperscale sockets?
You go ahead, Joe.
Okay, Randy, why don't I talk about edge device AI first? Because your question basically talking about edge device AI and also, the ASIC opportunity. What am I updating about the edge device? Actually, in reality, right now, all of our smartphone, especially the flagship smartphone, we have actually the GA, Generative AI capability in there already. Basically, that's the transformer model.
... and so a few function or feature already in landing in our previous generation flagship. For the coming third generation flagship, I think we will just fix these, put on more functions on that, and we will have some demo basically in the coming months. We're going to update you guys more about that trend. Your question specifically asking about the dollar content. So far, it's actually really just part of the function. I think in terms of the dollars, you know, it's not a huge content increase yet. Our view is actually going forward when more and more distributed AI is being performed on the smartphone, I think the requirement for more computation power on AI will increase.
So far, we probably won't be able to provide any sort of taller content on that, but trend is definitely positive. That's definitely one.
Yeah, I might add to Randy. Other than smartphone, well, basically, the basic OEMs, the customers, are really also working very diligently to to implement the various models to to, to, to, I would say, explore the applications in their edge devices, actually be they smartphone or, for instance, automotive, the EV. We, we- I think we are on the we-- actually, we have seen some demonstrations by, by some of our customers, which I think, again, as David just said, we, we'll, I think we'll induce, we enable much more, much higher interest from the end users. I think that's the key. Yeah, we, I think in a virtual, virtual cycle, that, that will induce the demand and the shorten the replacement cycle.
As to the ASIC opportunities, what I can, I cannot be very specific. What I can say is, it is quite obvious and quite natural for all the hyperscaler guys, data center, to to accelerate their their chip development to for the generative AI enablement. We are seeing certainly a heightened activity levels. And we have the all the IP portfolio available. For instance, the high-speed SerDes, 224G. We are, we have obviously a very advanced advanced node design capability, and the packaging capability. We expect to, we're in deep discussion with people, and we'll see how it goes, hopefully, positively, in the future.
Okay. No, good luck on that. I'll, I'll just have one quick follow-up on that part. I, I think in prepared marks, you mentioned it's an enterprise ASIC. I just wanted to clarify it's enterprise versus cloud, where you're seeing the initial traction?
Randy, actually, so we, we hear some background noise, so can you repeat the question again? We can't hear clearly.
Yeah. The, the question was the ASIC traction. Is it enterprise or I think that's in the prepared remarks, or are you seeing traction with the cloud providers on their ASIC?
Oh, okay. I think for enterprise, we continue because we have certain ASIC products with enterprise customers are mostly in the switching ASICs. For the new opportunities, those mostly rely with the hyperscalers.
Okay, great. The, the second question, the margin outlook has remained very resilient, and I think that's how you've been trying to guide it, balancing share and pricing. If you look ahead at, at the margins, do you think as we go through continued slow environment, some of your competitors have inventory, how do you see margin range, if you think these levels continue to hold from what you can see? And do you see any help coming through from input, supply pricing, or is that still inflationary?
Randy, first of all, I think before we comment about the growth margin, maybe we can probably provide some color about the current, the competitive landscape or competitive situations out there. Well, first of all, the long story short, we didn't really see any material change about the competition landscape right now. On the other hand, as you can see, both from our second quarter growth margin, which is reported already, and also our guidance for 47% ± 1.5%, they're all showing a stabilizing, I would say, the balance, a dedicated balance, actually, is that we are aiming for, but among revenue, margin, and market share.
I think, and also another, I would say, the good news from margin perspective is both from the channel side, the customer side, and the channel side, and also from the vendor side. I think it's all inventory level incoming down to a pretty, heavy, healthy level. You won't see any players trying to rush out the inventory by taking aggressive pricing. With that, I guess, the guidance we give out for third quarters, even though we cannot provide a guidance for fourth quarter, but most likely, I think that should be a good reference, for the fourth quarter's, gross margins, you know, head start as well. Yeah.
Okay. Supply side, how is the inflation or deflation?
You mean inflation, deflations about the, the, the financial market or?
Sorry. Like foundry and back end and, some of your, your cost structure.
Well, I think next year's foundry cost actually is still in discussion, but overall, given the market situations, we are seeing probably, hopefully it's going to be flat or slightly up. In general, actually it's a, it's a manageable situation, so it won't be a extreme factor so far.
Okay. Great. Thank you, David. Thank you, Rick.
Thank you. Next one, Brett Simpson from Arete Research. Go ahead, please.
Thanks very much. Rick, I want to just ask you a bit more about AI and your APU plans.
Brett, Brett, Brett, your, your voice are all breaking up. We, we can't really hear you.
Can you hear me? Can you hear me now? Apologies.
Still can't. Yeah, it's still a lot of echoes and also it's breaking up voice.
Okay. I'll, I'll, I'll, sorry. I'll dial back and try again.
Maybe we'll move on to the next.
Yes, of course.
Thank you.
Thank you. The next one is from Laura Chen of Citi. Go ahead, please.
Hello. Hi, good afternoon. Thank you for taking my question. My first question is also about the AI-related opportunity in the ASIC business. I'm just wondering that other than current consumer-related application, I think Rick already mentioned that you are progress in the enterprise side. I'm just wondering that, that, other than that proprietary IP and also the advanced node design capability, can MediaTek also provide a CoWoS design service for the ASIC project? As we see that actually for this kind of ASIC chip, probably the advanced packaging, together with a lot of like interface or high bandwidth memory is also quite important. That's my first question. Thank you.
Yes, we do. We are in deep discussion with potential customer, of course, includes the advanced packaging capability, such as CoWoS, and also the all the inter. How should I say? Inter, IT, interconnect, capability. We, as I said earlier, we have a very complete IP portfolio, and we have the advanced node and advanced packaging capabilities, of course, through working with our key supplier.
Okay.
The answer is affirmative to your question.
Yeah. Following that, we know that TSMC also provide quite comprehensive, like, IP pool for that kind of advanced packaging. For MediaTek's current project, potential project, will you leverage that kind of open IP, or you would tend to focus more on your proprietary solution?
We actually, we work most close with TSMC. But also we will also look at the alternative sources when they are technically competent.
Got it. Any idea, like, when we will see that probably, more mature or these kind of projects, to start to bear fruit?
Usually, this kind of project takes, at least, 1.5-2 years.
Okay. Got it. Thank you very much. Also my second question is also about the edge AI. We know that for current, our smartphone SoC solution, we already embedded like APU or neural engine. Do we need to see that OS provider, like Google, to really enable the application to see that edge AI getting more mature or any other trigger you are close watch?
Yeah, Laura, actually, it's in addition to our own APU, I think right now, especially for the GAI or LLM model, it's very important we're actually porting some of the Open AI model on our smartphone. I think currently it's we've been capable of, for example, Meta's Llama 2 model, Stanford's Alpaca model, Berkeley's OpenLLaMA, and also Stable Diffusion. I think that all is up and running, is actually being portable on our edge device. I, I'm sure actually there will be more model in running. The overall, I think the cycle is actually, we have the hardware capability. We make sure the, you know, certain appropriate model can be run properly and smoothly on our device.
We provide that, you know, together with our development tool to our customers, so they can develop, the feature and function they are looking for, for the generative AI. Right now, actually, it's again, like I said, for the third generation flagships, we've been in serious discussion with our customer, trying to have some landing features, GenAI landing feature on the phone. Before that, we will definitely demo few possible ideas in, I think, October, November timeframe.
Okay, thank you very much. That's very clear. Thank you, David.
Thank you. Next one, Gokul Hariharan, JP Morgan. Go ahead, please.
Yeah, hi. Good afternoon. Thanks for taking my question. My first question goes back to smartphone. Could you talk a little bit about what is the demand sentiment among your customers? Is there any improvement on China smartphone demand? Do we feel that China smartphone market is likely to remain at these depressed levels, given there's not really been any major recovery? Do we think that next year we will see a better China smartphone market? What are you hearing from your customers, and what is your own read? On that also, seems like Huawei is kind of like the one brand which is kind of coming back and growing quite nicely in China. My understanding is that MediaTek currently doesn't have a license to sell to Huawei.
Do you think that is something that could change over the near to medium term, given Huawei seems to be gaining a recently, a big share of the high-end market in the recent quarters? That's my first question. Thank you.
On the demand for the smartphone, I said we, we, we're starting our remarks. Basically, we're seeing, I would say, it's a good improvement in the shipment of the smartphone. Well, the inventory, inventory, lowered, I think both, actually, a fairly pretty well, inventory going down pretty well in the second quarter. We believe that inventory reduction will continue into third quarter at a very healthy pace. As such, the the reordering of the SoCs has moved up, moved up in its pace also. The end market demand, I think, as I where still remain cautious. We, we also believe that's the case. However, we don't believe the end demand is going to get worse.
It probably will grow, but grow at a moderate rate going forward. That's the how we view the the smartphone outlook. As to the impact of the Huawei, I want to be very clear that we do not have a license to ship to Huawei of smartphone SoC or flash we SoC to Huawei. And the impact of Huawei in the market, I think, the most important part is to look at the segments they are playing in. The most importantly, in the very, very high end, their Mate and their, what's it called? There are two of us, two series, very high-end.
P40.
Yeah. Yeah. So I think we believe the competition directly against our chips remain fairly far apart. We do not see a major impact from Huawei from that point of view. Thank you.
Okay. That's very clear. Thank you. My second question, could you talk a little bit more detail on the NVIDIA partnership in auto? What will be the structure of this partnership? Is it going to be like a JV structure, or you would be selling the chips, but you will be face, paying some kind of IP or a royalty fee to NVIDIA for the IP and the chiplets that you use from NVIDIA? NVIDIA already has their own standalone auto business with a fair number of autonomous drive and some ADAS-related design wins. In those areas, how does MediaTek enter?
Like, does it mean that MediaTek doesn't really participate in that ADAS autonomous drive market, or, is that structure also going to change for NVIDIA as well, that they will, they will kind of partner with you for many of their future autonomous drive and ADAS-related design wins also?
Okay. There is no joint venture structure. I want to say that clearly. It is a partnership structure. By that we mean at first from a technology point of view, as you just said, we integrate the IPs, the key IPs, such as GPU, such as their AI capability and their software technology, into our SoCs. We provide the other computing capabilities and the multimedia functions or the interconnects. The chips made this way will be marketed by MediaTek. We know NVIDIA has already their product portfolio and their customer base already, so we work basically in a partnership manner to go to the customers.
In that way, together, we can provide, as I said, the full spectrum of automotive in-cabin solutions from the very high end to the, to the mid-entry end, to our customers, potential customer base. Of course, in addition, we provide the other connectivity capabilities such as 5G, Wi-Fi, and some other component capability for, for a full portfolio. ADAS, I think that's one area we will work, work together on. Right now, the first thing is to have a very successful, very competitive in-cabin solutions for the market as soon as we possibly can. Thank you.
Okay, thank you very much. Thanks.
Thank you. Next one, Charlie Chan of Morgan Stanley. Go ahead, please.
Thank you. Good afternoon, gentlemen, and Jessie. My first question is also about the AI on the phone. First of all, do you envision that GAI, generative AI, can be a killer app on a phone, or is just a nice to have a feature? Can I get some, some insights from the management?
Yeah, Charlie, I think, now it's actually right now still in the early stage of GAI, but we already see the how powerful it is. Long story short, we do believe that could be, I won't use the word killer application. I think there will be a really nice features which could either increase the dollar content of the semiconductor or to the minimum, shorten the replacement cycle. Because actually, there's lots of useful functions out there. I think the key word here is not really, GAI only on the smartphone. It's something we call the distributed AI. Because for certain functions, for example, you know, right now, for Stable Diffusion, you can do some Photoshop on your phone with the natural language, on the edge.
Sometimes due to privacy issue, you don't really want to send all your pictures, which you want to do some Photoshop or likes activity on that, all to the cloud and someone processing, finish it and come back. Another example is actually when we talking to our phone customer, they're talking about some, some of the phone assistant to help you navigate through different applications. That actually is not as powerful as ChatGPT, probably is not even in the same league, but it's become very handy personal assistant. We, the key word truly going forward is the distributed AI model.
The phone maker and also we actually provide a tool. We will figure out what's the right functionality, which will be performed on the edge, and for certain functionality will be performed on the cloud, and it will be seamlessly connected. With that, we do believe actually there will be a very key driver to either increase the semiconductor content or to the minimum, but that's a new features, shorten the replacement cycle.
Thanks, David. You, you mentioned-
And from the-
Sorry, Rick, please go ahead.
I have some, I've been thinking about that part.
Mm-hmm.
Of course, not, not meaning I'm working on that myself. I have talked to quite a few people. The feeling is this Generative AI, you know, through cloud and the edge devices, be they smartphone or PC or automotives, and then provide another platform. It's not the same, but the nature being kind of similar to the 4G, to the 4G era, which I mean, which provide this platform and enable and excite a lot of the developers and the entrepreneurs to come up with new applications and new business models. Of course, I think, well, if you think about 4G, 4G was launched back in about 2011 timeframe.
Then we, when the, then one obvious example being, for instance, TikTok, it, it just came out of nowhere, and it became a huge success. I cannot say there will be a TikTok, but the, but the, I think GAI provides in such a powerful capability that the many entrepreneurs will take advantage of that and come with applications. That, that's just my just my thoughts. Thank you.
Thank you. Very, very helpful. So I, I believe MediaTek has success in four markets far before the 4G era. I remember you kind of enabled those, the emerging markets, a lot of the interesting features. Based on the comments you just made, Rick and David, two realistic question, right? First of all, for the emerging markets, especially China markets, do you think the policy, you know, control is kind of a restriction to your future application developments? Also, to David's point, I really appreciate that distributed AI computing, but who should be this kind of a location of computing?
You said, the MediaTek, the platformer, to decide, what, what kind of a computing will be done by cloud, what should be done by phone, or it should be, AI developers, or OEM, makers. Can, can you give us some, some color? Thank you.
Yeah. Charlie, I think most likely will be a combination of all those, you know, parties you mentioned. You know, for example, when we talk about Stable Diffusion, some of our phone customers are talking about some function provided by phone OEM. The, in that case, they probably don't really need to work with the cloud guy, if you consider that's actually just on your phone. For sort of an internet player, of course, they have the cloud capability, but they are also talking about whether or not they can offload some of the computation on the phone. Bear in mind, every AI calculation on the cloud, the cloud guys or internet guys need to pay for that running expense. If they can somehow distribute part of that on the phone, then they'll be, quote, unquote, "free" from the service provider perspective.
There are also some incentives, even from the cloud service or Generative AI cloud service guy. Again, answer your question directly, it will be a combination of, you know, all those parties you mentioned.
Yeah, we do consider.
... I'm sorry. But also it really depends on the edge. What do we call edge device? I think here is a spectrum of edge devices, not just smartphone. Smartphone actually being the, probably the, the smallest form factor one. But despite its, the power the smartphone has, but smartphone does have its limitation in form factor. And for, and, along with, for instance, some limitation in the memory size, the phone can, can carry. But then you, if you move to automotive, then the memory size, for instance, can go up quite drastically. And the size of the model that can be implemented in that kind of a device is much different from a 7, let's say, 7 giga, 7 billion kind of a model. So again, this thing is so new.
Remember, ChatGPT came up in November last year, and it's already amazing that how much work and the has been already done both at the cloud and the at the edge devices. I think people are working to try to find that a balance or equilibrium from different devices. It also, that's my, again, my, my, my thoughts. Your, your question on government, I think that's of, of course, out of our league. I, I, I, we, we cannot comment really directly. Hopefully, the government also the successes, you mentioned China, so we use China. The successes that China had with their internet economy provides a good guidance for them and the GAI will prosper all over in the world.
Got it. Thanks. Thanks, Rick. Let me switch gear to second topic, if that's okay. It's about the smartphone replacement cycle, because the company just introduced your mainstream 5G SoC. Do you think that can bring down the price points of smartphone to a sweet spot? Do you think that replacement cycle will restart anytime soon? Because that replacement has been sell, right, for almost two years. Last point is really about the secondhand markets, right? Do you think the new phones from your mainstream 5G SoC can compete with the secondhand phone in terms of the pricing points? Thank you.
Yeah. Charlie, actually, I think earlier this year, as we kind of provide our view for the global smartphone shipment, it's actually coming down from, in the past, roughly 1.4 billion to 1.2 billion. I think part of the reason, I would say a big part of the reason, is actually the replacement phone. When we talk about 1.2, it's already been taken into consideration.... offer replacement phone, especially for the, maybe the iOS system. For the Android side, actually, we didn't really see a big impact on the, on the refurbished phone on that, at that point.
On the replacement cycle of the phone, on the other hand, in general, in the past, roughly is 18 months, but this time, actually, the cycle being extended to, you know, almost two years. We do believe actually, as time over time, people still going to get a new phone, so people still going to get a new phone. Most likely next year, we, we feel, in terms of replacement cycle, it should be on the positive side rather than actually just keep waiting, and, you know, still waiting on that.
Got it. Thank you very much. Thanks for taking my questions. Thank you.
Thank you. Next one, Sunny Lin, UBS. Go ahead, please.
Thank you. Good afternoon. My first question is on a long-term outlook. It's been a while since the last time the company provided the adjustable market forecast by segments. I wonder if now is a good time that the management could provide your latest expectations on the long-term adjustable market by mobile, ASIC, automotive, and the others?
Sunny, actually, we would probably normally provide that long-term stock in the year end. Maybe we delay that question to the fourth quarter. Normally, in the year end, when we're doing the annual planning, we will take a look about the long-term basically addressable market. Why don't we delay that to the next quarter's conference?
Let's apply these to the original text: "Got it. No problem. Um, well, so, uh, perhaps if I may, uh, given, uh, lots of, uh, investor interest, uh, in your, uh, collaboration with NVIDIA, uh, and automotive does, does have lots of potential, uh, in terms of the market opportunities. Uh, you mentioned, uh, revenue will become more significant, uh, from 2026 . So I think, uh, uh, if you have any visibility at this point, that maybe in 4-5 year time frame, how many automotive account for your total revenue?
Oh, let me get you. Yeah. I think total revenue this year for automotive, we are looking for $200 million-$300 million, ballpark range, depends on how it goes in the end of this year. Relatively speaking, as I think absolute scale, I think it's still high growth, you know, but the relative scale is still small. Currently, we, we, we don't have actually, we didn't really provide actually the, the forecast guidance. Another way we can kind of talk about for the design pipeline, design pipeline revenue right now is over $1 billion.
Got it. That would start to be realized, after 2026?
Yes, it will be beyond 2026. Yeah. Normally, design, as we know, so the, we call the SOP cycle will be probably 2+ years. Okay.
Got it. Thank you. That's very helpful. My second question is on platform. How should we think about the product mix over the next maybe 12 to 18 months? I think on one hand, China platform could start to see a better improvement on the volume side. To drive that volume, I think some people wonder if that could be coming from lower end. I think on the other hand, if we look at the new product launch by smartphone makers in China, I think there's increasing focus on high end as they pursue better profitability. Based on your engagement with China's smartphone makers, how should we think about your 5G product mix maybe in second half of this year and into 2024?
Probably the better way to think about that is actually think about 4G smartphone as a total, which including 4G and 5G. We are actually right now promoting pretty heavily for, you know, transforms or convert from 4G to 5G entry smartphone. That's why actually, I mean, earlier, I think the other analysts were asking about the new entry-level 5G. On a blended basis, actually, when you think about the product portfolio or segmentation as a whole, 4G in the bottom, you actually see 4G being converted operate into 5G entry. I think that's point number one. When you look at the totality of the smartphone, it should be getting better. If we're only looking at 5G, okay, I think we normally separate it as flagship, high end, and mainstream.
I think for high end, probably will stay over there. Okay, right now, we are making good progress on the flagship. Currently, for this year, I think our market share on the flagship, we forecast to be 20%+. That one, actually, with our new portfolio and also getting more and more market traction, we believe actually the market share, our own contribution from the flagship should be increasing. On the high end right now, it's actually we see that being stabilizing. I think for the segmentation perspective, I would say it's actually it's a pretty stabilizing, and 4G conversion to 5G entry, that should be positive to the overall smartphone product mix.
Got it. Thank you very much.
Thank you. Next one, Brett Ling, BOA, Merrill Lynch. Go ahead, please.
Thank you for taking my question. I have two questions. One is on AI and another is on the automotive. While we learn the upcoming on the AI proliferation beyond server AI, also by many other supply chain partners like TSMC, AMD and Intel. For near term, would you please share the current contribution and what are the key offering, offerings, and how fast do you expect it to grow for MediaTek, maybe, on the in the next two years? Should we expect the ASIC to come earlier than, well, smartphone AI? Thank you.
Sorry, can you, no. Again, sorry, there's some background noise. Can you just repeat your first question again? You're talking about smartphone AI, or you're asking something else?
Yeah. Basically, for the near term, we, I just want to ask that what are the current contributions from the overall AI for MediaTek? What are the current offering right now? How fast do you expect the AI to grow for MediaTek? Would that be, well, smartphone AI or ASIC? Thank you.
Okay. Like we explained earlier, so far for the smartphone AI, it's really just a part of the smartphone. We deal with still a separate AI chip. It's part of the die area, part of the function. We do believe actually that going forward, the contribution of the die area, which equivalent to the semiconductor content, should be increasing. You know, we didn't really have a separate contribution or analysis on AI contribution. The only way we can separate that will be on our ASIC business. Basically, for the ASIC business, if the end customer is AI, we can somehow categorize it as the AI revenue. Like we explained earlier, right now, there are several projects that are ongoing. Normally take 1-1.5 years to ramp.
So far there's no revenue, no direct revenue from AI. We only have indirect revenue from AI, indirect revenue from AI.
Got it. Thank you very much. Do you have a, well, estimate on when the AI will kick off?
We don't, we don't have a fixed plan, but probably the better way of thinking about that is actually is that, you know, all product right now have the AI function in there. Even though we're only taking smartphone, for example, we talked about, we've been porting a lot of open AI model on the smartphone. Because actually it's, we are doing something similar on other platforms as well. I think probably the better way to think about that, because after all, we are not just on the cloud trending side. If you're thinking about it's a separate dedicated chip only for AI, probably not the best way to take a look about our revenue or our revenue contribution from AI.
The AI is really the driver for demand and driver for more computation for our AP in general, for our AP in general. Probably that's the better way to look at that.
Thank you very much. My second question will be on automotive. I was glad enough to be able to attend MediaTek and NVIDIA's press conference during the Computex, and I pretty was pretty encouraged to learn the partnership. May we follow up the progress of the development? Also, I remember the firm say that the partnership will target every single chip in an automotive and provide one-stop shopping platform for clients. Would you please share with the target markets, how the target market will expand? Should we also expand our power management IC, also penetrate into some of the automotive as well in the next few years, too? Well, maybe a slight follow-up is on the I know we are not going to build a JV, so and then that will be marketed by MediaTek.
Well, how would the partnership rework with MediaTek and NVIDIA? Thank you.
Okay. Well, first of all, actually, it's like we explained during the Computex, the first project is cooperation between MediaTek and NVIDIA. It's ongoing as well. To be precise, I think that product will be carry MediaTek brand name, and MediaTek will sell that product. But within that product, we are getting, you know, GPU chiplet from NVIDIA, but the whole thing will be MediaTek revenue. I think we are targeting to have the first time revenue somewhere toward the end of 2025, give the first product. There will be few more product right now is on the planning and kicking off. Overall, I guess for the revenue through this quote, unquote, partnership, probably will start to ramp in 2025.
I think that's the point number one, mainly on the cabin side. Mainly on the cabin side. In addition to that, like our CEO explained, we'll also leverage and partnership work together on NVIDIA existing business portfolio. For example, they have ADAS, they have a high-end stack cabin, and all those solution or platform will have MediaTek socket opportunity. For example, it could be 5G, telematics, it could be connectivity, it could be gaming, it could be something else. For example, even for, you know, display related. We will also trying to realize those synergy as well. Overall, like we explained during the Computex, I think this partnership is very synergetic and complementary, and both from technology and product portfolio.
Either way, you can think of the cabin as one product, but for other part, it will not be joint product portfolio or joint product development, but will be joint go-to-market and joint business platform. I think that's probably the better way to think about that.
Thank you very much. That's very clear. Just one slight follow-up is that, should we expect some potential collaboration opportunity beyond automotive with NVIDIA, say AI in the well, long, long run? Thank you.
Why not? We are focusing on getting this in-cabin chip out and selling. We have also, we are also working on, next, next ones, for, for complete, portfolio, roadmap. I cannot say, we don't, collaborate further. Yeah. Right now, we, we just want to make sure this automotive part work. We need that success.
Thank you. We're definitely looking forward to that. Thank you very much for the color.
Thank you.
Thank you. Ladies and gentlemen, we are going to take the last question. The last one is from Brett Simpson, Arete Research. Go ahead, please.
Yeah, thanks for squeezing me in. Can you hear this time?
Yeah, perfect.
Okay, great. Rick, I wanted to just come back on the AI side, the smartphone AI strategy. Can you talk about what size and model MediaTek can support with their APU next year? I guess when I sit back and look at this whole strategy for MediaTek, you know, developers that can offload AI workloads onto the phone, they save a lot of money not having to run their application in the cloud. So how, how does MediaTek get paid for enabling these models to run locally, on your APU, on your software stack? Is this something you can monetize beyond selling, selling, selling a chip? Thank you.
Well, the models, actually, I think David brought it up earlier. Usually the seven, six, seven billion parameters models such as Meta's Llama 2, Alpaca, you know, Stable Diffusion, et cetera, et cetera, for the smartphone. As I said, I believe a larger model, a larger size model will be implemented into PC and the automotive. I'll tell you right now, tell the truth, Brett, again, this is very new, and we are all just working very hard to enable the Generative AI to work well with the edge devices, without which I think it's just very shaky.
So I, I must say, we haven't. What we believe, at, at least first order, is the once people, once the users and some of the probably what they call KOLs, once they find out how to use generative AI for new applications and the to, to appeal to the general public, I think the most important replacement cycle being shortened. Then the, we, the, the, the, TAM will get back to a, a more better growth pattern. That's, that's the first order thinking that we have. I hope I'm answering your question.
Yeah. Yeah, that, that helps a lot. I guess maybe just to follow on from that, can you give us your perspective on how APUs drive up silicon content in phones? I guess it's, it's difficult for us to, you know, understand the extent to which this drives up your ASPs in smartphones, but, I mean, are we talking about this going up 50%, 100%, 10%? Any sense as to how you think AI contributes to your ASP expansion over the next one, two, three years? Thank you.
This is a very good question. Again, Brett, you, you have to bear in mind any of the smartphone chip that we're discussing here today for this year and for next year, I don't... I mean, it's not just us, other people too, are already in the. Actually, we, we, this year we have silicon out already, and next year is well, very much in design. It was planned just about a year ago, when, when the Generative AI was not, was not very prominent in people's mind at all. Anything that you hear right now is because we're already putting a quite powerful APUs in the our SoC, and we certainly are still working with our customers, trying to implement Generative AI with these already defined, I should say, APUs.
I think I, I, my, my feeling is the for the generation, maybe 2025 phone chips, there will be more another hierarchy of the processors architecture. You see, there's a CPU, GPU, and the APUs, and the proportion of which should be implemented in to optimize the GAI applications, I think, needs some work. We have to make that work so that both we have the capability, the computing power for the GAI applications, while we do not unnecessarily increase the cost by too much. Any of those silicon space takes money. Distribution of the silicon area among CPU, GPU, and APU is a very difficult subject, but that's what we're working on now.
One thing is quite sure, APU size of the silicon will increase by quite a lot, by quite a bit.
Yeah. That's very helpful. Thank you. Thank you, Rick.
Thank you. Ladies and gentlemen, we thank you for all your questions. I'll hand it over to Miss Jessie Wang for closing comment. Miss Wang, please go ahead.
Ladies and gentlemen, this concludes MediaTek 2023 second quarter conference call, and an audio replay will be available in 1 hour after the call at the investor session of MediaTek's website. We would like to thank you for your participation, and you may now disconnect.
Thank you. We thank you for your participation in today's conference. You may now disconnect.