Taiwan Semiconductor Manufacturing Co Ltd

TWSE:2330

Earnings Call Transcript

[Jeff Su, executive]

[Foreign Language] Good afternoon, everyone, and welcome to TSMC's Fourth Quarter 2024 Earnings Conference and Conference Call.

This is Jeff Su, TSMC's Director of Investor Relations and your host for today.

Today's event is being webcast live through TSMC's website at www.tsmc.com, where you can also download the earnings release materials. [Operator Instructions] The format for today's event will be as follows: First, TSMC's Senior Vice President and CFO, Mr. Wendell Huang, will summarize our operations in the fourth quarter 2024, followed by our guidance for the first quarter 2025. Afterwards, Mr. Huang; and TSMC's Chairman and CEO, Dr. C. C. Wei, will jointly provide the company's key messages. Then we will open both the floor and the line for the question-and-answer session.

As usual, I would like to remind everybody that today's discussions may contain forward-looking statements that are subject to significant risks and uncertainties which could cause actual results to differ materially from those contained in the forward-looking statements. Please refer to the safe harbor notice that appears in our press release.

And now I would like to turn the microphone over to TSMC's CFO, Mr. Wendell Huang, for the summary of operations and the current quarter guidance.

[Jen-Chau Huang, executive]

Thank you, Jeff. Good afternoon, everyone. Thank you for joining us today.

My presentation will start with financial highlights for the fourth quarter of 2024. After that, I will provide the guidance for the first quarter of 2025.

Fourth quarter revenue increased 14.3% sequentially in NT, supported by strong demand for our industry-leading 3-nanometer and 5-nanometer technologies. Gross margin increased by 1.2 percentage points sequentially to 59%, mainly reflecting a higher capacity utilization rate and productivity gains, partially offset by the dilution of 3-nanometer ramp-up. With operating leverage, total operating expenses accounted for 10% of net revenue. Thus, operating margin increased by 1.5 percentage points sequentially to 49%.

Overall, our fourth quarter EPS was TWD 14.45. And ROE was 36.2%.

Now let's move on to revenue by technology. 3-nanometer process technology contributed 26% of wafer revenue in the fourth quarter. 5-nanometer and 7-nanometer accounted for 34% and 14%, respectively. Advanced technologies, defined as 7-nanometer and below, accounted for 74% of wafer revenue. On a full year basis, 3-nanometer revenue accounted for 18% of 2024 wafer revenue; 5-nanometer, 34%; 7-nanometer, 17%. Advanced technologies accounted for 69% of total wafer revenue, up from 58% in 2023.

Moving on to revenue contribution by platform. HPC increased 19% quarter-over-quarter, to account for 53% of our fourth quarter revenue. Smartphone increased 17%, to account for 35%. IoT decreased 15%, to account for 5%. Automotive increased 6%, to account for 4%. DCE decreased 6%, to account for 1%.

On a full year basis, HPC increased 58% year-on-year. Smartphone, IoT, automotive, DCE increased 23%, 2%, 4% and 2%, respectively, in 2024. Overall, HPC accounted for 51% of our 2024 revenue. Smartphone accounted for 35%. IoT accounted for 6% and automotive accounted for 5%.

Moving on to the balance sheet. We ended the fourth quarter with cash and marketable securities of TWD 2.4 trillion or USD 74 billion. On the liabilities side, current liabilities increased by TWD 184 billion mainly due to the increase of TWD 71 billion in accounts payable and increase of TWD 99 billion in accrued liabilities and others.

In terms of financial ratios, accounts receivable turnover days declined by 1 day to 27 days, while inventory days decreased by 7 days to 80 days primarily due to shipment of N3 and N5 wafers.

Regarding cash flow and CapEx. During the fourth quarter, we generated about TWD 620 billion in cash from operations, spent TWD 362 billion in CapEx and distributed TWD 104 billion for the first quarter '24 cash dividend. Overall, our cash balance increased TWD 241 billion to TWD 2.1 trillion at the end of the quarter.

In U.S. dollar terms, our fourth quarter capital expenditures totaled $11.2 billion.

Now let me recap our performance in 2024. Due to the strong demand for our 3-nanometer and 5-nanometer process technologies, we continued to outperform the foundry industry in 2024. Our revenue increased 30% in U.S. dollar terms to USD 90 billion or increased 33.9% in NT to TWD 2.89 trillion. Gross margin increased 1.7 percentage points to 56.1%, mainly reflecting improvements in overall capacity utilization, partially offset by 3-nanometer dilution and higher electricity costs. With operating leverage, our operating margin increased 3.1 percentage points to 45.7%.

Overall, full year EPS increased 39.9% to TWD 45.25. And ROE increased 4.1 percentage point to 30.3%.

On cash flow, we spent USD 29.8 billion or TWD 956 billion in CapEx, generated TWD 1.8 trillion in operating cash flow and TWD 870 billion in free cash flow. We paid TWD 363 billion in cash dividends in 2024, up 24.5% year-over-year.

I've finished my financial summary. Now let's turn to our current quarter guidance. We expect our business in the first quarter to be impacted by smartphone seasonality, partially offset by continued growth in AI-related demand. Based on the current business outlook, we expect our first quarter revenue to be between USD 25 billion and USD 25.8 billion, which represents a 5.5% sequential decline or a 34.7% year-over-year increase at the midpoint.

Based on the exchange rate assumption of USD 1 to TWD 32.8, gross margin is expected to be between 57% and 59%, operating margin between 46.5% and 48.5%. Regarding tax rate: Our effective tax rate was 16.7% in 2024. For 2025, we expect our effective tax rate to be between 16% and 17%.

This concludes my financial presentation. Now let me turn to our key messages. I will start by talking about our fourth quarter '24 and first quarter '25 profitability.

Compared to third quarter, our fourth quarter gross margin increased by 120 basis points sequentially to 59% primarily due to a higher capacity utilization rate and productivity gains, partially offset by dilution from the continued ramp-up of our 3-nanometer technology. We have just guided our first quarter gross margin to decrease by 100 basis point to 58% at the midpoint. This is primarily due to ramp costs associated with N2 and CoWoS expansion and the start of dilution from our overseas fabs.

As a reminder. 6 factors determine TSMC's profitability: leadership, technology development and ramp-up, pricing, cost reduction, technology mix, capacity utilization and foreign exchange rate. Looking at full year 2025, given the 6 factors, there are a few puts and takes I would like to share. On the one hand, we are working hard to increase our value. The dilution impact from our N3 ramp is expected to gradually reduce. And we expect our overall utilization rate to moderately increase in 2025. On the other hand, as we have said before, we forecast 2% to 3% margin dilution impact from the ramp-up of our overseas fabs. The impact is less than 100 basis point in the first quarter of '25, but we expect it to grow more pronounced throughout the year as our fabs in Kumamoto and Arizona ramp up.

We also expect inflationary costs, including higher electricity prices in Taiwan, to impact our gross margin by at least 1% in 2025. In addition, there are some ramp-up costs associated with N2 and further conversion of N5 to N3 capacity, which together we expect to impact our gross margin by about 1%. Finally, we have no control over the foreign exchange rate, but that may be another factor in 2025. Longer term, excluding the impact of foreign exchange rate and considering our global manufacturing footprint expansion plans, we continue to forecast a long-term gross margin of 53% and higher is achievable.

Next, let me talk about our 2025 capital budget and depreciation. Every year, our CapEx is spent in anticipation of the growth that will follow in the future years. And our CapEx and capacity planning is based on the long-term market demand profile. At TSMC, a higher level of capital expenditures is always correlated with higher-growth opportunities in the following years. In 2024, we spent USD 29.8 billion as we continued to invest to support our customers' growth.

With our strong technology leadership and differentiation, we are well positioned to capture the multiyear structural demand from the industry megatrends of 5G, AI and HPC. In 2025, we expect our capital budget to be between USD 38 billion and USD 42 billion as we invest to capture the future growth. Out of the USD 38 billion to USD 42 billion CapEx for 2025, about 70% of the capital budget will be allocated for advanced process technologies. About 10% to 20% will be spent for specialty technologies. And about 10% to 20% will be spent for advanced packaging, testing, mask making and others.

Our depreciation expense is expected to increase by high single-digit percentage year-over-year in 2025, as newly incurred depreciation will be partially offset by other nodes rolling off depreciation. Even as we invest for the future growth with this level of CapEx spending in 2025, we remain committed to delivering profitable growth to our shareholders. We also remain committed to a sustainable and steadily increased cash dividend per share on both an annual and quarterly basis.

Now let me turn the microphone over to C. C.

[C.C. Wei, executive]

Thank you, Wendell. Good afternoon, everyone.

First, let me start with conclusion of 2024 and our 2025's outlook. 2024 was a mixed year of recovery for the global semiconductor industry. AI-related demand was strong, while other applications saw only a very mild recovery as macroeconomics condition weighed on consumer sentiment and the end market demand. Concluding 2024, the Foundry 2.0 industry, which we define as all logical wafer manufacturing, packaging, testing, mask making and others, increased 6% year-over-year, slightly lower than our previous forecast. Supported by strong demand for our leading-edge process technologies, TSMC's revenue increased 30% year-over-year in U.S. dollar term, outperformed the foundry industry growth.

Entering 2025, we expect fabless semiconductor inventory to have returned to a healthier level exceeding 2024. We forecast Foundry 2.0 industry to grow 10% year-over-year in 2025, supported by robust AI-related demand and a mild recovery in other end market segment. Supported by our technology leadership and broad customer base, we are confident we can continue to outperform the industry growth. We expect 2025 to be another strong growth year for TSMC and forecast our full year revenue to increase by close to mid-20s percent in U.S. dollar term.

Now I will talk about AI demand and TSMC's long-term growth outlook. We observed robust AI-related demand from our customers throughout 2024. Revenue from AI accelerators, which we now define as AI GPU, AI ASICs and HBM controller for AI training and inference in the data center, accounted for close to mid-teens percent of our total revenue in 2024. Even after more than tripling in 2024, we forecast our revenue from AI accelerator to double in 2025 as the strong surge in AI-related demand continues.

As a key enabler of AI applications, the value of our technology platform is increasing as customers rely on TSMC to provide the most advanced process and packaging technologies at scale in the most efficient and cost-effective way. To address the structural increase in the long-term market demand profile, TSMC is working closely with our customer to plan our capacity and investing in leading-edge specialty and advanced packaging technologies to support the growth.

As we have said before, TSMC employs a disciplined and [ rough ] capacity planning system to evaluate and judge market demand to determine the appropriate capacity build. This is especially important when we have such high forecasted demand from AI-related business. At the same time, we are committed to earning sustainable and healthy return that enable us to continue to invest to support our customers' growth while delivering profitable growth for our shareholders.

Underpinned by our technology leadership and broader customer base, we now forecast the revenue growth from AI accelerators to approach a mid-40% CAGR for the 5-year period starting off the already higher base of 2024. We expect AI accelerators to be the strongest driver of our HPC platform growth and the largest contributor in terms of our overall incremental revenue growth in the next several years.

Looking ahead, as the world's most reliable and effective capacity provider, TSMC is playing a critical and integral role in the global semiconductor industry. With our technology leadership, manufacturing excellence and customer trust, we are well positioned to address the growth from the industry megatrend of 5G, AI and HPC with our differentiated technologies. For the 5-year period starting from 2024, we expect our long-term revenue growth to approach a 20% CAGR in U.S. dollar term, fueled by all 4 of our growth platform, which are smartphone, HPC, IoT and automotive.

Next, let me talk about our global manufacturing footprint update. All our oversea decision are based on our customers' need as they value some geographic flexibilities and a necessary level of government support. This is also to maximize the value for our shareholder. In the U.S., we have a long-standing, good relationship with the U.S. government dating back to even before our Arizona fab project announcement in May 2020. We have received a strong commitment and support from the U.S. customers and the U.S. Federal state and city government and are making substantial progress.

Building on the successful result of our earlier engineering wafer production, we were able to pull ahead the production schedule of our first fab in Arizona. Our first fab has already entered a high-volume production in 4Q '24 utilizing N4 process technology, with a yield comparable to our fabs in Taiwan. We expect a smooth ramp-up process. And with our strong manufacturing capability and execution, we are confident to deliver the same level of manufacturing quality and reliability from our fab in Arizona as from our fab in Taiwan.

Our plans for second fab and third fab in Arizona are also on track. These fab will utilize even more advanced technologies such as our N3, N2 and A16, based on our customers' need. There's -- TSMC will continue to play a critical and integral role in enabling our customers' success while remaining a key partner in enabling of the U.S. semiconductor industry.

Next, in Japan, thanks to the strong support from the Japanese central prefecture and local government, our progress is also very good. Our first specialty technology fab in Kumamoto has started volume production at the end of 2024, with very good yield. Construction of our second fab, specialty fab, is scheduled to begin this year. In Europe, we have received strong commitment from the European Commission and the German federal state and city government. We are progressing smoothly with our plans to build a specialty technology fab in Dresden, Germany focusing on automotive and industrial application.

In Taiwan, we continue to receive support from Taiwan government. And we are investing in and expanding our advanced technology and packaging capacities. Given the robust multiyear demand for our 3-nanometer technology, we continue to expand our 3-nanometer capacity in Tainan Science Park. We are also preparing for multiple phases of 2-nanometer fabs in both Hsinchu and Kaohsiung Science Park to support the strong structural demand from our customers. We are also expanding our advanced packaging facilities across several location in Taiwan.

As we have said before, under today's fragmented globalization environment, oversea fab costs are higher for everyone, including TSMC and all other semiconductor manufacturers. We are leveraging our fundamental competitive advantage of manufacturing technology leadership and large-scale manufacturing base to be the most efficient and cost-effective manufacturer in the regions that we operate while supporting our customers' growth.

Finally, I will talk about N2 and A16 introduction. Our 2-nanometer and A16 technologies leads the industry in addressing the insatiable need for energy-efficient computing, and almost all the innovators are working with TSMC. We expect the number of the new tape-out for 2-nanometer technology in the first 2 years to be higher than both 3-nanometer and 5-nanometer in their first 2 years, fueled by both smartphone and HPC applications.

N2 delivered full node performance and power benefit with 10 to 15 speed improvement at the same power or 20% to 30% power improvement at the same speed, and a more than 15% chip density increase, as compared with N3E. N2 is well on track for volume production in second half of 2025 as scheduled, with a ramp profile similar to N3. With our strategy of continuous enhancement, we also introduced N2P as an extension of N2 family. N2P features further performance and power benefits on top of N2. N2P will support both smartphone and HPC applications. And volume production is scheduled for second half 2026.

We will also introduce A16 featuring Super Power Rail or SPR as separate offering. TSMC's SPR is a innovative, best-in-class backside power delivery solution that is first in the industry to incorporate a novel backside metal scheme that preserves gate density and device width flexibility to maximize product benefit. Compared with N2P, A16 provide a further 8% to 10% speed improvement at the same power or 15% to 20% power improvement at the same speed, and additional 7% to 10% chip density gain. A16 is the best suitable for specific HPC product with a complex signal route and dense power delivery network. Volume production is scheduled for second half 2026.

We believe N2, N2P, A16 and its derivative will further extend our technology leadership position and enable TSMC to capture the growth opportunity well into the future.

These conclude our key message, and thank you for your attention.

[Jeff Su, executive]

Thank you, C. C.

This concludes our prepared statements. [Operator Instructions] Questions will be taken from both the floor and from the call. Should you wish to raise your question in Chinese, I will translate it to English before our management answers your question. [Operator Instructions] Now we will begin the Q&A session. We'll take the first few questions here from the floor and then go to online. I think maybe left, middle, right.

[Jeff Su, executive]

So why don't we start? I think, first question, Gokul Hariharan from JPMorgan.

[Gokul Hariharan, analyst]

My first question is on the TSMC's U.S. future strategy. There have been a lot of changes recently. Taiwan relaxed the n-1 restriction. There was a news about that a week back. C. C., you met Elon Musk as well recently. So you said there are a lot of developments that you've discussed. Your key IDM competitor seems to be struggling as well, while your Arizona fab seems to be ramping up quite well, so in light of all these, I just wanted to understand the longer-term strategy. Would you consider investing in latest node in the U.S.? Because so far, it has been n-1. Now you don't have the restriction from the Taiwan government to go and invest in the latest node. What has been your feedback in whatever discussions you have had with the incoming President Trump administration? Because they have talked a lot about CHIPS Act and everything, but they're also supportive. Your original investment was during President Trump's first term. And lastly, I think -- Wendell, I think, last time you had mentioned you're not very keen on taking over any IDM fabs. Has that thinking changed, especially given TSMC has a potential to become a even more stronger partner for the U.S. in terms of bringing up U.S. local manufacturing? Sorry, long question.

[Jeff Su, executive]

Yes. Okay, thank you, Gokul. Indeed a very long question. I think Gokul's question is looking at TSMC and our strategies in terms of global expansion, particularly in the U.S. He notes that Taiwan has recently relaxed or said they relaxed the n-1 rule. And C. C. has met several, many of our large customers in the U.S. And our Arizona fab is ramping quite well, so his question really is on the longer-term strategy, I believe, 3 parts. Number one, what is the feedback or sort of discussions ongoing with the next administration in the U.S.? Secondly, would we consider taking over IDMs' fabs? Has that thinking changed? And last, on the new node. Maybe we'll go one by one.

[C.C. Wei, executive]

I almost forgot your question already. Okay, first one, the technology node. Actually it's not we don't want to ramp up the same technology as in Taiwan, but if you look at the -- when we're ramping up, introduce a new technology into manufacturing, the fab is -- the process is so complicated, so it should be very close to the R&D people. So the initial phase of the ramping up always come from the fab close to R&D, so in that sense, we want to ramp up the same kind of technology in U.S., but that practically is a little bit difficult. So Taiwan will always be first. Did that answer your question? It's not because of, like, n-1's limitation, no. It's practically we just have to ramp up a new node in Taiwan, [ yes ].

And the second, do we change our strategy to expand faster or something? Again this is -- we always say that we build the capacity oversea is due to customer's need. If my customer has a very high demand, what should I do? I build more fab, right, with necessary government support, by the way. Okay, talking about government, let me assure you that we have a very frank and open communication with the current government and with the future one also. I cannot say anything more than that. Okay, what is...

[Gokul Hariharan, analyst]

IDM fab...

[C.C. Wei, executive]

That's my customer. And we -- again, our strategy is not based on my IDM competitors' status. They are our very good customers. I like them. And they are very important to TSMC's business also. That all I can say. Thank you.

[Gokul Hariharan, analyst]

Maybe my next question, going to gross margins. So Wendell, we are almost approaching 60% gross margin. Last cycle, we peaked at about 60% towards the peak of the cycle. You are expecting the cycle to even strengthen, based on guidance that C. C. provided for both AI as well as some improvement in non-AI, so how should we think about gross margins in this cycle? Is it realistic that we can get to more than 60% gross margin in this up cycle?

And related to that, could you help us understand the U.S. -- or especially the U.S. fab, overseas fabs but especially U.S. fab, dilution. What are the key factors there? Because as you mentioned, yield is already approaching or almost close to Taiwan yield, so is it basically cycle time is longer? Or is it that some other costs are much higher in the U.S. fab? Because new fab depreciation is probably fairly similar compared to the Taiwan fab.

[Jeff Su, executive]

Yes, okay. So Gokul's second question is on gross margin and again 2 parts. He notes gross margin is almost approaching 60%. In 2022, the last cycle, it was also around this type of level. We have said that this year is another very strong growth year for TSMC, so his question is how should we think about gross margins in this current cycle. Can we approach or get to 60% or low 60s type of again? And then the second part is more specific to the U.S. in terms of the cost gap. What are the U.S. cost factors leading to the dilution impact?

[Jen-Chau Huang, executive]

Okay, Gokul, first question, on the gross margin. As we said, there are 6 factors affecting the profitability. Every year, different factors play different roles, but for example, if the utilization is extremely high like last cycle, it is not impossible for us to reach what you just said, okay? And secondly, the U.S. fab cost, it is more expensive in the U.S. mainly because of several reasons: number one, the smaller scale, right; now number two, the higher price in the supply chain; and number three, the very early stage of the ecosystem. So if you add all these up, as we said, 2% to 3% dilution from our overseas fabs every year in the next 5 years.

[Gokul Hariharan, analyst]

If I use the 2% to 3% and do some math, it feels like the overseas fab is starting at, I don't know, 10% gross margin or 5% gross margin, just adding up factors. Obviously it's not how it works, but I'm just doing outside-in. Is that right? Is that the right kind of ballpark in terms of thinking about margin?

[Jen-Chau Huang, executive]

All we can share is the 2% to 3%, yes.

[Gokul Hariharan, analyst]

I don't think TSMC has ever started a fab at 10% gross margin.

[C.C. Wei, executive]

Gokul, we are working hard to improve it.

[Jeff Su, executive]

Okay, thank you, Gokul. We'll go to the middle, Laura Chen from Citigroup, Citi Bank.

[Chia Yi Chen, analyst]

Congratulation for the good result. I just want to have more details about your [ review ]. I mean I think people are kind of looking for your updated long-term CAGR growth. So I believe that 20% starting from a very -- already very high base in 2024 is a really good long-term objective but just wondering that, aside from the strong AI demand, what's your view on the traditionals, applications like PC and the smartphone, growth and particularly for this year.

[Jeff Su, executive]

Okay, so Laura's first question is looking -- she notes that we have updated our long-term CAGR to be approaching 20% revenue growth in U.S. dollars, starting off even the high base of '24, so her question is, of course, AI demand is part of that, but what about smartphone and PC. And I think your question is specific to this year, C. C., 2025.

[C.C. Wei, executive]

This year is still mild growth for PC and smartphone, but everything is AI related, all right, so you can start to see why we have confidence to give you a close to 20% CAGR in the next 5 years. AI: You look at a smartphone. They will put AI functionality inside, and not only that. So the silicon content will be increased. In addition to that, actually the replacement cycle will be shortened. And also they need to go into the very advanced technology because of, if you want to put a lot of functionality inside a small chip, you need a much more advanced technology to put those [indiscernible]. Put all together, that even smartphone, the unit growth is almost low single digit, but then the silicon and the replacement cycle and the technology migration, that give us more growth than just unit growth; similar reason for PC.

[Chia Yi Chen, analyst]

So we can kind of expect, those AI-capable edge devices, they will all base on 2-nanometer next year, perhaps second half.

[C.C. Wei, executive]

Leading-edge technology, that's all I say.

[Chia Yi Chen, analyst]

And also my next question is about AI. I noted that, these times, you included the HBM controller into your AI business revenues definition. So can you provide us more update about what the HBI-based die business opportunities could be? And previously, TSMC kind of announced a cooperation with key memories suppliers globally. Can you give us more details or updates on the progress of this business engagement?

[Jeff Su, executive]

Okay, thank you, Laura. So Laura's second question is on HBM controllers. She notes that our definition of AI accelerators includes memory controllers or HBM controllers, so her question is, well, how do we see this opportunity. Or what is the opportunity for TSMC? And what is the progress of this working with our memory partners?

[C.C. Wei, executive]

We are working with all the memory suppliers, all of them. And that is because of TSMC's logic chip or logical technology more advanced and that meet our customers' requirement. So all of them are working with TSMC. Now we start to see some of the product coming out, but the high volume, probably you will need to wait for another half or 1 year to see the high volume and big contribution to TSMC's revenue.

[Jeff Su, executive]

We'll move to this side of the room. I guess we have Charlie Chan from Morgan Stanley.

[Charlie Chan, analyst]

I think it's going to be a very exciting year given your bullish outlook and also lots of news going on, right? So let me start with that, overnight, the U.S. seems to put a new framework on restricting China's AI business, right? So I'm wondering whether that will create some business impact to your China business. And how are you going to manage that? And also, for some chips in the middle, high performers like crypto mining, autonomous driving chip, do you think these count as cloud AI? And would TSMC be able to continue to service your China customers?

[Jeff Su, executive]

Okay, thank you, Charlie. So Charlie's first question, if I may sort of extrapolate or summarize, is about the announcements of different types of U.S. export restrictions this week pertaining to China and AI-related chips. So his question is what is the impact to TSMC. How does it impact our business?

[C.C. Wei, executive]

So far, we look at -- we don't have all analysis yet, but the first look is not significantly. It's manageable. So that meaning that, my customers who are being restricted [ or something ], we are applying for the special permit for them. And we believe that we have confidence that they will get some permission, so long as they are not in the AI area, okay, especially automotive industry. Or even you talk about crypto mining, yes.

[Charlie Chan, analyst]

That is super helpful. My second question is actually a very hot topic recently as well, the CPO. I think, your main partner, Jensen, came to Taiwan, this time probably -- besides meeting you, right, probably also he want to enable this supply chain. So based on your recent technology symposium, right, you already get ready for your COUPE optical engine, but do you think the Taiwan supply chain can really facilitate this CPO? Because with all these key components, the next-generation Rubin schedule could have some issue. So I think this is part of -- first part of the question, about how you are going to facilitate this CPO supply chain.

And secondly, to TSMC, your foundry service, right? Do you see significant upside when an [ optical networking ] migrate to CPO? Because -- I ask these because there are -- some conventional product, like a optical transceiver DSP, could be replaced.

[Jeff Su, executive]

Well, Charlie's second question is a very specific topic. He wants to know -- well, if I can generalize because we certainly don't comment on customers or their products. But in terms of our progress on silicon photonics and CPO, how are we working with customers? How we're preparing as part of our advanced packaging solutions. And what are the opportunities for TSMC as optical moves to silicon photonics and other type of solutions, on a general basis?

[C.C. Wei, executive]

Charlie, that a very technical question. Silicon photonics, we are working on it, as you said. And we've got a good result also. However, a big volume, I don't think it will be in this year. Or probably we have to wait for 1 or 1.5 years to you can see that -- the contribution [ of ] the volume production. The initial results are quite good, no doubt about it, and so my customer are quite happy.

[Jeff Su, executive]

Okay, thank you, C. C. Operator, we'll now move to the questions online. We'll take the first call from the online participant, please.

[Operator]

Yes. First question, Brett Simpson, Arete Research.

[Brett Simpson, analyst]

Can I just say, congratulations on reaching 100 billion in annual sales in Q4. It's quite a milestone. So my first question is on Arizona. I think, Wendell, you mentioned that we need to see some higher scale, so can you update us on the status of Phase 2? It looks like the construction of the shell is nearly complete, but it would be great to understand more about how you see P2 developing over the course of 2025. And in terms of pricing U.S. wafers, how are you planning to do this? Will you have a U.S. price and a Taiwan price? Or are you more likely to have a global price regardless of where you make the wafers?

[Jeff Su, executive]

Okay, so Brett's question, first question, is on Arizona, maybe split into 2 parts. First is in terms of we have already started the volume production of the first fab. So Brett would like an update on the progress of the second fab in terms of the construction of the buildings and the shells, et cetera. And then the second part would be on the pricing of overseas. As we say, there's value to our customers. He wants to know, do we charge a separate price? Or is it part of the overall pricing, et cetera, et cetera?

[C.C. Wei, executive]

Let me answer the second question first. Do we charge a little bit higher? Yes, we did because of we have a value of geographics of -- flexibility, right? And you guy knows [ made in U.S.A. ] is a premium product. Yes, we discuss with our customer. And they are -- all agree and happy to work with TSMC so that we can -- [ maybe ] because of the cost structure over there. So it's a little bit higher price over there.

The progress of the first fab is right now in volume production. Second fab, we've almost finished all the building; and start to put the facility, et cetera, et cetera. Then we expect that we move the tools this year also. We have a plan that our third fab probably will start very soon; and we will announce it in later days, okay?

[Jeff Su, executive]

Okay, thank you, C. C. Brett, does that answer your first question? And do you have a second one?

[Brett Simpson, analyst]

Yes, very clear. And the second question: I wanted to get your perspective. Broadcom's CEO recently laid out a large SAM for AI hyperscalers building out custom silicon. I think he was talking about million clusters from each of the customers he has in the next 2 or 3 years. What's TSMC's perspective on all this? I'm sure you've spent a lot of time verifying what hyperscalers are planning over the years to come. And how comfortable are you with the scale of what's being implied here?

[Jeff Su, executive]

Okay, so Brett's second question is looking at AI, I guess, specifically AI custom chips or ASICs. He notes that one of our customers recently laid out a very strong or large addressable SAM, market for AI hyperscalers using custom silicon; lots of them talking about clusters of 1 million chips. So he wants to know. What is TSMC's view? How do we see this trend in terms of AI ASICs as part of the AI demand megatrend?

[C.C. Wei, executive]

Brett, I'm not going to answer the question of the specific number, but let me assure you that, whether it's ASIC or it's graphic, they all need a very leading-edge technology. And they're all working with TSMC, okay, so -- and the second one is, is the demand real. Was -- is -- as a number that my customers said. I will say that the demand is very strong. Is that enough to answer your question, Brett?

[Brett Simpson, analyst]

Yes. That's great.

[Jeff Su, executive]

Operator, do we have anyone else on the line? It seems not. Then let's...

[Operator]

No, we don't.

[Jeff Su, executive]

Okay, we don't. Then let's go back to the floor. I think, in -- on the left side, Bruce Lu from Goldman Sachs.

[Zheng Lu, analyst]

To be honest: I'm a bit surprised that the long-term gross margin target doesn't really change yet. I believe TSMC's value is definitely more than selling the -- pass on the cost. I believe that TSMC need to invest a lot more in R&D to maintain their leadership. TSMC suggested -- raised the gross margin target in 2022 with higher R&D requirement, with higher profitable target, right? So I asked the same question 2 quarters ago -- which is in the process of pricing negotiation, which is understandable, but I think the price negotiation is pretty much done. What's the discrepancy here? Why TSMC cannot raise the profitability target.

[Jeff Su, executive]

Okay, so Bruce's first question, he wants to know again, our long-term gross margin. Why are we not changing the target of 53% and higher, okay? He correctly notes that certainly TSMC's value is increasing. And certainly TSMC would need to invest a lot of money in R&D and capacity to support our customers' growth. So we have always had a focus on earning the right return. He also notes, in 2022 -- well, our gross margin used to be about 50%. Then we raised it to 53% and higher, so his question is why is it not "and higher," I guess.

[Jen-Chau Huang, executive]

All right, Bruce, as we said, 6 factors affecting the profitability. Every year, different factors have different weight. Now 2 things to note. Number one, starting from this year, overseas fab expansion, 2 to 3 percentage point impact every year for the next 5 years. The other things to note: macro environment uncertainty which may lead to impacting the global economy, which may lead to end-market demand. Now having said that, we are in a capital-intensive industry, so we'll need to have, to earn a healthy return to continue to invest to support our customer, support their growth and also deliver a profitable growth to our shareholders. And you mentioned about the raising of long-term gross margins back in 2022 to 53% and higher, and we have been able to deliver that "and higher" part since then. So given all the above, we continue to think that 53% and higher gross margin is achievable. And we work very hard to achieve on the "higher" part.

[Zheng Lu, analyst]

Okay, I'll try next 2 quarters. For the CoWoS capacity, TSMC has been very aggressive in expanding the capacity. However, the application is highly concentrated in AI at the current stage, which there are certain noise around it. When can we see non-AI application such as server, smartphone or anything else can be -- can start to adopt CoWoS capacity in case there is any fluctuation in the AI demand?

[Jeff Su, executive]

Okay, thank you, Bruce. So Bruce's second question is on CoWoS capacity. In his words, we have been very aggressive to expand the capacity, but his concern is highly concentrated with AI-related demand. So his question is, when do we expect or -- to see more non-AI application adoption of CoWoS solutions?

[C.C. Wei, executive]

Well, yes, today is all AI focused. And we have a very tight capacity and cannot even meet customers' need, but whether other products will adopt this kind of CoWoS approach, they will. It's coming and we know that it's coming. So that's all I can say.

[Zheng Lu, analyst]

When?

[C.C. Wei, executive]

It's coming.

[Zheng Lu, analyst]

Okay, I will try next quarter.

[C.C. Wei, executive]

On the CPU and on the server chip. Let me give you a hint.

[Jeff Su, executive]

We'll go to the middle, Arthur Lai from Macquarie.

[Yu Jang Lai, analyst]

C. C. and Wendell and Jeff, Arthur Lai from Macquarie. So first of all, congrats on the strong gross margin. I just have a very quick follow-up on the U.S. and JP expansion, as this is important. My client keep chasing me. So do you have a operational strategy to mitigate the cost gap between the oversea fab and Taiwan fab? Yes. I think, C. C., you hint that you work on it and improve the gross margin, but during Chinese New Year, I read Morris Chang's -- his autobiography. And he mentioned that -- on the strategy is -- it's actually from the Taiwan mother fab, right, so I want to understand how we maintain the high yield and also drive the costs down.

[Jeff Su, executive]

Okay, so Arthur's question is about our overseas expansion. His question is related to the cost gap and what is our operational strategies to mitigate the cost gap. How are we doing this internally in our fab operations and strategies to do so?

[C.C. Wei, executive]

[ Well, you mention ] my boss's book, okay. That's meaning that you really read it. What he say is a copy exactly is, whatever Taiwan's improvement, the U.S. will copy over there. It doesn't mean that this year, next year and the following year will be the same. We continue to improve. That's improve the cost structure both in Taiwan and in the U.S. And we also try very hard to find out a new methodology or whatever that I cannot share with you right now, but it will give Arizona fab some benefit. And so we will improve -- we will minimize the gap between the -- cost structure between U.S. and Taiwan. And we are working on that, but no matter what I say, we will be the best fab over there, okay?

[Yu Jang Lai, analyst]

And second, follow-up question probably is on Wendell. You just mentioned that there's a 200 bps or 300 bps margin dilution, right? So can you give us [ a one-level down cycle ], the variable costs and the fixed costs, maybe [ half them ] or maybe -- which one is higher?

[Jeff Su, executive]

Okay, so Arthur's second question is on the overseas dilution of 2% to 3%. He is asking if we can provide a further breakdown in terms of how much of that is composed from variable costs. How much of that is from the fixed costs, et cetera?

[Jen-Chau Huang, executive]

Arthur, we really don't give breakdown on these numbers, but both of them are higher. That's all I can share with you.

[Jeff Su, executive]

Okay, thank you. We'll move to the right side of the room. I think Rick Hsu from Daiwa Securities.

[Rick Hsu, analyst]

Yes. So the first one, C. C., can you share with us your view on this year's global semiconductor revenue forecast, ex memory; or any driver by applications in priority across the main application?

[Jeff Su, executive]

Okay, so Rick's first question, he is asking for our forecast of the semiconductor industry, what we used to provide as semi ex mem, but of course, we have already given Foundry 2.0. Then he would like the outlook by end market application, in terms of ranking; maybe just a comment on the overall end markets as a whole, yes.

[C.C. Wei, executive]

Rick, I think the memory business will grow this year also, but all I can say is that HBM will grow very fast. And I don't comment on other memories because of it's not logic.

[Jeff Su, executive]

And we have already provided, "Foundry 2.0 to grow 10% year-over-year." That's our industry forecast for 2025.

[Rick Hsu, analyst]

Just a quick follow-up. Can I use your Foundry 2.0 market growth as a proxy of the global semi ex memory?

[Jeff Su, executive]

So his question is can we use Foundry 2.0 as a proxy for semiconductor ex memory.

[Jen-Chau Huang, executive]

Yes.

[Jeff Su, executive]

Okay.

[Rick Hsu, analyst]

The second one is very quick, about your CoWoS and SoIC capacity ramp. Can you give us more color this year? Because recently there seemed to be a lot of market noises. Some add orders. Some cut orders, so I would like to see your view on the CoWoS ramp.

[Jeff Su, executive]

Okay, so Rick's second question is lots of market rumors here. So he would like to know any comment we can provide on CoWoS ramp in 2025.

[C.C. Wei, executive]

Rick, as you said, there's a lot of rumor. That's a rumor. I assure you. We are working very hard to meet the requirement of my customers' demand, so "cut the order," that won't happen. We actually continue to increase, so we are -- again I will say that. We are working very hard to increase the capacity.

[Jeff Su, executive]

Yes. Okay, let's move back to -- operator, is there anyone online?

[Operator]

Yes. We have next one, Robert Sanders, Deutsche Bank.

[Robert Sanders, analyst]

I just had a question on AI demand. Is there a scenario where HBM is more of a constraint on the demand, rather than CoWoS which seems to be the biggest constraint at the moment? And I have a follow-up.

[Jeff Su, executive]

Okay, so Rob is asking us to comment on AI demand and HBM status constraint? Or what is the bigger constraint in AI demand?

[C.C. Wei, executive]

I don't comment on other supplier, but I know that we have a very tight capacity to support the AI demand. I don't want to say I'm the bottleneck. TSMC, always working very hard with customer to meet their requirement. That's all I can say.

[Jeff Su, executive]

Do you have a second question?

[Robert Sanders, analyst]

Yes, just on SoIC. There have been more discussion in the market around your smartphone customers adopting SoIC. Can you just discuss if there's any kind of inflection point here, whether it's in the PC domain or the smartphone domain? Or is this still more of a data center story?

[Jeff Su, executive]

Okay. Well, Rob's second question is on SoIC adoption. His question basically, in a nutshell, is when do we see an inflection point for smartphone application to adopt SoIC.

[C.C. Wei, executive]

Today, SoIC's demand is still focused on AI applications, okay? For PC or for other area, it's coming but not right now.

[Jeff Su, executive]

Okay, thank you, Rob. Thank you, C. C. I think, in the interest of the time, we'll take the last 2 questions, please. Okay, I guess we will go to Sunny Lin from UBS.

[Sunny Lin, analyst]

And so my first question is to try to get a bit more clarity on the cloud growth for 2025. I think, longer term, without a doubt, the technology definitely has lots of potential for demand opportunities, but I think -- if we look at 2025 and 2026, I think there could be increasing uncertainties coming from maybe [indiscernible] spending, macro or even some of the supply chain challenges. And so I understand the management just provided a pretty good guidance for this year for sales to double. And so if you look at that number, do you think there is still more upside than downside as we go through 2025? Or how should we think about the demand profile for this year and next year?

[Jeff Su, executive]

Okay, well, Sunny's question is about the AI-related demand. We have said that, even after tripling, more than tripling, last year, it will double again in 2025. She wants to know, is there upside or downside to this? And also for us to provide an outlook on the 2026 AI growth.

[C.C. Wei, executive]

Sunny, I certainly hope there is upside, but I hope I get -- my team can supply enough capacity to support it. Does that give you enough hint? Okay. And we also forecast based on the 2024's high number. We also forecast of mid-40s is CAGR for the 5 years. That give you some kind of estimate that you can calculate.

[Sunny Lin, analyst]

Yes. Well, so mid-40% is the long-term expectation in terms of growth by next few years, but how should we think about the trajectory of the growth? For sure, this year, it's still pretty strong growth, but do you think, at some point, maybe we see a moderation of growth temporarily and then followed by another ramp?

[Jeff Su, executive]

Well, I think Sunny's question again is asking us to comment on 2026 outlook, which is a little bit early. Or is that -- how do we see the trajectory of the growth?

[C.C. Wei, executive]

Jeff already say it's a little bit too early.

[Jeff Su, executive]

All right...

[Sunny Lin, analyst]

Sure, no problem. So I will follow up maybe next quarter as well. And so my second question is on edge AI. And so last year, management did expect maybe 2025 to be the inflection point for -- to see more content related to edge AI, so based on your current visibility, are you seeing clients ramping for this year for the edge AI products, maybe into second half? And before, you also mentioned edge AI could potentially drive 5% to 10% die size increase. Will that be a onetime increase? Or do you think, beyond the 5% to 10% increase for the maybe first-gen product, there should be sustainable increase going forward?

[Jeff Su, executive]

Okay, so Sunny's second question is related to edge AI. She would like some more detail or color. Do we see customers ramping edge or what we call on-device AI products in the second half of this year? And the second part, in terms of the content increase, 5% to 10% increase: Is this a onetime thing? Is this an ongoing thing? How do we estimate the content benefit from on-device AI?

[C.C. Wei, executive]

Okay. On the edge AI, in our observation, we found out that our customers start to put up more neural processing inside. And so we estimated the 5% to 10% more silicon being used. [ Can it be ] every year 5% to 10%? Definitely it is no, right? So they will move to next node, the technology migration. That's also to TSMC's advantage. Not only that, I also say that, the replacement cycle, I think it will be shortened because of, when you have a new toy that -- with AI functionality inside it, everybody want replacing, replace their smartphone, replace their PCs. And [ I count that one ] much more than the -- a mere 5% increase. All right, did I answer your question?

[Sunny Lin, analyst]

Yes.

[Jeff Su, executive]

Operator, I think there's one more participant online, so we'll take the last question from online participant, please.

[Operator]

Apologies. I think the last caller just dropped the line.

[Jeff Su, executive]

Okay, then we'll take the last question from Brad Lin from Bank of America.

[Brad Lin, analyst]

So I would like to ask 2 questions. First question will be on the CoWoS as well. So we have observed an increasing margin of advanced packaging. Could you remind us the CoWoS contribution of last year? And do you expect the margin to kind of approach the corporate average or even exceed it after the so-called -- the value reflection this year? That would be my first question.

[Jeff Su, executive]

Okay, so Brad's first question is very specific to CoWoS. Basically he wants to know what is the revenue contribution from CoWoS last year. And what is the margin profile? Maybe we can talk about advanced packaging.

[Jen-Chau Huang, executive]

Brad, we don't break it down in different segments of the advanced packaging, but overall speaking, advanced packaging accounted for over 8% of revenue last year. And it will account for over 10% this year. In terms of gross margins, it is better. It is better than before but still below the corporate average. Thank you.

[Brad Lin, analyst]

That's very helpful. And then my second question will be on the IDMs. So we have seen IDMs increasingly rely on, well, TSMC. And then do -- we still expect the IDM to support our long-term growth...

[Jeff Su, executive]

Okay, so Brad's second question, I think, is on IDM and IDM outsourcing. He does note that we do see more IDM outsourcing business. So is this part of our long-term growth outlook CAGR?

[C.C. Wei, executive]

Again let me repeat again. They are our very good customers, and we work together. I don't say they rely on TSMC. We are partners. And I really hope that -- a long-term relationship will be there, for sure.

[Jeff Su, executive]

Okay, thank you, C. C. Thank you, Brad. Thank you, everyone. This concludes our Q&A session.

Before we conclude today's conference, please be advised that the replay of the conference will be accessible within 30 minutes from now. The transcript will become available 24 hours from now. And certainly both will be available through TSMC's website at www.tsmc.com.

So thank you, everyone, for joining today online and in person. We'd like to wish everyone a happy new year and hope everyone continues to stay well and hope you'll join us again next quarter. Goodbye and thank you. Have a good day.