Taiwan Semiconductor Manufacturing Co Ltd Q3 FY2023 Earnings Call

This is Jeff Su, TSMC's Director of Investor Relations and your host for today. TSMC is hosting our earnings conference call via live audio webcast through the Company's website at www.tsmc.com, where you can also download the earnings release materials. If you are joining us through the conference call, your dial-in lines are in listen-only mode. The format for today's event will be as follows. First, TSMC's Vice President and CFO, Mr. Wendell Huang, will summarize our operations in the third quarter 2023, followed by our guidance for the fourth quarter 2023. Afterwards, Mr. Huang and TSMC CEO, Dr. C. C. Wei, will jointly provide the Company's key messages. Then, we will open the line for the Q&A. 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 call over to TSMC's CFO, Mr. Wendell Huang, for the summary of operations and the current quarter guidance.

Thank you, Jeff. Good afternoon, everyone. Thank you for joining us today. My presentation will start with financial highlights for the third quarter 2023. After that, I will provide the guidance for the fourth quarter 2023. Third quarter revenue increased 13.7% sequentially in NT dollars or 10.2% in U.S. dollars as our third quarter business was supported by the strong ramp of our industry-leading 3-nanometer technology and higher demand for 5-nanometer technologies, partially offset by customers' ongoing inventory adjustment. Gross margin increased 0.2 percentage points sequentially to 54.3%, mainly reflecting higher capacity utilization, partially offset by the margin dilution from N3 ramp. Total operating expenses accounted for 12.6% of net revenue as compared to 12.1% in the second quarter, mainly due to higher R&D expenses to support our 3-nanometer and 2-nanometer development. Operating margin was 41.7%, down 0.3 percentage points from the previous quarter. Overall, our third quarter EPS was TWD8.14, and ROE was 25.8%. Now let's move on to the revenue by technology. 3-nanometer process technology contributed 6% of wafer revenue in the third quarter while 5-nanometer and 7-nanometer accounted for 37% and 16%, respectively. Advanced technologies, defined as 7-nanometer and below, accounted for 59% of wafer revenue. Moving on to revenue contribution by platform. HPC increased 6% quarter-over-quarter to account for 42% of our third quarter revenue. Smartphone increased 33% to account for 39%. IoT increased 24% to account for 9%. Automotive decreased 24% to account for 5%. And DCE decreased 1% to account for 2%. Moving on to the balance sheet. We ended the third quarter with cash and marketable securities of TWD1.55 trillion or US$48 billion. On the liability side, current liabilities increased by TWD159 billion, mainly due to the increase of TWD95 billion in accounts payable and the increase of TWD59 billion in accrued liabilities and others. Long-term interest-bearing debt increased by TWD30 billion, of which TWD10 billion from new issuance and TWD20 billion from foreign exchange rate movement. On financial ratio, accounts receivable turnover days increased 3 days to 35 days while days of inventory decreased 3 days to 96 days. Regarding cash flow and CapEx. During the third quarter, we generated about TWD295 billion in cash from operations, spent TWD227 billion in CapEx and distributed TWD71 billion for fourth quarter '22 cash dividend. Overall, our cash balance increased TWD35 billion to TWD1.31 trillion at the end of the quarter. In U.S. dollar terms, our third quarter capital expenditures totaled TWD7.1 billion. I have finished my financial summary. Now let's turn to our current quarter guidance. Based on the current business outlook, we expect our fourth quarter revenue to be between TWD18.8 billion and US$19.6 billion, which represents an 11.1% sequential increase at the midpoint. Based on the exchange rate assumption of US$1 to TWD32, gross margin is expected to be between 51.5% and 53.5%, operating margin between 39.5% and 41.5%. This concludes the financial presentation. Now let me turn to our key messages. I will start by making some comments on our third quarter '23 and fourth quarter '23 profitability. Compared to second quarter, our third quarter gross margin increased by 20 basis points sequentially to 54.3%, primarily due to a higher capacity utilization rate and a more favorable foreign exchange rate, partially offset by the margin dilution from the initial ramp-up of our 3-nanometer technology. Compared to our third quarter guidance, our actual gross margin exceeded the high end of the range provided three months ago by 80 basis points, mainly due to a more favorable foreign exchange rate. We have just guided our fourth quarter gross margin to decline by 1.8 percentage points to 52.5% at the midpoint, primarily due to the continued margin dilution from the steep ramp of our 3-nanometer technology. As a reminder, six factors determine TSMC's profitability: leadership, technology development and ramp-up pricing, cost reduction, capacity utilization, technology mix and foreign exchange rate. To manage our profitability in the next several years, we will work diligently on our internal cost improvement while continuing to strategically sell our value. Excluding the impact of foreign exchange rate, of which we have no control over, we continue to forecast a long-term gross margin of 53% and higher is achievable. Next, let me talk about our 2023 CapEx and depreciation. Every year, our CapEx is spent in anticipation of the growth that will follow in future years. Given the near-term uncertainties, we continue to manage our business prudently and have tightened our capital spending throughout the year where appropriate. We now expect our 2023 CapEx to be approximately US$32 billion. Out of the approximately US$32 billion CapEx for 2023, about 70% of the capital budget will be allocated for advanced process technologies, about 20% will be spent for specialty technologies, and about 10% will be spent for advanced packaging, testing, mask-making, and others. Our depreciation expense is now expected to increase by low 20s percentage year-over-year in 2023, as compared to our January forecast of approximately 30% year-over-year increase. Despite the near-term inventory cycle, our commitment to support customers' growth remains unchanged, and our disciplined CapEx and capacity planning remain based on the long-term structure market demand profile. We will continue to work closely with our customers to plan our long-term capacity and invest in leading-edge specialty and advanced packaging technologies to support their growth while delivering profitable growth to our shareholders. Now let me turn the microphone over to C. C.

Thank you, Wendell. Good afternoon, everyone. First, let me start with our near-term demand and inventory. We concluded our third quarter with revenue of US$17.3 billion, in line with our guidance in U.S. dollar terms. Our business in the third quarter was supported by the strong ramp of our industry-leading 3-nanometer technology and higher demand for 5-nanometer technologies, partially offset by customers' ongoing inventory adjustment. Moving into the first quarter 2023, while AI-related demand continues to be strong, it is not enough to offset the overall cyclicality of our business. We expect our business in the fourth quarter to be supported by the continuing ramp of our 3-nanometer technology, partially offset by customers' continued inventory adjustment. On the inventory side, we expect the fabless semiconductor inventory to have continuously reduced in the third quarter. However, due to the persistent weaker overall macroeconomic conditions and slow demand recovery in China, customers remain cautious in their inventory control. Thus, we expect the inventory digestion to continue in the fourth quarter. Having said that, we are observing some early signs of demand stabilization in the PC and smartphone end markets. Together with such level of inventory control, we forecast the fabless semiconductor inventory to further reduce and exit Q4 '23 at a healthier level. Next, let me talk about our global manufacturing footprint update. TSMC's mission is to be the trusted technology and capacity provider of the global logic IC industry for years to come. As we have said before, our strategy is to expand our global manufacturing footprint to increase customer trust, expand our future growth potential, and reach for more global talents. Our overseas decisions are based on our customers' needs and the necessary level of government support. This is to maximize the value for our shareholders. In Europe, after conducting extensive due diligence, we announced our plan to build a specialty technology fab in Dresden, Germany, focusing on automotive and industrial applications. We have received a strong commitment to support this project from our JV partners, the European Commission government, and German federal, state, and city governments. These fabs will utilize 22- and 28-nanometer and 12-, 16-nanometer technologies for semiconductor wafer fabrication. Fab construction is scheduled to begin in the second half of 2024, and production is targeted to begin in late 2027. In Arizona, we are receiving strong support from the city of Phoenix, State of Arizona, and U.S. federal government and continue to develop positive relationships and work closely with our local trade and union partners. We are making good progress on the fab infrastructure, utilities, and equipment installation issues in our first fab, and the situation is improving. We have also begun early preparation for our Arizona fab operations and hired close to 1,100 local TSMC employees so far. Many of them have been brought to Taiwan for extensive hands-on experience in our fab, so that they can further their technical fields, who are being immersed in TSMC operational environment and culture. We continue to target volume production of N4 process technology in the first half of 2025 and are confident that once we begin operations, we will be able to deliver the same level of manufacturing, quality, and reliability in Arizona as from our fabs in Taiwan. In Japan, we built a specialty technology fab, which will utilize 12- and 16-nanometer and 22- and 28-nanometer process technologies. We have hired approximately 800 local TSMC employees so far, with the majority having been brought to Taiwan for hands-on experience. Equipment moving has begun this month, and volume production is on track for late 2024. In China, we have recently received an extension from the U.S. Bureau of Industry and Security to continue our operation in Nanjing. We are currently in the process of applying for validated end-user authorization and expect to receive a prominent authorization in the near future. From a cost perspective, the initial cost of overseas fabs is higher than TSMC fab in Taiwan due to, first, smaller fab scale; second, higher costs throughout the supply chain; and third, the early stage of the semiconductor ecosystem overseas as compared to a matured ecosystem in Taiwan. TSMC's responsibility is to manage and minimize the cost gap to maximize the return for our shareholders. Our pricing will also remain strategic to reflect our value, which includes the value of geographic flexibility. We will also work closely with government to secure their support. At the same time, we are leveraging our fundamental competitive advantage of manufacturing technology leadership, large volume economies of scale to continuously drive our cost down. By taking such actions, TSMC has the ability to absorb the higher cost of overseas fabs and still deliver the long-term gross margin of 53% and higher and sustainable ROE greater than 25%. We remain firm in our commitment to maximize the value for our shareholders. Now let me talk about N3 and N3E ramp-up and progress. Our 3-nanometer technology, the most advanced semiconductor technology in both PPA and transistor technology, is already in production with good yield, and we are seeing a strong ramp in the second half of this year supported by both HPC and smartphone applications. We reaffirm N3 will contribute a mid-single-digit percentage of our total wafer revenue in 2023, and we expect a much higher percentage in 2024 supported by robust demand from multiple customers. N3E will leverage the strong foundation of N3 to further extend our N3 family with enhanced performance, power, and yield and provide complete platform support for both HPC and smartphone applications. N3E has passed qualification and achieved performance and yield targets and will start volume production in the fourth quarter of this year. We also continue to provide further enhancement of N3 technology, including N3P and N3X. With our strategy of continuous enhancement of our 3-nanometer process technologies, we expect strong multi-yield demand from our customers, and we are confident that our 3-nanometer family will be another large and long-lasting node for TSMC. Finally, I will talk about the N2 status. The recent surge in AI-related demand supports our already strong conviction that demand for energy-efficient computing will accelerate in an intelligent and connected world. The value of our technology platform is expanding beyond the scope of geometry scaling and increasing toward greater power efficiency. In addition, as process technology complexity increases, the lead time and engagement with customers also starts much earlier. As a result, we are observing a strong level of customer interest and engagement at our N2, similar to or higher than N3 at a similar stage from both HPC and smartphone applications. Our 2-nanometer technology will be the most advanced semiconductor technology in the industry in both density and energy efficiency when it is introduced in 2025. Our N2 technology development is progressing well and on track for volume production in 2025. Our N2 adopts a nanosheet transistor structure, which has demonstrated excellent power efficiency and will deliver full node performance and power benefits to address the increasing need for energy-efficient computing. As part of the N2 technology platform, we also developed N2 with a backside power rail solution, which is best suited for HPC applications. We are targeting the backside power rail to be available in the second half of 2025 to customers with production in 2026. With our strategy of continuous enhancement, N2 and its derivatives will further extend our technology leadership well into the future. This concludes our key messages, and thank you for your attention.

Thank you, C. C. This wraps up our prepared statements. Before we move into the Q&A session, I would like to remind everyone to please limit your questions to two at a time to give all participants a chance to ask their questions. If you would like to ask your questions in Chinese, I will translate them into English before our management responds. Now let's begin the Q&A session. Operator, can we please start with the first caller on the line? Thank you.

Yes, the first one to ask a question is Gokul Hariharan from JPMorgan.

Congratulations on a great result, and thanks for the details on N3 and N2. My first question is on the technology leadership. Given we are hearing a lot of competitive messaging from your U.S. IDM competitors/customers in the last few months, Intel seems to think that they will be getting into technology or process technology leadership in 2025. Just wanted to hear what does TSMC think of Intel's claim? And when TSMC thinks about customer engagement, do you feel you will lose a little bit of market share to Intel when it comes to the N2 or the first generation of nanosheet transistors? Or do you think your very high market share in N3 will continue into N2? That's my first question.

Okay. Thank you, Gokul. Please allow me to summarize your first question. So Gokul's first question is about technology leadership and competition with, I think, particularly IDM. He notes this IDM is very, messaging about taking over process technology leadership from TSMC. And so Gokul's question is, how do we see this? Are we concerned that we will lose market share at nanosheet or N2 to this IDM given their claims of regaining process technology leadership. Is that correct, Gokul?

Yes, that's right.

Well, Gokul, this is C. C. Wei. Let me answer your question with a very simple answer: no. But what I was stated a little bit longer? Actually, we do not underestimate any of our competitors or take them lightly. Having said that, our internal assessment shows our N3P, now I repeat again, N3P technology, demonstrated comparable PPA to 18A, my competitors' technology, but with an earlier time to market, better technology maturity, and much better cost. In fact, let me repeat again, our 2-nanometer technology without backside power is more advanced than both N4P and 18A, and it will be the most advanced technology in the semiconductor industry when it is introduced in 2025. Does that answer your question, Gokul?

Okay, that's clear. Can you discuss the demand related to AI? You mentioned strong demand on the data center side and that AI makes up about 6% of revenues this year, primarily from data centers. Are we seeing more engagement with AI demand in edge devices? Based on TSMC's expectations, will this be a significant growth factor for TSMC's leading-edge AI devices over the next one to two years? Do you believe this will be a larger driver for you?

Okay. Thank you, Gokul. So Gokul's second question is on AI-related demand. He notes, of course, that we have talked last time also about AI-related demand outlook and particularly focused on what we call server AI processes, or Gokul referring to data centers. So this question is really more about on the edge devices. Are we starting to see AI-related demand for edge devices? Do we expect it to be a big growth driver in the next one to two years for our leading-edge technologies as well?

Yes, we are observing activity from customers incorporating AI capabilities in end devices like smartphones and PCs. We hope this will enhance TSMC's AI business.

So do you see that happening in the next one year? Or is it something that will happen in a more long-term horizon? Just wanted to understand when to think about the cadence of this growth.

Okay. Let me answer briefly. It started right now, and we will expect that more and more customers will put AI capability into their end devices and products.

Okay. Thank you, Gokul. Operator, can we move on to the next participant, please.

Next one to ask a question, Charlie Chan from Morgan Stanley.

So my first question is about the cycle recovery. So much appreciate your comments about demand. So my question is about when do you expect that there will be an uptick in fab utilization, assuming demand is stabilizing and also inventory gets back to a healthy level? Because it's just very hard to believe your fab utilization outside of leading edge will stay at only 70%, 80%. So first question is about, do you see that overall fab utilization picking up anytime soon?

Okay. So Charlie, I believe your first question pertains to the cycle and specifically about when we can expect fab utilization to increase. This is really dependent on whether we think the cycle is reaching its lowest point and when we anticipate a recovery. Is that correct?

Yes.

Well, this is C. C. Wei again. Let me address your question. We are noticing some initial signs of demand stabilization in the PC and smartphone markets, which are the largest segments for TSMC. We anticipate that 2024 will be a strong growth year. However, we cannot definitively state if we have reached the lowest point yet, as it's too early to declare a significant rebound. Despite the ongoing uncertainties in the American market, we are considering customers' inventory management in the first half of 2024. We have already mentioned our strong technological leadership and our wide customer base, which are unique to TSMC and allow our customers to succeed in their markets. TSMC continues to experience healthy growth, which is why we can outperform the overall industry and are confident in achieving growth next year.

I just want to confirm something because we are noticing some positive trends and increased orders in the foundry sector. My second question is about AI. Over the past three months, have you revised your forecast based on the GPU or chip market? Recently, you mentioned implementing additional controls on AI shipments to China. Do you anticipate this will have any short-term or long-term effects on your expectations for AI semiconductor revenue growth?

Okay. Thank you, Charlie. So Charlie's second question is related to AI, two-parter. First, do we see over the last three months, I think his words were an upward revision in the demand from AI in the last three months? And then he wants to know, given the recent additional regulations announced, what would the impact to the AI demand be to TSMC, both for the short term and the long term?

The demand for AI is continuing to grow significantly. From TSMC's perspective, we currently face capacity limitations in supporting this demand. We are actively working on increasing our capacity to meet these needs. Recently, the U.S. government implemented new regulations that prevent certain products from being shipped to Mainland China. We are still assessing the situation, but so far, we believe the impact on TSMC is limited and manageable for the short term. For the long term, we are still evaluating the potential consequences.

Got it. So actually, there was one embedded question about the custom chip. So may I know your perspective about this custom chip long-term may kind of gain share or outgrow the GPU products? So what's your assumption for this custom chip in terms of the real contribution within AI or to TSMC? May I ask as a follow-up?

Okay. Let me answer. Whether a customer developed the CPU, GPU, AI accelerator, or ASIC for all types of AI applications, the commonality is that they all require the usage of leading-edge technology with stable yield delivery to support larger die sizes and a strong design ecosystem. All of those are TSMC's strengths. So, we are able to address and capture a major portion of the market in terms of semiconductor components in AI.

Operator, can we move on to the next participant, please?

Next one to ask a question is Bruce Lu from Goldman Sachs.

I believe the question is addressing the outlook for the next two years. Management previously mentioned a revenue compound annual growth rate of around 15% to 20% from 2021 to 2026. The smartphone business is expected to grow at or slightly below the corporate average, but it experienced a significant decline in 2023. Are we anticipating a sharp recovery for the smartphone business to align with the corporate average growth rate? Additionally, management indicated that the back-end business will grow in line with the corporate average. Will we observe stronger growth in the back-end business over the next two years?

Okay. Bruce, your question is really about our stated goal of achieving a revenue CAGR of between 15% to 20% in U.S. dollar terms from 2021 to 2026. You're interested in understanding the components of this, particularly regarding the smartphone market, which has experienced slower growth recently. How do we anticipate the smartphone market to grow over the next one to two years within this CAGR framework? Additionally, I believe you are also inquiring about the growth expectations for the back-end business, which we previously mentioned would grow slightly faster than the corporate average. What are our current expectations for that growth in the next one to two years? Is that correct, Bruce?

Bruce, this is Wendell. Yes, in the next two to three years, backing up to '21 to '26, we still expect that the smartphone, as a whole, will grow slightly slower than the corporate average. We still think that HPC will be the strongest one and will be the major growth contributor to our multiyear growth. This is your first question. As to the growth of back-end business, we still expect that the back-end business, as a whole, will grow slightly faster than the corporate average in the five-year time period.

So, the next question is for the technology cadence. I mean, for TSMC, for revenue contribution for 5-nanometer, we start to see the meaningful revenue contribution for final starting in the third quarter of 2020, and 3-nanometer is the third quarter of 2023. So the cadence looks like is longer for 3-nanometers versus 5-nanometer and 7-nanometers. What is the technology cadence moving forward? Are we able to see a meaningful revenue contribution by 2-nanometer in a two-year time frame or two years' time? I think the technology migration cadence is an important indicator.

Bruce's second question, again, as you said, is around the technology cadence. He notes that 5-nanometer started contributing revenue in Q3 '20, but I think you're saying 3-nanometer's revenue contribution in Q3 '23. So the cadence is growing longer to kind of three years between five and three. So what will be the technology cadence in the future? And thus, for 2-nanometer, when should we expect meaningful revenue as well? Is that correct, Bruce?

Okay. Bruce, let me answer the question. I think that we developed the technology to meet the customers' demand. That's the first priority for us. But then different customers may have different product schedule considerations. And as time goes by, the technology complexity actually becomes more and more complicated and our customers will design their product and react to the market situation. So let me answer the question in a very simple way: TSMC's technology cadence remains constant and supports our customers' growth. But when we get the same amount or some percentage of revenue, it will depend on customers' product schedule.

But the follow-up question is that if the customer doesn't really need the advanced technology as fast as before, maybe we slow it down a bit, which will make better returns, right?

Well, we don't slow down our technology development per se. We might slow down our capacity expansion according to customer demand. Did I answer your question, Bruce? That's what we are doing right now.

Operator, can we move on to the next participant, please?

The next one, we have Laura Chen from Citi.

I think my question is also similar to what Bruce mentioned about the capacity expansion plan. As we see that healthier inventory situation right now, and at the same time, the most advanced processes now depend on customers' demand. So just wanted to get your feeling about the overall CapEx outlook or capacity outlook into the next two years, do you feel that we'll be better to resume the year-on-year CapEx next year or later, considering there's still quite strong demand on N2 and N2 ramp-up? At the same time, the most advanced nodes now seem you will see maybe 2x more expensive on N2 versus N3. So my question is just about the future capacity expansion.

Okay. So Laura's first question again is on the capacity expansion plan. She notes that, of course, the inventory, as we said, is becoming healthier but also with N3 and N2. So she really wants to know what is the CapEx and capacity outlook planned for the next one to two years.

Right. Laura, it's Wendell. As mentioned, our capacity plan will depend on our customers' product plans. In terms of CapEx, we have invested heavily in recent years to capitalize on growth opportunities ahead. As we start to see returns from those investments, we expect the growth in our CapEx to stabilize over the next few years. This does not imply a decrease in dollar amounts, but we anticipate a reduction in capital intensity in the coming years. This is our current perspective.

I recall that previously, we mentioned about like we target longer-term to back to like mid-30s for our CapEx intensity. So is there any possibility we see we achieve that target next year?

It's not a target; it's a forecast based on customer demand, and it's also projected over three to five years, not just for the immediate next year.

Okay. Very clear. And also my second question is about N2 progress. I appreciate C. C.'s previous sharing on the progress and also the backside power timeline. I'm just wondering what would be the most challenging part if we migrate to like a backside power? And also, since the sister density will become totally different on nanosheet. So, I'm just wondering about the clients' migration or intention to adopt the most advanced nodes into the next two, three years? And what will be the most challenging technology perspective?

Okay. So Laura, let me make sure I got this right. So your second question is about N2, and you want to know, with the adoption of backside power rail, what is the most challenging part from a technology perspective, and then how do we see the customer adoption?

I'll answer that. As technology moves into more and more advanced nodes, the challenge is always there. Technology complexity increases dramatically. But we can do it, no doubt about it. And we still remain the technology leadership in this industry. If you ask me, what is the most challenging part, I would say it's cost. I mean, today inflation and everything, and the tools have become more and more expensive. Although we can do it on time to meet customers' requirements. Our challenge right now, actually, I would say, number one, cost. I want to reduce the cost so more customers can afford it. But even with that, actually, we have a lot of customers interested and engaged with TSMC today. Actually, it's probably higher than the N3 at a similar stage. Okay. Did I answer your question?

Yes, very clear, sir. Thank you, C. C.

Okay. Thank you, Laura. Operator, can we move on to the next participant, please?

Right now, we have Brett Simpson from Arete Research.

I have a question for C. C. about the hyperscalers. Major U.S. hyperscalers are currently hiring many chip designers and planning to develop their own AI silicon directly with TSMC, similar to what Apple has accomplished over the years. Is TSMC generally supportive of this trend? Additionally, can you provide your view on whether hyperscalers possess the necessary in-house IP and expertise to bypass the ASIC suppliers and go directly to TSMC?

Okay. Brett, thank you. So Brett's first question is looking at his observation, U.S. hyperscale companies are hiring a lot of people to do AI custom chips, silicon, and working directly or coming directly to work with TSMC. So his question is, is TSMC supporting such efforts and how do we see such type of customers, I guess? Is that your question, Brett?

Brett, okay. Those are hyperscalers. I don't comment on the specific customer. But all we know or our fundamental rule is whether the customer developed CPU, TPU AI accelerator, or ASIC for their own application or for any purpose in the AI area, we will support them, actually. And because of our technology leadership and our good manufacturing, so we are able to address and capture a major portion of the market. And so you are asking whether we support it or not, we support every customer all over the world.

Okay. That's very clear. And maybe just as a follow-up, I wanted to ask about some of the areas in the quarter that were weaker than expected, namely 7-nanometer, I think it's half year-on-year and also automotive that saw a decline sequentially, a meaningful decline sequentially. Can you just help us understand how you build back up 7-nanometer? What led to the weakness? And what's happening in automotive? And how do you assess prospects for automotive over the next six, nine months or so?

Okay. So two parts to Brett's question. Maybe I'll start with the second, by the way, looking at automotive demand and the weakness in the past quarter. How do we see the automotive demand? And then also, his question is about 7-nanometer, also year-on-year sharp revenue decrease. So how do we see the outlook for 7-nanometer as well.

Let me address the question regarding automotive demand. Over the past three years, automotive demand has been very robust, and we have met their requests. Currently, it appears that automotive demand has shifted into an inventory adjustment phase in the latter half of 2023. However, we anticipate that automotive demand will rebound in 2024 due to the increasing number of electric vehicles and additional functionalities being integrated into automotive designs. As for the 7-nanometer technology, the low utilization and decrease in revenue have surpassed our original expectations. We anticipated higher utilization of the N7 technology at this point, but that hasn’t happened. This has coincided with a significant drop in smartphone demand over the last decade, decreasing from about 1.4 billion units to approximately 1.1 billion units. This decline has affected N7 utilization, particularly due to a major customer delaying their product launch, which has contributed to the low utilization rate. Despite this, we are optimistic about filling our 7- and 6-nanometer capacity with additional specialty demand from consumer, RF, connectivity, and other applications, aiming to reach a healthy utilization level in the coming years. This situation is reminiscent of what we experienced with the 28-nanometer technology in 2018 and 2019 when it was also underutilized for a time. We collaborated with our customers and developers on specialty technologies, ultimately leading to an expansion of 28-nanometer specialty capacity. That experience reflects a similar trajectory here. Brett, did I answer your question?

Yes.

All right, Brett. Thank you. Operator, can we move on to the next participant, please?

Next one to ask questions, Brad Lin from Bank of America.

First of all, congratulations on the strong results and the impressive gross margin. I have two questions: one regarding end device AI and edge AI, and the other about the CPO. I appreciate the constructive comments and positive outlook on edge AI from management. Besides the interesting engagement with clients, what are the implications for wafer consumption for the firm? Additionally, focusing on computing power and energy consumption related to end devices with AI and their extra functions, should we expect a reacceleration in node migration for these end devices? That's my first question.

Okay. So Brad's first question is about edge and device AI. He wants to know what is the implications for wafer consumption and then with the increasing need for energy efficiency and power. He's wondering, does this reaccelerate or increase the node for, I guess, his words is node migration or adoption of leading-edge technologies.

The edge devices, including smartphones and PCs, are beginning to integrate AI functionality. We have observed an increase in the number of neural engines being added. While the overall unit production may not have risen dramatically, the die size has grown to mid-single digits. I expect this trend to continue, leading to more applications of AI being integrated into these edge devices. This necessitates very power-efficient chips, particularly for mobile devices. I believe my customers will soon transition to leading-edge technology nodes more swiftly in order to stay competitive in the market.

A bit of a follow-up is that it currently accounts for some mid-single digit increment in die size for a chip. Are we seeing this potentially grow to mid-teens or even larger in the mid- to long-term?

So Brad's quick follow-up is, if the AI portion is currently around mid-single digits, what percentage can we expect it to reach in a few years? Could it be mid-teens or something higher?

Well, I will answer the question. Actually, we see the increase in the die size for we cannot nail down, we say, the mid-single digit, but I expect it to start to increase. And whether that will increase our forecast and our growth or something, it's still too early to say at this moment.

Yes, we are still assessing the impact of this development. Therefore, we are sticking with our previous statement that we anticipate growth to approximately mid-teens of our revenue in about five years.

Yes. I think, Brad, probably just very simply, as we said, edge AI, we do see some activity. It will drive silicon content, but this will occur over time, okay? And we don't have any quantitative number to share, right?

Got it.

Okay. What is your second question?

Okay. So the second question is on CPO. So basically, we have learned that TSMC is doing quite well, also leading the industry in CPO or so-called silicon photonics and has introduced a platform to clients with the technologies. So, we learn about the opportunities and implications of the new technology for the industry and for our firm. And also, should we expect the platform to offer additional competitive advantage for TSMC in the long run?

So, Brad, your second question is on silicon photonics. Is that correct?

Okay. Let me answer that question. Silicon photonics actually is growing in importance because of just a larger amount of data needs to be collected to process and transfer in an energy-efficient manner. Silicon photonics tends to be the best fit for that role. And TSMC has been working on silicon photonics for years and most importantly, we're collaborating with multiple leading customers to support their innovations in this field. It takes a lot of time to develop the technology and to build the capacity. And when we increased the volume production, we believe that TSMC's silicon photonics will be the best technology when customers roll out all their innovations. But as I said, it's gradually increasing in their activity and gradually increasing their demand as of today.

Got it. Thank you very much.

All right. Thank you. Operator, can we move on to the next caller, please?

Next one, we have Sunny Lin from UBS.

So my first question is on advanced packaging. Incrementally, we are hearing more customer interest in the adoption to achieve better heterogeneous integration. But want to get your thoughts on what could be the potential impact of customers relying more on packaging to improve system performance and perhaps last on the price migration given cost considerations? Meanwhile, SOIC has been introduced for quite a while, whereas the customer adoption still seems to be limited at this point. And so when should we expect a more meaningful pickup of SOIC? And what could be the major catalyst?

Okay. So Sunny, sorry, I may have missed a little bit of the first part. But I think her question is on overall advanced packaging, looking at this trend and the move to more, of course, heterogeneous integration. What are the cost implications and how does advanced packaging work and go together with the process technology standpoint? And then also a question about the update or progress of SOIC, is that correct, Sunny?

Let me address that. The increase in costs is not due to the more advanced node. In fact, our customers are focused on maximizing system performance, which is a significant factor. This involves improving speed and reducing power consumption, among other aspects. Cost is also a consideration we have noted. More customers are transitioning to very advanced technology nodes and adopting chip-based approaches. Regardless, TSMC offers industry-leading solutions in both cutting-edge technology and advanced packaging, collaborating with our customers to achieve optimal system performance. Regarding the SOIC, when it will reach high volume and provide substantial revenue for TSMC, that is on the horizon. The customer is already prepared to announce their new product, which is set to gain widespread adoption. I expect that starting now and into next year, the SOIC will begin generating revenue and become one of the faster-growing advanced packaging solutions in the coming years.

Got it. If I may, a quick follow-up. Three months ago, you had a target to double your CoWoS capacities. And just now you mentioned AI demand continued to surprise on the upside. So I wonder if there's any update on your CoWoS capacity expansion?

Okay. So I will take this as your second question, Sunny, but she's asking about CoWoS expansion. We had said that we will double the capacity three months ago. Can we provide an update on the overall as CoWoS capacity, and I guess, CapEx and capacity go hand-in-hand. What is our plan?

Well, Sunny, the last time we said that we will double our CoWoS capacity. We are working very hard to increase the capacity more than double, but today is limited by my suppliers' capability and their capacity. So we still maintain that we will double our CoWoS capacity by the end of 2024. But the total output actually is more than double from 2023 to 2024 because of a very high demand from our customers. So actually then, this kind of trend will continue to increase our CoWoS capacity to support our customers even into 2025.

Okay. Sunny, does that answer? Okay. Thank you. Operator, can we move on to the next caller, please?

Next one, please welcome Mehdi Hosseini from FIG.

I understand there are several new products being introduced at the end of this year and into the first half of 2024 for various end markets. I would like to know how the introduction of these new products will affect seasonality. Is it possible that the ramp-up in the first half, particularly with these new products, could mitigate some of the seasonal impacts? I would appreciate any insights on this, and I have a follow-up question.

All right, Mehdi. Well, Mehdi's first question is in terms of new products, which, of course, customer products, we don't comment on, but he said we are ramping products into the second half. So how will this ramp of new products go into as we go into the first half '24? And can this offset or mitigate some of the seasonality?

Yes. Mehdi, I don't think we can comment on specific customer products, but I can tell you that we're not seeing any dramatic change in the seasonality as of now.

Okay. Because I'm looking at your year calendar '23, and given your Q4 guide, you're actually doing better than what you guided three months ago, once you said revenues could be down 10% U.S. dollar and now it could actually be down high single-digit. Is that a combination of a stronger new product ramp and better pricing? Is that a fair assessment?

Okay. Mehdi noted that three months ago, we projected a decline of about 10% in U.S. dollar terms for this year. With the guidance for the fourth quarter looking slightly better, he is asking about the implications of this change.

Well, let me give you one simple reason because our ramp-up of N3, because of the demand of N3 is strong. So we ramped up quickly to meet customer demand. So, the final result is better than we expected three months ago.

And we have also said that strong ramp of N3 will continue next year, okay? That's about all the seasonality color we can give.

Got you. Okay. And then perhaps, if I were to ask a second question, I just want to better understand your view on your customers' inventory correction. We're reaching the bottom where we don't have any visibility on how quickly they're going to refresh inventory. The slope of the recovery is still not clear. Did I understand you correctly?

So, Mehdi's second question, he would like to clarify. So are we saying that customers' inventory is reaching or approaching the bottom, but the slope of the inventory is not clear, is that what we are saying?

I will respond to your question. Over the past few months, we have noticed demand stabilizing in the PC and smartphone markets. We have even received urgent purchase orders requesting more devices to be shipped to meet this demand. This suggests that inventory is improving more than we anticipated. While uncertainty in the macro environment may persist, we expect to be very close to a healthy condition. Therefore, we believe 2024 will be a year of healthy growth for TSMC. Mehdi, did I address your question?

Great. Thank you.

Thank you, Mehdi. To save time, we'll take questions from the final two participants.

Next one to ask a question is Krish Sankar from TD Cowen.

I have two of them. First one is on gross margin. Where do you expect N3 to reach corporate average gross margin? And as you look into next year, as more mature node capacity comes online across the industry, how to think about mature node gross margins also? And I have a follow-up after that.

Okay. So Krish's first question is on gross margin. When do we expect 3-nanometer, or N3, to reach the corporate average gross margin? And how do we see the gross margin trend for the more mature nodes?

Yes. In the past, our leading node typically reaches gross margins comparable to the corporate margin in about eight quarters. However, as we develop more leading nodes, it will become increasingly challenging for a number of reasons. Firstly, our corporate margin is now higher than it was before. Secondly, the complexity of leading nodes is increasing. Additionally, unexpected inflation pressures in recent years have also added to the higher costs in N3. As a result, it will be quite challenging for future leading nodes to achieve the corporate margin within the same timeframe as before. Regarding mature nodes, I can share that the gross margins for our mature nodes are closely aligned with the corporate average within a narrow range because we concentrate on specialty technology rather than commodity capacity.

That's very helpful. As a follow-up, regarding Arizona, you mentioned that you've hired about 1,100 local employees. Is that enough to start 4-nanometer production, or do you need to reach a different employee target before you can actually begin this production, especially since you're still aiming for output in the first half of 2025?

Okay. So Krish's second question on is about our first fab in Arizona. He notes that we have said we hired 1,100 local employees. So his question is this enough critical mass or enough people basically to support the ramp of the first fab as we plan, as we said today in the first half.

Of course, we continue to hire the local talent to join TSMC's fab in Arizona. So, when we start to have volume production, we are confident that we will have enough resources to support our ramp-up in Arizona.

Okay. All right. Thank you, Krish. Operator, can we move on to the last participant, please?

The last one to ask a question is Charles Shi from Needham.

First off, I really want to congratulate TSMC for delivering good results for Q3 and very good guidance for Q4. But I want to really call out the reported revenue for 5-nanometer in the third quarter. It looks like you are showing some really good countercyclical strength and probably a record high. I want to understand the rebound in the 5-nanometer business in Q3. Is that going to be more in the following quarters? And what's behind that? And relative, let's say, your expectation like three to six months ago when you were reducing the '23 outlook, is 5-nanometer doing better than expected? And how has the demand been trending in the last two, three months for 5-nanometer?

Okay. So Charles' first question is about 5-nanometer. He's asking in the very near term. He notes that he saw a very strong sequential revenue increase in the third quarter. So, he's wondering what is driving this. And then he's asking about what is the outlook for the next three to six months, particularly 5-nanometer.

Yes, I can share that the increase in revenue for N5 in the third quarter mainly comes from two platforms, HPC and smartphones. HPC also includes AI-related demand, while the smartphone demand is influenced by some customers' seasonal product cycles. I won't provide forward-looking guidance at this moment, but we will update you in January regarding the actual N5 revenue for the next quarter.

The overall revenue. We don't provide revenue by process node, okay? What's your second question, Charles?

Yes. Thanks, Jeff. The other question is about CapEx. It sounds like that you're expecting CapEx on an absolute dollar might still grow. Going forward, I know that's a long-term comment. But I look at the near term, TSMC CapEx seems to be running at US$7 billion per quarter in the second half of '23, which kind of is at US$28 billion annualized run rate. But if we are expecting total CapEx for '24 to grow in dollar terms over '23, it seems like you're expecting CapEx ramp in 2024. Maybe that's your planning for some of the CapEx ramp in '24. Is that the right way to think about CapEx? Is US$7 billion really the bottom level run rate for TSMC CapEx at this point?

Charles' second question is about capital expenditures. He is noting that based on our guidance, our CapEx is running at approximately US$7 billion per quarter. He is presuming, although we do not discuss 2024, that the CapEx amount for next year may increase. If we maintain a US$7 billion run rate, does that suggest a total of US$28 billion? How should he understand this?

Charles, every year, the CapEx is invested based on the future opportunity to grow, and we invest to capture those future opportunities. Too early to talk about 2024, really, we will share the guidance with you in January quarterly release.

Okay. All right. Thank you, Charles. 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. Both of these you can find through TSMC's website at www.tsmc.com. So thank you, everyone, for taking the time to join us today. We hope everyone continues to be well, and we hope to see you join us again in January. Goodbye, and have a good day.