[Non-English content] Good afternoon everyone, and welcome to TSMC's first quarter 2021 earnings conference call. This is Jeff Su, TSMC's Director of Investor Relations and your host for today.
To prevent the spread of COVID-19, TSMC is hosting our earnings conference call live, 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 first quarter 2021, followed by our guidance for the second quarter 2021. Afterwards, Mr. Huang and TSMC's CEO, Dr. C.C. Wei, will jointly provide the company's key messages. We will open the line for 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 on our press release. Now, I would like to turn the call over to TSMC's CFO, Mr. Wendell Huang, for the summary of operations and current quarter guidance.
Thank you, Jeff. Good afternoon, everyone. Thank you for joining us today. My presentation will start with financial highlights for the first quarter 2021. I will provide the guidance for the second quarter 2021. First quarter revenue increased 0.2% sequentially in TWD, or 1.9% in USD. Our first quarter business was supported by HPC related demand, balanced by a milder smartphone seasonality than in recent years. Gross margin decreased 1.6 percentage points sequentially to 52.4%, mainly due to relatively lower level of capacity utilization and an unfavorable foreign exchange rate. Total operating expenses slightly increased by 0.8 billion TWD, mainly due to higher level of R&D activities for the N5 family. Operating margin decreased by 2 percentage points sequentially to 41.5%. Overall, our first quarter EPS was TWD 5.39 and ROE was 29.5%. Let's move on to revenue by technology.
5 nm process technology contributed 14% of wafer revenue in the first quarter, while 7 nm accounted for 35%. Advanced technologies, which we now define as 7 nm and below, accounted for 49% of wafer revenue. Moving on to revenue contribution by platform. Smartphone decreased 11% quarter-over-quarter to account for 45% of our first quarter revenue. HPC increased 13% to account for 35%. IoT increased 10% to account for 9%. Automotive increased 32% to account for 4%. DCE increased 10% to account for 4%. Moving on to the balance sheet. We ended the first quarter with cash and marketable securities of TWD 797 billion. On the liability side, current liabilities increased by TWD 45 billion, mainly due to the increase of TWD 49 billion in short-term loans and increase of TWD 50 billion in accrued liabilities and others, partially offset by the decrease of TWD 51 billion in accounts payable.
Long-term interest-bearing debt increased by TWD 23 billion, mainly as we raised TWD 21.1 billion of corporate bonds during the quarter. On financial ratios, accounts receivable turnover days increased one day to 40 days. Days of inventory increased 10 days to 83 days, primarily due to N5 wafer prebuild. Let me make a few comments on cash flow and CapEx. During the first quarter, we generated about TWD 228 billion in cash from operations, spent TWD 248 billion in CapEx, and distributed TWD 65 billion for second quarter 2020 cash dividend. Short-term loans increased by TWD 52 billion, while bonds payable increased by TWD 18.5 billion due to the bond issuance.
Overall, our cash balance increased TWD 4.6 billion to TWD 665 billion at the end of the quarter. In U.S. dollar terms, our first quarter capital expenditures total $8.8 billion. I have finished my financial summary. Let's turn to our second quarter guidance.
Based on the current business outlook, we expect our second quarter revenue to be between $12.9 billion and $13.2 billion, which represents a 1% sequential increase at the midpoint. This revenue guidance includes the minor impact from the power outage that occurred yesterday at our Fab 14 in Tainan. Based on the exchange rate assumption of one U.S. dollar to 28.4 TWD, gross margin is expected to be between 49.5% and 51.5%. Operating margin between 38.5% and 40.5%. The sequential decline in second quarter gross margin is mainly due to the margin dilution from higher 5 nm contribution, the slower rate of cost improvement as our fabs continue to run at a very high level of utilization, and the absence of positive inventory revaluation. This concludes my financial presentation. Let me turn to our key messages. I will start with our near-term demand and inventory.
We concluded our first quarter with revenue of TWD 362.4 billion, or $12.9 billion, which was in line with our guidance. The slight sequential increase was mainly driven by HPC-related demand, balanced by a milder smartphone seasonality than in recent years. Moving into second quarter 2021, we expect our revenue to be flattish as HPC-related demand will continue to grow, offset by smartphone seasonality. On the inventory front, our fabless customers' overall inventory was healthy exiting fourth quarter of 2020. Amidst the lingering macro and supply uncertainties, we expect our customers and the supply chain to gradually prepare higher levels of inventory throughout the year as compared to the historical seasonal level. We expect this to persist for a period of time, given the industry's continued need to ensure supply securities.
Looking ahead to the second half of the year, we expect our capacity to remain tight throughout the year, supported by strong demand for our industry-leading advanced and special technology. For the full year of 2021, we now forecast the overall semiconductor market, excluding memory, to grow about 12%, while foundry industry growth is forecast to be about 16%. For TSMC, we are confident we can outperform the foundry revenue growth and grow by around 20% in 2021 in US dollar terms. Next, let me talk about our capital budget for this year. Every year, our CapEx is spent in anticipation of the growth that will follow in future years. As we enter a period of higher growth, underpinned by the multi-year structural megatrends of 5G related and HPC applications, we believe a higher level of capital investment is necessary to capture the future growth opportunities.
In order to meet the increasing demand for our advanced and specialty technologies in the next several years, we have decided to raise our full year 2021 CapEx to be around $30 billion. About 80% of the 2021 capital budget will be allocated for advanced process technologies, including 3 nm, 5 nm, and 7 nm. About 10% will be spent for advanced packaging and mask making, about 10% will be spent for specialty technologies. Let me turn the microphone over to C.C.
Thank you, Wendell. We hope everybody is staying safe and healthy during this time. Let me talk about the capacity shortage and demand outlook. Our customers are currently facing challenges from the industry-wide semiconductor capacity shortage, which is driven by both a structural increase in long-term demand as well as short-term imbalance in the supply chain. We are witnessing a structural increase in underlying semiconductor demand as a multi-year mega trend of 5G and HPC-related applications are expected to fuel strong demand for our advanced technologies in the next several years. COVID-19 has also fundamentally accelerated the digital transformation, making semiconductors more pervasive and essential in people's lives. The need to ensure supply security is creating short-term imbalance in the supply chain, driven by supply chain disruption due to COVID-19 and uncertainties brought about by geopolitical tensions. Let me talk about TSMC's investment plan and disciplines.
TSMC's mission is to be the trusted technology and capacity provider for the global logic IC industry for years to come. In order to support our customers' growth, TSMC is taking several actions to help address the capacity shortage for our customers. We are working hard to increase our productivity to drive more output to help support our customers for the near term. To address the structural increase in the long-term demand profile, we are working closely with our customers and investing to support their demand. We have acquired land and equipment and started the construction of new facilities. We are hiring thousands of employees and expanding our capacity at multiple sites. TSMC expects to invest about $100 billion through the next three years to increase capacity to support the manufacturing and R&D of leading-edge and specialty technologies.
Increased capacity is expected to improve supply certainty for our customers and help strengthen confidence in global supply chains that rely on semiconductors. Our capital investment decisions are based on four disciplines: technology leadership, flexible and responsive manufacturing, retaining customers' trust, and earning the proper return. At the same time, we face manufacturing cost challenges due to increasing process complexity at leading node, new investment in mature nodes, and rising material costs. We are continuing to work closely with customers to share our value. Our value includes the value of our technology, the value of our service, and the value of our capacity support to customers. We are looking to firm up our wafer pricing to a reasonable level. We will continue to work diligently with our suppliers to deliver on cost improvement.
By taking such actions, we believe we can continue to earn a proper return that enables us to invest to support our customers' growth and fulfill our mission as trusted foundry partners. With our technology leadership, manufacturing excellence, and customer trust, we are well-positioned to capture the growth from the favorable industry mega-trend. We reiterate our long-term revenue to be 10%-15% CAGR from 2020 to 2025 in U.S. dollar terms. Let me talk about the automotive supply update. The automotive market has been soft since 2018. Entering 2020, COVID-19 further impacted the automotive market. The automotive supply chain was affected throughout the year, and our customers continued to reduce their demand throughout the third quarter of 2020. We only began to see sudden recovery in the fourth quarter of 2020. The automotive supply chain is long and complex, with its own inventory management practices.
From chip production to car production, it takes at least 6 months with several tiers of suppliers in between. TSMC is doing its part to address the chip supply challenges for our customers. In January of this year, TSMC announced that capacity support for automotive customers is our top priority. Since then, we have worked dynamically with our other customers to reallocate our wafer capacity to support the worldwide automotive industry. However, the shortage further deteriorated due to the unexpected snowstorm in Texas and the fab manufacturing disruption in Japan. Together with our productivity improvement, we expect the automotive component shortage from semiconductor to be greatly reduced for TSMC's customers by the next quarter. Now I will talk about Taiwan water supply update. The water supply in Taiwan is currently tight due to the lack of rainfall in the past one year. We have been prepared for this.
TSMC has a long-established enterprise risk management system in place, which covers water supply risk as well. Through our existing water conservation measures, we are able to manage the current water usage reduction requirement from the government with no impact on our operations. We also have detailed response procedure to handle water shortage at different stages. We will continue our collaborative effort with the government and the private sector on water conservation and new water sources. With our comprehensive enterprise risk management system, we do not expect to see any material impact to our operations. Finally, I will talk about the N5 and N3 status. TSMC's N5 is the foundry industry's most advanced solution with the best PPA. N5 is already in its second year of volume production, which yield better than our original plan.
N5 demand continue to be strong, driven by smartphone and HPC applications, and we expect N5 to contribute around 20% of our wafer revenue in 2021. N4 will leverage the strong foundation of N5 to further extend our 5 nm family. N4 is a straightforward migration from N5 with compatible design rules, while providing further performance, power, and density enhancement for the next wave of 5 nm products. N4 risk production is targeted for second half of this year, and volume production in 2022. Thus, we expect demand for our N5 family to continue to grow in the next several years, driven by the robust demand for smartphone and HPC applications. N3 will be another full node stride from our N5 and will use FinFET transistor structure to deliver the best technology maturity, performance, and cost for our customers. Our N3 technology development is on track with good progress.
We continue to see a much higher level of customer engagement for both HPC and smartphone applications at N3 as compared with the N5 and N3 at a similar stage. Risk production is scheduled in 2021, and volume production is targeted in second half of 2022. Our 3 nm technology will be the most advanced foundry technology in both PPA and transistor technology, when it is introduced. Thus, we are confident that both our 5 nm and 3 nm will be large and long-lasting nodes for TSMC. This concluding our key message. Thank you for your attention.
Thank you, C.C. This concludes our prepared statements. Before we begin the Q&A session, I would like to remind everybody to please limit your questions to two at a time to allow all participants an opportunity to ask their questions. Should you wish to raise your question in Chinese, I will translate it to English before our management answers your question. For those of you on the call, if you would like to ask a question, please press the zero, then one on your telephone keypad now. Questions will be taken in the order in which they were received. If at any time you would like to remove yourself from the questioning queue, please press zero two . Now, let's begin the Q&A session. Operator, can we please proceed with the first caller on the line?
The first one to ask question, Randy Abrams, Credit Suisse.
Okay, yes. Thank you. I wanted to ask the first question just about Intel. They did announce their plans to reengage in the foundry sector, and also, I think making it clear their goals are to get back to manufacturing leadership. Could you discuss how you're viewing them now as a customer? The assurances you're getting on business sustainability and how you're managing the potential risk if they improve manufacturing and pull back on some of the outsourcing plans.
Okay, Randy. Let me summarize your first question. Randy's first question relates to Intel and their recent announcement to reengage on foundry and get back to a manufacturing leadership position. Randy's question is: how does TSMC view Intel as a customer? What kind of business assurances are we getting on the sustainability of the business? How do we manage any of the potential risks?
Randy, let me start with TSMC is everyone's foundry. Support all our customers openly and fairly. Intel is an important customer, and we will collaborate in some areas and compete in other areas. We always work with our customers to develop the necessary technology to support their products. Now let me comment a little bit on the competition. As a leading pure-play foundry, TSMC has never been short on competition in our 30+ year history, and we know how to compete. We will continue to focus on delivering technology leadership, manufacturing excellence, and earning our customers' trust. The last point, customers' trust, is very important because we do not have internal product that competes with our customers. We can be the trusted technology and capacity provider and for years to come. If you ask other comment, how we support Intel, we support them as an important customer.
We plan our capacity for the long-term industry mega-trend also. It's not for the short-term demand.
Okay. Does that answer your first question, Randy?
No, that's good on the first question. The second question and topic I wanted to discuss, you mentioned in your prepared remarks about there is a bit more geopolitical pressure with particularly U.S., but also Europe and China. They all are being aggressive about domestic capacity. If you could give an updated view on your strategy, if any shifts at the margin, where you've traditionally wanted a high scale in Taiwan. I'm curious for U.S., with the big land if there's any plans to accelerate positioning with the potential eventually you could do mega fab. If you could give an update on Nanjing, if there's plans to expand from the current, I think you're at 20K. If from a customer level, you're seeing shifts where more customers are starting to consider geographic location in the foundry consideration.
Okay, Randy. Let me try to summarize your second question. I think there's quite a few parts.
Oh, okay.
I think, first, Randy's question is on looking at sort of the geopolitical landscape, and looking at this talk of fabs in different countries. I think Randy first wants to know what is our progress or how do we see particularly U.S. manufacturing. Secondly, in other areas. Thirdly, he would like an update on the Nanjing expansion. Lastly, how do customers feel about the need to manufacture in different countries? Is that correct, Randy?
Yeah, that's correct. Yeah. Thank you.
Randy, that's a lot of questions.
Yeah. Sorry. You only give us two.
Let me try to answer. The first one, actually, I would like to say TSMC is helping a global company. We have a lot of manufacturing sites outside Taiwan, U.S., mainland China, Singapore. Let me comment on the U.S. first. We have been in the U.S. for a long time, though. We set up a WaferTech, that an 8-inch fab located in upstate Washington back in 1996. It's continued to operate and manufacture chips for our customers today. Now we are increasing our presence in the U.S. with an advanced 12-inch semiconductor fab in Arizona, and the progress is executing to our plan. We are happy that we are joined the effort to support semiconductor manufacturing in the U.S. You also ask about our status in Nanjing. Did you ask that?
Yes.
The fab in Nanjing is progressing well. We already completed the first phase of 20,000 wafers capacity installation, and actually, it's in production for a while. We also have a plan, depending on the customer's demand and depending on the economics, we have a plan to expand the capacity also. Okay. Other question in Taiwan, why we
I think also-
Yeah, two.
Two others. One, Randy is asking, with other regions also.
We never rule out any possibility with other regions. Today, we already announced plans. We currently have no further fab expansion plan in other areas such as Europe. We did not rule out any possibility. However, I want to emphasize, Taiwan will continue to be the main focus for TSMC. Our center of R&D and the majority of production line will continue to be located in Taiwan. Okay. Did that answer your questions?
Yeah, that's clear. Maybe if you can clarify, for customer decisions, just that approach with Taiwan , are you starting to see it get raised a bit more about customers choosing location? Are they still focused on the traditional just getting the best PPA, the best cost, and delivery time?
Right. Randy's, the last part of his question is from a customer's perspective, is there a push by customers for this geographic diversification or what do customers want?
Well, our customer welcome we establish the new fab in Arizona State. Let me say that. However, the most important one to them is the technology, is the manufacturing, is the efficiency that TSMC provide. Okay. Actually, that's the most important one, other than the consideration of geopolitical locations.
Okay, Randy?
Okay, great. That's clear. Yes, thank you. That's clear.
Okay. Thank you, Randy. All right. Operator, can we move on to the next person on the line, please?
Next one on the line, Gokul Hariharan, JP Morgan. Go ahead, please.
Thanks for taking my question. My first question, could we talk a little bit about the TWD 100 billion capacity plan for the next three years? Is that primarily a CapEx number? Just assuming that this year already we are spending about $ 30 billion. Should we be assuming that our CapEx is going to run around these levels or even higher in the next two years as well? Just wanted to clarify because there was some confusion about whether that's a CapEx number or a CapEx plus R&D number.
Okay. I think, Gokul, your first question is that we intend to spend $ 100 billion in the next three years. Is that a CapEx number? What does that mean for the spending at the CapEx in the next two years?
Hi, Gokul. This is Wendell. Yes, $100 billion is CapEx number. We've already guide that this year will be $ 30 billion. We're not going into the linearity in the next two years. You can actually have a feeling about what we will be spending in the next two years.
Got it. That's very clear. My second question is on the inventory cycle and a lot of the capacity expansion that we are seeing in older technologies. TSMC also is spending about roughly $ 3 billion based on the new guidance on special ty technologies as well. The industry also seems to be spending quite a bit of capacity there. What is TSMC's take in terms of when we are going to see a bit more normalization in some of this older capacity? Does TSMC also subscribe to the view that even in 2022, we are likely to see some degree of capacity tightness or capacity shortage? TSMC feels that we will likely resolve this towards the end of this year or early next year?
Okay, Gokul. Thank you. Let me try to summarize your second question. Your second question is asking about the inventory cycle, and particularly on the mature nodes. Looking at the expansion in the mature nodes, and Gokul wants to know, I think that on the mature nodes, could we see some type of overcapacity, or can the tightness continue to persist, or will we start to see some kind of overcapacity or oversupply towards the end of this year or in 2022? Is that correct, Gokul?
Yeah. I think many of your competitors are talking about 2022 also being undersupplied in many of these process nodes. Just wanted to hear TSMC's view on that.
Well, Gokul, let me answer the question carefully because we cannot rule out the possibility of an inventory correction or overbooking, something like that. Actually, we expect the structural demand to continue, and we will work with our customer closely, actually, and to develop some technology solution to meet customers' requirement and create a different differential and long-lasting value to our customer. As a result, we actually see the demand continue to be high and the shortage will continue throughout this year and may be extended into 2022 also. Did that answer your questions?
Okay. Do you also feel that customers will continue to hold on to a higher level of inventory for quite some period of time? Is that the way you think about inventory as well?
Yeah. We expect the customer, almost all of them, to prepare a higher level of inventory. That is because of today, geopolitical tension continue to persist. Even the COVID-19 will recede sometimes, we hope as soon as possible, but it will continue for a while, and put factors together, and we do expect them to prepare a higher level of inventory. I believe Wendell already saying that.
Yeah. Okay, Gokul. Does that answer your second question?
Thank you.
Okay. Thank you. Operator, let's please move on to the next caller on the line, please.
Now we have Sebastian Hou from CLSA.
Good afternoon, gentlemen. Thanks for taking my questions. The first one is on the pricing strategy. I remember that six months ago, the company talked about sticking to the principle of respecting long-term partnership with customers, and the company doesn't seem to want to change the pricing on the mature technology nodes, which I mean 28 nm above. I'm wondering if that's still the case now, or if the company now considers some upward adjustment. If it's the latter, what has changed versus six months ago? Thank you.
Okay, Sebastian. Sebastian's first question is regards to pricing. He says that we always talk about long-term partnership with our customers, and he's saying, in particular, on the older nodes, 28 nm and such, would we raise the price? He's asking sort of what is the pricing strategy today, and what has changed versus previously.
Sebastian, let me answer that. For more than 30 years, TSMC has provided stable and predictable pricing, we have refrained from opportunistic or short-term actions. Now, as I said in my statement, the cost structure start to structure change, because we have to invest in the leading-edge technology, which is more complex than ever. We also increase the mature technology node capacity, which a lot of them have already been fully depreciated. Now we have to invest in the new tools. We refrain from our opportunistic and short-term action, but we also have to share our value. We are working with our customer closely, and we want to firm up our wafer pricing to a reasonable level.
We are also working with our supplier to deliver the cost reduction, and we want to earn a proper return that enables us to continue to invest to support our customers' growth. In today's term, capacity support is the most important one they are looking for. Okay.
Okay, Seba-.
Got it. That's fair.
Do you have a second question?
Yeah, sorry. Go ahead.
Yes, I do. I have a second question. Thank you. I think for many reasons, we have seen that many countries globally, they plan or they want to increase or build their own semiconductor fabrication capacity domestically. We're also seeing some IDM. They are forced to increase their insourcing or add some internal capacity because of the chip crunch. My question is that IDM outsourcing has been one favorable driver for the foundry industry and TSMC growth in the past three decades, and how TSMC see this trend evolving in coming years? Would you be concerned this could lead to some overcapacity in a few years, even if some of those may not be effective? This is my second question. Thank you.
Okay, Sebastian, let me summarize your question. Your observation that countries are pushing for more domestic manufacturing, and IDMs are also looking at expanding capacity. Sebastian's question is looking at IDM outsourcing. Do we see this trend slowing down, or how do we see it in the next coming few years? Could this capacity that's being built result in excess capacity?
Well, Sebastian, let me say that in our long-term forecast, we continue to see the IDMs outsourcing continue to increase. We prepare the capacity for that also. We don't think that IDM trying to expand their own capacity will result in overcapacity situation. Technology is the most important thing, let me say that. We expand our capacity based on the customer's need, and with the technology leadership that provided their product to be very competitive in the market. They are all happy to work with TSMC in developing their product for now, for the future. As a result, we continue to see the increased outsourcing from IDMs.
Okay, Sebastian, does that answer your second question?
Yes. It does. Thank you.
Okay, thank you. Operator, can we move on to the next person on the line, please?
Next one to ask question, Charlie Chan from Morgan Stanley.
Hi, good afternoon. Thanks for taking my question. First of all, can I ask about the change of the 2021 revenue guidance? Can you explain where is the upside coming from? By applications would be great. Does that include some pricing adjustments for just the revised out the revenue guidance? Thank you.
All right. I think Charlie's first question is relating to our 2021 revenue guidance, from now of around 20%, to say what has changed versus last time. He also wants to know, can we talk about by application, what is driving this change. What was the last part of the question, Charlie, sorry?
Actually, that's it. Thank you.
Okay.
Does that capture?
Oh, okay.
Price hike as well? Thank you.
His question is, what is driving the change in the growth guidance for this year? He would like to know which applications are driving it. Does it include some price increases in this guidance?
Okay, Charlie. First of all, we don't comment on price. I can share with you that we are everyone's foundry. Our CapEx and capacity planning are based on the long-term demand profile underpinned by the industry mega trends, not short-term cyclical factors. We are seeing stronger engagement with more customers on 5 nm and 3 nm as compared to three months ago. We work closely with our customers to plan the capacity, and we'll continue to focus our investments on advanced and specialty technology to support our customers' structural growth. This year, in terms of platform, we expect that HPC and automotive platform growth will be higher than the corporate average, and the smartphone and IoT will be close to the corporate average.
Okay, understood. It seems like the upside coming across border or just some specific application. I know that HPC, automotive are growing better, just compared to last guidance, what is driving the upside?
Okay. Actually, all the platforms have upsides compared to three months ago.
Okay. Got you. Thanks for the Wendell. My next question is about your capital intensity in the long term. I think one or two quarters ago, company updated their capital intensity, and at some point can fall back to 35% capital intensity. Linked to that, what does that mean to the long-term gross margin trend? Because in today's conference call, I keep hearing some comments about structural cost increase. I'm not sure if you said about the chemicals or equipment price, would that impact company's long-term gross margin trend? Thanks.
Sure. Charlie, let me share with you. I'm sorry. Go ahead.
Yeah.
It's okay.
Yeah. I think, Charlie, your question is on capital intensity, looking at what is the capital intensity looking like the next few years, and how does this correlate with our stated long-term capital intensity of mid-30s range? He also, on the back of that, what does this mean for the long-term gross margin trend?
Okay.
Yes.
Charlie, in terms of capital intensity, I've actually given out several points already. First of all, if you look at, we're saying the next three years we'll be spending $100 billion. This year will be TWD 30 billion. We also say that in the next five years, we expect to grow between 10%-15% revenue CAGR. If you do math, you probably will have a good idea about where our capital intensity will be in the next three years. Now, at this moment, we still expect that the capital intensity will go back to mid-30% level in the longer term. That's the capital intensity. In terms of gross margins, I think as C.C. has already mentioned, we see some challenges from manufacturing costs due to the increasing complexity of leading nodes, the new investment in mature nodes, and some rising material costs.
Therefore, we're taking actions to ensure that we earn a proper return by firming up our price, working with the supplier to drive the cost improvement. We expect that the 50% gross margin remains our target and is achievable.
Okay. Yep.
Yeah.
That's very clear. Can I assume part of that TWD 100 billion CapEx is also associated to the cost increase? If that is the case, how much of that is due to the cost increase versus demand?
I think the last part of his question is that out of the TWD 100 billion in this higher capital intensity, how much is due to the cost versus the demand?
Yes, exactly.
Okay. Charlie, basically we're seeing more engagement of our demand in the next few years. I would say most of the CapEx comes from the strong demand for our advanced technology and specialty technology, especially 5 and 3 nm .
Okay. Understood. Yep. Thank you.
Thank you, Charlie. All right. Operator, thank you. Can we please move on to the next caller, please?
Next one to ask questions, Bruce Lu from Goldman Sachs. Go ahead, please.
Hi. Thanks for taking my questions. My question, I want to stick with the $ 100 billion CapEx. I think this is the first time for TSMC to announce a multi-year CapEx. I think this suggests very strong growth even beyond 2023 or through 2025. Can you give us a little bit more color about what kind of patient demand, which is strong enough to give the company such a high confidence for the CapEx? We've seen through the various cycles, how can we have confidence for the demand three to five years down the road? Assuming TSMC mostly invests in advanced nodes, do you foresee the mature node capacity tightness continue, and how and when this can be resolved? Thank you.
Okay. Bruce has two questions. First is related to our CapEx. With such a high level of spending, what is giving us the confidence that we see out over the next several years to intention to spend this TWD 100 billion? His second Well, maybe we'll go that first, and then second question.
Okay, let me answer that one first. In fact, we are seeing stronger engagement with more customers on 5 nm and 3 nm . The engagement is so strong that we have to really prepare the capacity for it, and that's the main reason. What is the second?
His second question is looking at our CapEx, with the majority of our CapEx being on advanced nodes. On the mature nodes, will the supply-demand gap in mature nodes further widen?
We did see the gap that mature node capacity not enough to support all the products in the market. We are working with our customer closely to analyze the gap, and we are also preparing to invest on the mature node, as I said in my statement. Most important, we are developing the technology, specialty technology with a mature node, and to support our customers' need so their product can be very competitive in the market, and so we can have demand secured for the next few years, and we decide to invest on the mature node capacity.
Okay, Bruce, does that answer your two questions?
Can I take one?
We have to limit it to two, sorry. There's a lot of people still waiting. Thank you.
I will take that.
Operator, can we move on to the next, please?
Next one to ask questions, Rob Sanders from Deutsche Bank. Go ahead, please.
Yeah, my first question, thanks for taking my question, is regarding your CapEx rising up to the mid-30s by 2023. Are you asking customers to commit earlier than normal on that capacity? Are you considering asking customers for prepayments? How do you de-risk those capacity plans, and are you seeing an increased willingness to single source? My second question was, how far are you actually booked out on capacity, and at which node is the biggest gap between demand and your capacity? Thank you.
Okay, Robert. We will take your questions one at a time. His first question is looking at our CapEx for the next few years. With this level of spending, do we see customer commitments that are earlier than normal? Are we looking for things like prepayments from customers to secure their commitments? This is the first question.
Okay. Robert, let me answer this first. The TWD 100 billion CapEx is decided because we see the fundamental structure demand increase from the multi-year megatrend, and the acceleration of the digital transformations. We cannot disclose the detail of our commercial terms with our customers. For us to make the investment decision will definitely require proper returns and secure customer commitments.
Okay. Robert, second question is, how far are we booked in advance in terms of our demand, and which nodes do we see the biggest gap between what customers may want?
Well, I cannot comment on which node, because almost all the nodes are in high demand today. However, let me stress again that our investment in the capacities are for the future many years to come, because we work with customer closely and to plan for the next few years as capacity support to them. The customer talking to TSMC, and they lay out their product plan for the next few years, at least three to four years, and we plan the capacity for that.
Okay. Thank you.
C.C., thank you, Robert. Operator, can we please move on to the next caller, please?
Now we have Brett Simpson from Arete Research.
Yeah, thanks very much. I had a question on the crypto activity at TSMC. I guess we've seen record hash rate expansion around Bitcoin ASICs and Ethereum GPU mining in the last couple of quarters. Can you maybe share with us what portion of HPC sales is crypto at the moment? Then as we get into the second half of the year, should we expect this to decline? I wanted to get your perspective. A couple years ago, we had extreme volatility around crypto. Bitcoin is now a trillion-dollar market cap. Is this good business for TSMC? Do you think this time will be different? Just wanted to get your perspective on this. Thanks.
Okay, let me repeat your first question, Brett. He's asking about within HPC, looking at cryptocurrency, he's asking what is the contribution we're seeing from cryptocurrency, or crypto mining I should say, to our revenue. How do we expect this to go in the second half of this year? A longer-term question, which is how do we view this business?
Let me answer the question. TSMC's technology is a leading technology, and that's why even cryptocurrencies are mining using TSMC technology a lot. I cannot comment on what is the percentage or how much of this particular market sector to our revenue. However, I can say that cryptocurrency mining today is more mature than it was two or three years ago. It remains a volatile market. However, we will continue to work closely with our customers in this field.
Okay, Brett, do you have a second question?
Okay, great.
Yeah, sorry.
Yeah, thanks. I wanted to talk about the inventory levels at TSMC at present. It grew quite significantly, and I think you mentioned it was N5-related. Many of your smartphone customers are saying they have shortages at leading edge, and you're building inventories at 5 nm , so how do we reconcile that? Just looking at Q2, would you expect inventories to rise again in the June quarter? Thanks.
Okay, Brett. You're asking about TSMC's inventory days, right? Brett is asking what is leading to the increase in the inventory days at the end of first quarter, and then how do we expect this to trend in the second quarter?
Okay, Brett. We pre-build for our customers during seasonal low level, as we did before. Now, when we start to ramp in the higher season, the inventory usually come down naturally, as before.
Okay. Does that answer your question, Brett?
Yeah. Thank you.
Yep. Great. Perfect. Thanks, Brett.
That's great. Thanks.
Operator, can we move on to the next person on the line?
Now please welcome Roland Shu from Citigroup. Please go ahead.
Hi, good afternoon. My first question is also for this TWD 100 billion CapEx. Can you clarify, is this year's CapEx of TWD 30 included in this TWD 100 billion or not? Also, I use your long-term capital intensity target. Last time you said long-term is 3-5 years. I use this above TWD 30 billion CapEx, maybe in 2024. It implies that your revenue in 2024 will likely to exceed TWD 90 billion or even bigger, which is more than double than 2020's level. My question is, are there any challenges to you to recruit and train up enough amount of the talent to support such a fast growth for you going forward?
Okay. Let me summarize your questions, Roland. First, Roland is asking this TWD 100 billion CapEx, does this include 2021 of around TWD 30 billion? Then he's asking about if we look at the longer-term capital intensity, what does this kind of imply for 2024 and 2025 CapEx and capital intensity? Then, another part is that with this pace of growth, how do we recruit the talent to support our operations?
Hi, Roland. Yes, TWD 100 billion include this year's CapEx. We've talked about the three-year TWD 100 billion, 2021, 2022, and 2023. The capital intensity, I think, as I said earlier, you can probably do some calculation and have a feeling about the capital intensity in those three years. Longer term, we do see that the capital intensity will go back to about mid-30s level at this moment.
His second question is how do we recruit talent to support our growth?
Roland, this is a very good question.
Very good question.
The talent people recruiting is one of our top priorities in recent years. Fortunately, we have communicated with the government and get Taiwan government's strong support. They are now pushing for a new program and to actually allure the student to be in the semiconductor area, major in this area. Internally, TSMC also have a very robust system right now we just established to train all the newcomer, all the new engineer to be more, they can grow faster. Externally, we got the help from government. Internally, we do our own part also to enhance the training. That's the way that we try to meet the requirement of enough talent people inside TSMC.
Okay. Thank you, Roland.
Operator I think that was-
Yeah, but I think this actually one question.
That's two questions, Roland, okay? Operator, can we move on? There's still quite a few people. Roland, we're happy to have you get back in the queue. Let's move on to the next caller, please, for now.
Next one for question is Andrew Lu from SinoLink Securities.
Yes, thank you for taking my question. My first question is can we know what kind of % of capacity increase on eight-inch specialty foundry and 12-inch mature and 12-inch advanced for the next three years? Maybe just the average will be fine.
Okay. Andrew's first question is on the capacity increase. He wants to know in the next three years how much of capacity are we increasing on 8-inch, and then how much capacity are we increasing on the 12-inch.
Andrew, let me share with you. We don't disclose that kind of details. Basically, 80% of the CapEx will be spent in advanced technology, about 10% in advanced packaging and mask making, and another 10% in specialty technologies.
In the next three years.
Okay, Andrew, do you have a second question?
Yes, I do have the second question. The first question doesn't really answer. Can I have two more?
No, we do not comment on the capacity by 8-inch or 12-inch. I think Wendell has just said.
Okay. Second question is not related to price. Assuming the next year our rebate to the customer has been removed, what kind of % additional growth we should factor into our model? Thank you.
Okay, Andrew's second question is assuming next year that the rebates have been removed, how much will this drive additional growth in next year, and how should he factor this into his model?
Well, this kind of pricing is strictly confidential between TSMC and TSMC's customer. I don't think that we can comment on that one, whether it's a rebate, whether it's any other activities.
Okay. Thank you.
Thank you.
Thank you. Operator, can we please move on to the next caller, please?
Next one to ask questions, Sunny Lin from UBS.
Hi, good afternoon. Thank you for taking my question. My first question is also on CapEx. When you plan for CapEx for this year, next few years, do you think the equipment supply could be a potential bottleneck in terms of the additional upside that you can spend? I think several equipment makers have mentioned that based on this year's industry CapEx, they are already at extreme supply tightness, especially for EUV. Any color will be appreciated.
Okay. Sunny's first question is that with our CapEx plan, do we see or face any equipment bottlenecks in terms of securing the tools and equipment? I think part of your question is also particularly with regards to EUV.
Well, let me answer the question. In fact, when we plan TWD 100 billion CapEx, we also work closely with our supplier to prepare in advance. We don't expect, certainly we don't expect any bottleneck, whether it is EUV or not. Actually, we work closely with them.
Got it. Right. Would it be fair to assume that when you announced the TWD 100 billion CapEx for years, you already have commitment from your suppliers?
The answer is yes.
Got it. My second question is on 3 nm . Now we are just about a year before the mass production in second half of 2022. At this point, how should we think about the revenue contribution in its first year of commercial production? I think for 5 and 7 nm , they could get to high single digits of revenue or even close to 10% in first year. Just want to get your thought on that.
Okay. Sunny's second question is looking at 3 nm , and with the schedule for production, how should we think about the revenue contribution from 3 nm in its first year?
Sunny, that's too far to talk about that. We will update you later on. Now it's about two or three years away. Yeah. We do expect it's a big and long-lasting node, just like the former N5.
Okay. Thank you, Sunny.
Sure. Thank you.
Operator, can we move on to the next caller, please?
Right now, we have Laura Chen from KGI. Go ahead, please.
Yes. Hi, good afternoon. Thank you for taking my question. My first question is still same, similar to previous question about inventory days and inventory level. I think both Wendell and C.C. Mentioned already that high inventory probably will persist for a while. In what level we may start to worry about that, or what would be the checking point? Because so far we all know that the demand outlook and TSMC is, in particular in the advanced node, are quite tight. What would be the checking point that we are closely following? That's my first question. Thanks.
Okay, Laura's first question is with regards to inventory and inventory levels. She understands that demand is tight, but do we worry about inventory levels? What are the type of checking points that we would look at?
Well, let me answer that question. Yes. I did say that our customer want to secure the supply actually at this moment. That's due to some imbalance in the supply chain, they are preparing for the future also. How we are going to do to test this, what is the checking point? Actually, let me say that we are working with our customer closely. If not daily, it's at least we check in very often, and we make sure that all the demand to TSMC has been secured, and we prepare the capacity for that.
Okay, thank you. My second question is also about the mature node. I think C.C. mentioned about some specialty design, special technology for mature node. I recall you mentioned before about the CIS progress and also the gallium nitride progress. Can you give me more update or some special technology you are working now with the mature node, which may be the expansion in the next few years?
Okay. Laura's second question is looking at the mature nodes and that C.C. mentioned that our strategy is to work with customers to develop specialty technologies at those mature nodes. She's wondering if we can give a little more examples of what types of specialty technologies. Is that correct, Laura?
Yes. Thank you. Also FDSOI as well, if it's possible.
FDSOI and other areas.
Well, let me answer the last one first. We don't work on the FDSOI per se.
Okay.
We developed the specialty technology for CMOS image sensor, as I mentioned previously, and the technology continue to improve because if you look at the application of the CMOS image sensor in the smartphone, in the automotive, they are a lot. Okay. In fact, the most important one also is ultra-low power. We develop the technology to meet the requirement of the mobile world. I mean that everything is portable. Ultra-low power is very important. Gallium nitride, all those kind of specialty, we continue to work with our customer and for the future high-frequency application or the high voltage applications. We also work on the RF technology, radio frequencies. It's important because of 5G era. The RF become very important in application in the Wi-Fi communication area and a lot of them.
Following that question, do we have space or any capacity to further expand those technology here in Taiwan?
Laura's asking, will space be a constraint or limitation for the specialty?
Good question. We are working with a customer to expand our capacity whenever necessary.
Okay.
Okay, thanks.
All right. Thank you, Laura. Operator, can we move on to the next person on the line, please?
Yeah. Next on the line is Rick Hsu from Daiwa Securities.
Yeah. Hi. Good afternoon, guys, thank you so much for taking my call. I just got one question here. Regarding your second quarter guidance, the revenue is going to grow sequentially in U.S. dollar terms. If I don't remember wrong, Wendell did say that your inventory increase in Q1 was mainly because your customers pre-built inventory for 5 nm That assumes that your 5 nm contribution will also increase in the second quarter. The exchange rate, also not getting worse, right? Against the backdrop of these three positive factors, revenue increase, 5 nm increase, and favorable exchange rate, why your gross margin guidance for the second quarter is below your first quarter?
Okay. Rick's question is looking at the second quarter, and looking at the gross margin guidance, why is it basically lower than the first quarter, or a sequential decline if you use the midpoint?
Okay. Rick, let me explain to you. The sequential decline is mainly due to mix, as the contribution from N5 will increase, but it still carries a dilutive effect. Secondly, we see a slower rate of cost improvement as our fabs continue to run at a very high level of utilization, leaving less time to do cost improvement activities. Lastly, a more technical thing is the absence of a positive inventory revaluation in the quarter.
I see. Thank you so much.
Okay. Thank you, Rick. Let's move on to the next caller, please.
Next one to ask questions is Aaron Jeng from Nomura Securities.
Thanks for taking my question. This question was not asked before, so I wish to ask before end of the call. You said customers' engagement on N3 and the N5 are stronger than what you saw three months ago, which drives your TWD 100 billion CapEx for the next three years. Okay. Let me ask in this way: compared with three months ago, are you now projecting a bigger market share gain potential over the next three years at pure foundry market on your widening technology leadership at the node? This is my only question.
Okay, Aaron. Aaron's question is looking at the fact that we said customer engagement at 5 nm and 3 nm are stronger than what we saw a few months ago. Does this mean that we're going to expect to gain bigger or larger market share as a result? Is that correct, Aaron?
Yes, a bigger market share than expected three months ago.
Versus three months ago.
Okay. Let me answer that question. Certainly, as compared with three months ago, we have some progress in engaging with the customer to get their commitment to work with TSMC on 5 nm and 3 nm . Whether this one is the indication of TSMC's technology leadership, I would happily say yes. The most important thing, actually, is that we are continuing to work with customer to develop the technology they need for their product. Each customer has a different kind of preference, and we always can meet their demand.
Okay. Thank you, Aaron.
No problem. Thank you.
Okay, operator, can we move to the next caller, please?
Next one to ask question, Mehdi Hosseini, SIG.
Yes, sir. Thanks for taking question. My first question has to do with some of the comparisons that you provided during last earnings conference call. You were comparing the capital intensity and the growth prospect to the period of 2010 and 2015. In that context, my question has to do with depreciation. Back then, during the period of 2010 through 2015, depreciation increased at a growth rate of 20%. How do you see that growth rate changing in the period of 2020 and 2025? I have a follow-up.
Okay, Mehdi's first question is looking at, I think basically looking at depreciation and looking at as we enter a higher period of growth, what does our depreciation look like? Also he is asking about the depreciation growth or increase, given that we expect to grow between 10%-15% in 2020 to 2025 CAGR period. What does the depreciation growth look like this year and then beyond?
Okay. I can share with you that the depreciation this year will be around 30% higher than last year.
We are not ready to share with you the rest of the five-year period's depreciation at this moment.
Okay. Do you have a second question, Mehdi?
Sure. Yes, I have a second question. You raised your CapEx spending given the increased demand by your customer, but your revenue growth target remains the same at 10%-15%. Why aren't you raising the revenue target as you're raising the CapEx?
Okay. I think Mehdi is asking that we raised the CapEx spending, and so what is our view of the growth target 10%-15%? Why are we not raising that as well?
Mehdi, actually, if you think about this, 10%-15% five-year CAGR, it's pretty big range. From what we currently forecast, the revenue target is still within that range, maybe closer to the higher end than last time.
Okay. Thank you, Mehdi.
Thank you.
Thank you. Operator, can we move on to the next caller, please?
Now we have Randy Abrams from Credit Suisse. Go ahead, please.
Okay. Yes. Thank you for the follow-ups. The first one on the back end, that's keeping pace. Could you give an update on the spending and momentum you're seeing for the new SOIC, and then also how the CoWoS and InFO are progressing?
Randy's first question is on our advanced packaging solutions. He wants to know an update on how SOIC is progressing, as well as CoWoS, and the other solutions. Is that right, Randy?
Yeah, that's correct.
Okay.
Okay. Let me comment on the SOIC first. This is the most advanced back-end technology, I think, that we offer to our customer. It will start to small volume production in 2022, and it's also actually adopted by a very high-performance HPC applications. As for InFO and CoWoS, we continue to expand our customer portfolio, and I expect that the business from InFO and CoWoS will continue to increase in the next several years.
Okay, great. Just one quick one on that. You mentioned very high-performance applications. SOIC, in a couple years as it ramps, also, do you expect a pretty big ramp like we saw in the past for InFO, where it should be a good volume runner for TSMC? I wanted to ask a second question, just a couple clarification on gross margin in the second half. N5 will be getting more mature, I'm curious, factoring your depreciation guidance, N5 getting more mature, if your view is 50% gross margin, or if you're running very tight utilization, we may be able to stay a bit above the long-term range in second half.
Okay. A quick one. Randy wants to know, SOIC, will it be a large contributor? How large can it be in a few years' time? Also on gross margin, the gross margin outlook for second half.
We hope that SOIC going to be adopted by all those HPC applications customer. I cannot nail down what is the specific revenue number in the future. We do expect our back-end service will continue to grow, and the growth rate will be a little bit higher than the corporate average.
Okay.
Okay, great.
Randy, about second half gross margin, it's a bit too early to give details on that. You've already mentioned several things. N5 will become bigger in contribution to the revenue. It still carries a negative or dilutive effect on the margins, about 2-3 percentage point. Utilization is still pretty tight, and we continue trying to improve our cost under this very high utilization environment. That's all the things we can share with you at this moment.
Okay. If I could fit in to the.
Sorry, Randy. I'm sorry, that's two questions, because we-
Okay. Sorry. Thank you.
Thank you. Operator, in the interest of time, I think we'll take the last two callers, please.
Okay. The next one will be Gokul Hariharan from JP Morgan.
Thanks for taking my flow of questions. My first question is, how should we think about HPC in terms of the demand cadence for second and third wave? I think when smartphone was our big growth driver in the last 10 years, we had leading edge growth from processors, et cetera, but we also had a lot of other ICs in smartphones as well as other applications, which drove the second, third, fourth wave demand for any process node. Now that HPC seems to be one of the key drivers for growth, how do we think about second, third wave demand? Would it keep up with the first and second waves for N5, N3, et cetera? Should we think about TSMC will be doing more capacity conversion compared to in the past? That's my first question.
Gokul's first question is looking at HPC and looking at how HPC is also, along with smartphone, becoming the first-wave adopters of our leading-edge nodes. He's wondering then, though, for the second and additional waves of demand, how do we see HPC driving additional waves of demand, or will we convert capacity?
Let me make some comment. Actually, the HPC application includes many different sub-segments, such as CPUs, GPU, networking, FPGA, AI accelerator, video gaming, et cetera. Each one will have their own migration path, and product life cycle also. We expect to see HPC not only in the first wave, but in additional waves of demand to support our leading node in the future, actually. Did that answer your questions?
Okay. That's clear.
Do you have a second?
Second question. Yeah, just wanted to follow up on any thoughts from TSMC on the Arizona fab capacity. I think you already announced 20,000 of wafer per month of 5 nm coming up in 2024. What has been your discussion with customers regarding any potential upside to this capacity? Are customers asking for more capacity there? Do you feel that right now this seems more like a N-1 cadence because 5 nm started in Taiwan in 2020 already? Do we feel that we will move to a more shorter or a quicker cadence for leading edge in, let's say, Arizona or U.S. capacity? Just wanted to understand how TSMC is thinking about this right now.
Okay. Gokul's second question is on our U.S. manufacturing and our fab in Arizona. He wants to know that, are customers asking for more capacity or more production? We start with N5. I guess your question is, what about the future plans for bringing additional technologies there and the cadence?
Okay. We are executing our plan in Arizona according to the schedule, construction will start this year. Phase I production, as you said, you are starting 2024 with a 20,000 wafer per month. This is a 5 nm technology. In fact, we have acquired a large piece of land in Arizona to provide flexibility. The further expansion is possible, but we will ramp up to phase I first, Based on the operation efficiency and cost economics, Also the customer's demand to decide what the next steps we are going to do. Our customer welcome us to build a capacity in the U.S., Our fab in Arizona will be available to support all our customers from around the world. Just like all the TSMC's fab, no matter where they are and no matter where they're located.
Okay. Thank you. Operator, in the interest of time then, can we let the last caller ask their questions, please?
Yes. The last one to ask question, Sebastian Hou, CLSA.
Yeah. Thank you. I only have one question. Just follow up on C.C.'s comments earlier, that C.C. mentioned that TSMC has been working closely with customer to analyze the gap between capacity and demand on the trailing edge nodes. I'm wondering if you could share some color with us, if we exclude the overbooking portion, and based on your best analysis, does the demand still significantly exceed supply? How big is the gap, if you have any rough number that can be shared? Furthermore, based on the CapEx you and your peers are investing in the capacity expansion lead time, when do you think the tightness can be eased or the whole shortage situation can be removed? That's the only question I have. Thank you.
Let me try to summarize your question, Sebastian. You're asking on the mature nodes, the fact that TSMC works with our customers very closely, but also in looking at the supply-demand of those older nodes. With the additional capacity added, when and will we eventually see an easing of the supply tightness at the mature nodes? Is that correct, Sebastian?
Yes. Also if we exclude the overbooking part-
All right.
Your best estimate, whether the demand still significantly exceeds supply right now. Thank you.
All right. To be frank with you, as I said, we work with the customer closely. The overbooking is not in our calculation. Although we did not exclude this possibility. We do the very detailed analysis internally, and as I said, work with customer closely. We prepare the mature nodes capacity for them. However, building a fab from a green fab start and also to install the capacity, it won't be available until 2023. This year and next year, I still expect the capacity tightness will continue and probably also next year. 2023, I hope that we can offer more capacity to support to our customers. At that time, we start to see the supply chains tightness will release a little bit.
Okay, Sebastian, does that answer your question?
Yes. Is it fair for us to conclude that in the next 18 months, it is very safe to assume that we will still be in the supply tighter situation? Is that right?
For our customers, we are working with them, let me say that. It's still very tight. Yes, you are right. Okay?
Okay. Thank you.
Thank you, Sebastian. 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 four hours from now, and the transcript will become available 24 hours from now, both of which are going to be available through TSMC's website at www.tsmc.com. Thank you for joining us today. We hope everyone continues to stay healthy and safe, and we hope you will join us again next quarter. Goodbye and have a good day.