Time: (Beijing Time) January 15, 2026, 2:00 PM Participants: CEO C.C. Wei, CFO Wendell Huang, Director of Corporate Relations Jeff Su [Transcript translated and compiled by Guohai Securities Overseas Team, does not represent the views or stance of Guohai Securities. If there are discrepancies between this translation and the official conference call document, please refer to the official document.]

Summary of Q4 2025 Industry & Operations In 2025, AI-related demand demonstrated robust growth throughout the year, while the non-AI market bottomed out and showed a modest recovery by year-end. The "Foundry 2.0" industry grew 16% year-over-year. For 2026, the "Foundry 2.0" industry is projected to grow approximately 14% year-over-year. AI accelerator business accounted for 15% of TSMC's total revenue in 2025. The revenue growth forecast for the AI accelerator business has been raised, with an expected Compound Annual Growth Rate (CAGR) in the mid-to-high 50% range for the five-year period from 2024 to 2029. TSMC's overall revenue CAGR (in USD) is anticipated to be close to 25% from 2024 to 2029. Arizona, USA: Phase two construction is complete, with tool installation expected in 2026. Due to strong customer demand, production schedules are being accelerated, with mass production targeted for the second half of 2027. Permits are currently being sought for a fourth fab, which will also be the first advanced packaging facility. Wafer shipments (12-inch equivalent): 3.961 million units (QoQ -3.0%, YoY +15.9%).

Financial Performance Net revenue: NT$10,460.9 billion / US$33.73 billion (NT$ value QoQ +5.7%, YoY +20.5%), compared to guidance of US$32.2-33.4 billion and a consensus estimate of NT$10,315.1 billion. Net income attributable to shareholders: NT$505.74 billion (QoQ +11.8%, YoY +35.0%). Gross margin: 62.3% (QoQ +2.8 ppts, YoY +3.3 ppts), compared to guidance of 59%-61% and a consensus of 60.6%, primarily driven by favorable currency exchange rates, optimized capacity utilization, and cost improvement initiatives. Net profit margin attributable to shareholders: 48.3% (QoQ +2.6 ppts, YoY +5.2 ppts). EPS: NT$19.50 (QoQ +11.8%, YoY +35.0%), compared to a consensus estimate of NT$18.12.

Revenue Breakdown & Capital Expenditure Capital Expenditure (CapEx): NT$356.91 billion (QoQ +24.2%, YoY -1.4%).

Q1 2026 Guidance Assumptions: Exchange rate of NT$31.6 to US$1. Revenue: US$34.6-35.8 billion (midpoint QoQ +4%, YoY +38%), compared to a consensus estimate of NT$1,050.77 billion / US$33.22 billion. Gross margin: 63%-65%, compared to a consensus of 59.6%. Operating profit margin: 54%-56%.

Full-Year 2026 Guidance Revenue: Approaching 30% year-over-year growth in USD terms, compared to a consensus estimate of US$150.44 billion, implying ~25% growth. 2024-2029 five-year revenue CAGR of 25%. CapEx: US$52-56 billion, compared to a consensus of US$46.04 billion, with 70%-80% allocated to advanced processes, 10% to specialty technologies, and 10%-20% to advanced packaging, testing, mask making, and others. Depreciation expenses: Expected to increase over 10% year-over-year. Gross margin: Long-term target remains above 56%. Outlook for full-year 2026: (1) Overall capacity utilization is expected to grow moderately. (2) Continue to earn our value (price increases). (3) Improve production efficiency. (4) Continuously optimize capacity across nodes, including flexibly adjusting capacity between N7, N5, and N3 nodes to enhance profitability. (5) Overseas fab capacity ramp-up is expected to dilute gross margin by approximately 2-3% initially, potentially expanding to 3-4% in later stages. (6) Currency exchange impact.

Q&A Session Q: TSMC is significantly increasing CapEx, but there are concerns about an AI bubble and risks. Could you share feedback from your customers and their customers, and how long you believe this cycle will last? A: You are essentially asking if AI demand is real. Of course, I am nervous too. We need to invest about US$52-56 billion. If we mishandle this, it would be a significant disaster for TSMC. So, over the past three months, I spent considerable time communicating with my customers and their customers. Their responses were very reassuring. They showed me evidence that AI is genuinely helping their business grow and stated their financial returns are substantial. I further verified their financial health, and they are all very financially sound. I specifically inquired about application scenarios. For instance, they told me a certain CSP's social media software is benefiting, leading to continuous user growth. I believe this, and based on our own experience with AI applications, we've improved our fab efficiency. That's why, even in this high-cost environment, our gross margin is satisfactory. In summary, I believe AI is not only real but is increasingly integrated into daily life. We are confident that US AI customers have substantial demand for US fabs. Therefore, we must accelerate our fab expansion in Arizona. We are also significantly boosting capacity in Taiwan. In the US, we are working hard to expedite progress, which is going smoothly with government support, though we still need to fulfill all permit requirements. So, both in Taiwan and Arizona, we are accelerating capacity expansion to meet AI demand. Capacity is extremely tight. We are constantly working to close the gap. This year and next, we must double our efforts to narrow it.

Q: How does TSMC balance AI with other factors like power supply? A: My primary concern is Taiwan's power supply. We need ample, unrestricted power for expansion. Regarding global AI data center construction, a customer's customer informed me they started planning power supply five or six years ago. Their current message is that TSMC's chip supply is the bottleneck, urging me to focus there for now. However, we are monitoring global power supply, especially in the US, including related suppliers like turbines, nuclear plants, as well as rack and cooling system availability. Currently, things seem on track. Therefore, we must focus on closing the chip supply-demand gap.

Q: What was advanced packaging's revenue contribution in 2025? What are the 2026 advanced packaging CapEx priorities? SoIC? Or panel-level packaging? A: Advanced packaging contributed nearly 10% to revenue in 2025, approximately 8%. For 2026, we expect it to be slightly above 10%. We anticipate its growth rate will outpace the company's overall growth over the next five years. Regarding CapEx, last year it was about 10%, slightly under. Now, we estimate advanced packaging, mask making, and others will account for 10%-20%. So, you can see the investment is higher. We are investing in advanced packaging technologies our customers require, including those you mentioned.

Q: What is the outlook for non-AI demand, especially given rising memory costs? How will this affect PC and smartphone shipments? And what about markets like networking and general-purpose servers? A: While termed 'non-AI', they are closely related to AI. Network processors still need AI data for scaling. Growth remains strong. For PCs and smartphones, we expect higher memory prices to lead to very limited shipment growth. However, for TSMC, we haven't observed changes in customer behavior. We supply mostly high-end smartphones, which are less sensitive to memory prices, so demand remains robust. We are still working to close the supply gap for these customers.

Q: Is there a risk of TSMC losing market share to foundry competitors like Intel? A: My answer is "no". Money alone isn't sufficient for competition nowadays. I wish investors in Intel would also invest in TSMC. The fundamental point is that technology is extremely complex. Designing with a very advanced technology takes two to three years to master fully. After preparation, product design takes another one to two years to scale. So, while we face a strong competitor, firstly, it takes time. Secondly, we do not underestimate their progress. But are we afraid? For over thirty years, we have competed. So, no, we are confident in maintaining business growth as expected.

Q: What is TSMC's strategy for mature nodes? Reports suggest TSMC is exiting 8-inch mature node business and shifting capacity to advanced packaging. Is this true? If so, why? A: We have indeed scaled down 8-inch and 6-inch capacity. But I assure you, we fully support all our customers. We communicate with them to utilize resources more flexibly and efficiently. If a customer's business is healthy, even in smaller 8-inch wafer areas, we will continue our support.

Q: Impact of rising memory prices and weak demand on 2027? A: For TSMC, no impact. As mentioned, our customers focus on high-end smartphones or PCs, which are less price-sensitive. They continue to provide optimistic forecasts for this year and next.

Q: Supply seems tight in 2026. With increased CapEx, do you expect a better supply-demand balance in 2027? Also, could engineering resources, especially fab engineers, be a constraint for expansion in Taiwan and the US? A: Building a new fab takes two to three years. So, even with the US$52-56 billion CapEx, contribution this year is nearly zero, and minimal in 2027. We are really focusing on 2028 and 2029 supply, hoping to close the gap then. For 2026 and 2027, we focus on short-term production efficiency gains, which have been consistently improving. In 2028 and 2029, we will significantly increase CapEx if AI demand trends continue as expected.

Q: In 2025, TSMC's wafer ASP increased about 20% for the second consecutive year, including both pricing and product mix factors. Is this the new normal? A: Each new node is priced higher than the previous generation, so the blended ASP increases. But I believe you are asking about pricing's contribution to profitability. Profitability is affected by six factors; price is just one. We continuously work to earn our value (price increases). However, in recent years, price increases have largely offset inflation in equipment, materials, labor, etc. Other factors contribute more to higher profitability: high capacity utilization due to strong demand, supported by rigorous capacity planning; excellent manufacturing capabilities and ongoing production efficiency improvements; and continuous optimization of capacity across nodes, including upgrading N5 to N3 capacity. These efforts sustain healthy, sustainable returns, enabling continued investment to support customer growth.

Q: Historically, revenue from a new node declined a few years after introduction. Now, even nodes four or five years old generate high revenue. What are the financial implications and competitive landscape impact? A: Observing semiconductor products, the trend requires maintaining low power consumption alongside high-speed performance. TSMC's technological differentiation is increasingly strong, offering both advantages. Consequently, we see continuous demand waves from advanced node customers, sustaining long-term demand. This is the key difference. It requires technological leadership, which must be improved annually. This is why customers continue partnering with TSMC to keep their products competitive.

Q: What is the projected revenue contribution from 2nm in 2026? A: Demand for the 2nm process is stronger than it was for 3nm at a comparable stage. However, discussing revenue contribution percentage at new node launch is less meaningful now because the company's overall revenue base is much larger. In absolute terms, N2 volume will be larger, but the percentage contribution is not particularly indicative.

Q: What specific value does N2 provide that motivates customer adoption? A: Because products demand low power and high speed, and our technology delivers this value. Furthermore, we improve every year. That is the real value. If transistor cost increases, we see that performance increases proportionally. The main challenge, if it can be called that, is the supply-demand gap, which we are striving to close.