ab 23 March 2026 Global Research Asia Semiconductors Robust Cloud AI outlook in 2026 across GPUs and ASICs Equities Asia Semiconductors Sunny Lin Analyst sunny.lin@ubs.com +886-2-8722 7346 Ryan Sun Associate Analyst ryan-za.sun@ubs.com +886-2-8722 7267 This report has been prepared by UBS Securities Pte. Ltd., Taipei Branch. ANALYST CERTIFICATION AND REQUIRED DISCLOSURES, including information on the Quantitative Research Review published by UBS, begin on page 157. UBS does and seeks to do business with com ...

27 Mar 2026 10:38:33 ET │ 24 pages TSMC (2330.TW) AI megacycle accelerates CITI'S TAKE We further upgrade our earnings projections for TSMC and introduce our 2028 estimate. The strong AI demand is not limited to GPU/ASICs but extends to the broader ecosystem, including server CPUs, networking chips, and CPOs. The tightening supply environment also supports continued wafer price increases, reinforcing TSMC's pricing power and margin resilience despite rising depreciation from higher capex and also inflationa ...

J P M O R G A N Asia Pacific Equity Research 24 March 2026 (852) 2800-8564 gokul.hariharan@jpmorgan.com J.P. Morgan Securities (Asia Pacific) Limited/ J.P. Morgan Broking (Hong Kong) Limited Jennifer Hsieh (886-2) 2725-9868 jennifer.hsieh@jpmorgan.com J.P. Morgan Securities (Taiwan) Limited In this report, we address the most frequent questions on TSMC from investors: Overweight 2330.TW, 2330 TT Price (23 Mar 26):NT$1810.0 Price Target (Dec-26):NT$2250.0 Technology and Telecoms Gokul Hariharan AC TSMC Ten q ...

SEMICON China 感受：AI 引爆万亿美元市场，先进封 装与光互连成焦点 20260329 摘要 半导体行业进入加速增长期，预计 2026 年全球市场规模达 9,750 亿美 元，存储市场增速近 190%为核心驱动。 算力瓶颈由芯片转向存储与连接，驱动 CPO、全光互联及先进封装 （CoWoS/PLP）需求爆发。 先进工艺产能持续紧缺，国内 2nm/3nm 及主流工艺（28-65nm）需 求旺盛，中国晶圆产能占比 2030 年有望达 32%。 投资优先级排序：原材料 > 存储 > 设备 > 代工 > 芯片设计；后道设 备增速预期将超越前道，具备估值修复潜力。 AI 眼镜预计 2026-2027 年大规模爆发，技术路径由无显示向硅基 OLED/Micro-LED 显示升级，成为核心 Agent 终端。 地缘政治导致估值倍数下降及成本上涨，但对全球 AI 总需求影响有限 （中东占比 <5%），供应链风险可控。 Q&A 您在参加完近期一系列 SEMICON 活动（包括 SEMICON China）后，对全 球半导体行业的整体市场情况和发展趋势有何观察与判断？ 参加完包括 SEMICON China 在内的 ...

证券研究报告 | 行业策略 gszqdatemark 2026 03 25 年 月 日 电子 2026 年度策略：算力闭环加速，重塑价值新锚点 存储：AI 存力大时代已至。本轮存储有何不同？我们认为，本轮存储行情 已经跳出此前的周期循环，进入人工智能重塑底层逻辑的成长大时代。从 存储板块业绩表现来看，台股存储业绩不断提升，其中南亚科因 DDR4 涨 价迅猛表现尤为亮眼，业绩层面进一步确认景气度，存储板块的业绩确定 性也不断提升。从国内公司来看，我们认为在供需紧缺背景下，能拿到稳 定、高品质颗粒供给的模组厂商有望充分受益于 AI 时代供需紧缺下存储 升级+涨价+国产化三重β，业绩弹性显著、估值提升空间广阔。存储是有 业绩（有安全边际）、有长逻辑（AI+国产化）的板块，我们看好存储涨价 周期持续，重视 AI 存力大时代投资机遇。 PCB：AI 硬件架构与性能升级推动 PCB 价值量加速通胀。在产能紧缺背 景下，行业进入新一轮 AI 资本开支周期，各大厂商纷纷着手加速推进产 能建设，把握需求热潮。进入 2026 年我们预计 PCB 行业高景气度仍将持 续，一方面，经历 2025 年 GB200 服务器的产能及良率爬 ...

Stacking in packaging such as CoWoS, SoIC, and HBM is enabling faster chip performance by increasing the bandwidth of chip interconnections and is a key enabler of AI chips. It is also being extended to frontend processes to drive upcoming node migrations. Such technologies could create big opportunities for packaging equipment and material companies. This Blackbook discusses technologies, competition, and major beneficiaries for this multi-year trend. See Disclosure Appendix of this report for important di ...

Summary of Key Points from the Conference Call Industry Overview - **Industry Focus**: Greater China Semiconductors, specifically the semiconductor industry in the Asia Pacific region [1] - **Investment Outlook**: The semiconductor industry is viewed as attractive, with a focus on AI infrastructure spending and TSMC's capital expenditures (capex) [4] Core Insights - **AI Infrastructure Spending**: There is a significant increase in spending on AI infrastructure, which is expected to drive demand for semiconductors [1][6] - **TSMC's Capex**: TSMC is leading the industry in capital expenditures, which is anticipated to drop significantly due to robust revenue growth [30][28] - **Long-term Demand Drivers**: Key companies identified include TSMC, SMIC, MediaTek, and others, with a focus on AI-related semiconductor demand [7] - **Tech Inflation**: Rising costs in wafer, OSAT, and memory are creating margin pressures for chip designers in 2026 [7] - **AI Cannibalization**: AI is expected to replace some human jobs, leading to a shift in semiconductor supply chains prioritizing AI semiconductors over non-AI semiconductors [7] Financial Metrics and Valuation - **Valuation Comparison**: A detailed comparison of various semiconductor companies, including TSMC, UMC, and SMIC, highlighting P/E ratios, EPS growth, and market cap [8][9] - **TSMC Financials**: As of March 18, 2026, TSMC's stock price is 1,905.0 TWD with a target price of 2,288.0 TWD, indicating a 20% upside potential [8] - **Memory Sector**: Companies like Winbond and GigaDevice are highlighted as top picks in the memory sector, with significant upside potential [58] Market Dynamics - **Broader Semiconductor Cycle**: The logic semiconductor foundry utilization is reported at 70-80% in the first half of 2026, indicating a recovery phase [12] - **AI vs. Non-AI Growth**: Excluding NVIDIA's AI GPU revenue, non-AI semiconductor growth was slow at only 10% year-over-year in 2024 [14] - **Inventory Trends**: A decrease in inventory days historically correlates with an increase in semiconductor stock indices, with a noted decline in inventory days in Q3 2025 [16] TSMC's Strategic Plans - **Fab Expansion Plans**: TSMC has outlined a comprehensive fab expansion plan, including new facilities in Tainan, Arizona, and Japan, with a focus on advanced technologies [23][25] - **Leading-edge Wafer Demand**: TSMC's leading-edge wafer demand is expected to significantly contribute to its revenue, with projections indicating that AI semiconductors could account for over 30% of TSMC's revenue in 2026 [49][48] Additional Insights - **Cloud AI Impact**: The global semiconductor market size is projected to reach $1 trillion by 2030, driven by cloud AI demand, with a total addressable market (TAM) for cloud AI semiconductors estimated at $235 billion by 2025 [93] - **NAND and NOR Flash Supply**: There is an expected shortage in NAND and NOR flash memory due to increased AI storage demands [66][72] This summary encapsulates the key points discussed in the conference call, providing insights into the semiconductor industry's current state, future outlook, and specific company strategies, particularly focusing on TSMC and its role in the AI semiconductor market.

Summary of Key Takeaways from the 2026 Tech Tour Industry Overview - The 2026 Tech Tour involved meetings with 25 companies across Taiwan, Korea, and Japan, focusing on the semiconductor and hardware sectors, particularly memory and AI technologies [2][3]. Core Insights - **AI Demand**: There is a strong and consistent demand for AI across various sectors including foundry, memory, and semiconductor equipment. This trend is expected to continue into 2027 with significant capacity build-up in wafer production and advanced packaging technologies [2][3]. - **Memory Cycle**: The memory market is experiencing a stronger and longer cycle, with potential price increases for certain products expected to double quarter-over-quarter in Q2 2026. The consensus anticipates continued price hikes in the second half of 2026, although at a potentially slower rate [3]. - **Supply Chain Resilience**: Major companies like TSMC, Samsung, and SK Hynix have improved their supply chain resilience post-Ukraine conflict, with no current disruptions reported despite rising energy prices [3][18]. Company-Specific Insights TSMC - **Capacity Expansion**: TSMC's second fab in Arizona is set for volume production in 2H27, with plans for additional fabs contingent on business needs. The cost of production in the US is significantly higher than in Taiwan [17][18]. - **AI Capex**: TSMC's capital expenditure decisions are driven by long-term demand for AI, with a focus on maintaining pricing discipline and not overcharging during up-cycles [18][20]. - **Smartphone Market**: TSMC anticipates a decline in global smartphone shipments but expects flagship models to perform better, with a conservative outlook on overall demand [20]. MediaTek - **Revenue Growth**: MediaTek expects a 17% CAGR for non-AI revenue from 2024 to 2029, supported by strong demand for its flagship SoCs and a focus on optimizing chip design [19][20]. - **Competitive Position**: MediaTek sees advantages in IP capability and supply chain management compared to ASIC competitors, with a focus on maintaining gross margins despite rising costs [21][23]. Samsung - **Outperform Rating**: Samsung is rated as outperform with a price target of KRW 140,000, reflecting strong growth expectations in the semiconductor sector [13][14]. SK Hynix - **Outperform Rating**: SK Hynix is also rated as outperform with a price target of KRW 750,000, indicating confidence in its market position [14]. Micron - **Outperform Rating**: Micron is rated as outperform with a price target of USD 330.00, reflecting positive market sentiment [14]. Novatek - **Market-Perform Rating**: Novatek is rated as market-perform with a price target of NT$390.00, indicating a cautious outlook [10]. Delta Electronics - **Growth in AI Power Supply**: Delta expects significant growth in its power supply unit revenue, driven by AI demand, with projections of revenue rising to USD 6-7 billion this year [26]. Additional Insights - **Memory Price Trends**: Companies like ADATA are experiencing rapid increases in memory prices, with expectations of continued upward trends throughout 2026. ADATA is pursuing a bullish inventory strategy to capitalize on anticipated shortages [30][31]. - **Market Dynamics**: The competitive landscape is shifting as CSPs (Cloud Service Providers) move from GPU to ASIC solutions, with major players like AWS and Google leading this transition [25]. Conclusion The semiconductor and memory sectors are poised for growth driven by AI demand and strategic capacity expansions. Companies are focusing on maintaining supply chain resilience and optimizing their product offerings to navigate the evolving market dynamics.

Core Viewpoint - The AI industry is entering a "chip shortage era," which is expected to last until at least 2027, highlighting the importance of supply chain management alongside technological capabilities [37]. Group 1: AI Chip Demand and Supply - Anthropic generated an additional $6 billion in annual recurring revenue in just one month, primarily through its AI programming tool, Claude Code [4]. - The demand for AI chips, particularly those using TSMC's 3nm process, is expected to consume nearly 60% of TSMC's 3nm capacity this year, rising to 86% next year, squeezing out traditional mobile chip customers [11][12]. - TSMC's 3nm capacity is under pressure as major AI chip manufacturers like NVIDIA, AMD, Google, and AWS are all vying for this advanced process technology [8][9]. Group 2: Supply Chain Dynamics - NVIDIA has strategically locked in supplies of logic wafers and memory components, positioning itself as a major beneficiary in the ongoing supply chain competition [33][34]. - The shift in focus from power supply to silicon wafer availability indicates that while data centers and power supply have expanded, the chip supply has not kept pace [28][32]. - The production of high-bandwidth memory (HBM) is also facing challenges, as HBM consumes 3 to 4 times the wafer capacity compared to standard DDR memory, exacerbating the supply constraints [17][22]. Group 3: Market Implications - The competition for chip resources is leading to a "reallocation of bits," where AI applications are prioritized over consumer electronics, potentially resulting in higher prices and slower product cycles for smartphones and PCs [23][38]. - The pricing dynamics for HBM are shifting, with DDR memory prices rising, which may reduce the incentive for manufacturers to shift production capacity from DDR to HBM [22]. - The AI industry's rapid growth is outpacing hardware supply capabilities, leading to a scenario where access to chips becomes a critical factor for success in AI deployment [38]. Group 4: Future Outlook - TSMC's role has become increasingly pivotal, as its capacity allocation decisions directly impact the competitiveness of major players like NVIDIA, Google, and AMD [38]. - The ongoing competition for silicon resources may lead to a significant transformation in the AI landscape, where the ability to secure chips becomes more crucial than algorithmic advancements [38]. - The consumer electronics sector may face significant challenges as AI demand continues to dominate chip production, potentially leading to a decline in smartphone demand and increased costs for consumers [38].

Summary of Key Points from the Investor Presentation on Global and China AI GPU Industry Industry Overview - The presentation focuses on the **AI GPU industry**, particularly the competitive landscape between **China** and the **US** in AI semiconductor production and demand [1][4][98]. Core Insights - **Long-term Demand Drivers**: - The AI semiconductor market is expected to grow significantly, with **cloud AI** being a major growth driver, potentially reaching a total addressable market (TAM) of **US$235 billion by 2025** [12][18]. - **China's AI chip TAM** is projected to grow to **US$67 billion by 2030**, with self-sufficiency expected to reach **76%** [109][111]. - **Market Dynamics**: - The **cloud capital expenditure (capex)** is robust, with estimates of nearly **US$632 billion** in 2026 from the top 10 global cloud service providers (CSPs) [12]. - **Nvidia's CEO** estimates global cloud capex could reach **US$1 trillion by 2028**, including sovereign AI [14]. - **Supply Chain Challenges**: - The semiconductor supply chain is prioritizing AI semiconductors over non-AI semiconductors, leading to potential shortages in other areas [10]. - **Tech inflation** is expected to impact demand for tech products, with rising costs for wafers, OSAT, and memory creating margin pressures for chip designers in 2026 [10]. - **China's AI GPU Development**: - The presentation raises critical questions about whether China can supply competitive AI GPUs at scale and the potential size of the domestic AI GPU market [100]. - The **local AI chip market** is expected to surpass US chips in value by **2027**, with **Huawei** projected to maintain over **50%** market share in local AI chips from 2026 to 2030 [148] [150]. Important Data Points - **NVIDIA's Production Estimates**: - TSMC is expected to produce **7-8 million GPU chips in 2025**, with NVIDIA's server rack chip consumption projected to reach **60,000-70,000** units [64][66]. - **AI Semiconductor Consumption**: - AI computing wafer consumption could reach **US$26 billion in 2026**, with NVIDIA accounting for the majority of this demand [56]. - **TSMC's Capacity Expansion**: - TSMC plans to expand its CoWoS capacity to **125k wafers per month by 2026** due to strong AI demand [47][52]. Other Notable Insights - **Geopolitical Risks**: - The presentation discusses potential geopolitical risks affecting the supply chain, including restrictions on foreign foundries and export controls on critical technologies [154]. - **Inference Economics**: - Domestic chips in China are reported to have lower total cost of ownership (TCO) and comparable costs per token for AI inference compared to NVIDIA's processors [158]. - **Strategic Responses**: - Recommendations for overcoming wafer process constraints include packaging more dies into a single chip and expanding manufacturing capacity [130]. This summary encapsulates the critical insights and data points from the investor presentation, highlighting the competitive landscape and future outlook for the AI GPU industry, particularly in the context of China and its efforts to close the technological gap with the US.