Group 1 - The core viewpoint of the article highlights an unprecedented capital expenditure surge among global cloud service providers (CSPs) driven by the AI arms race, with total spending expected to exceed $520 billion by 2026 [1][2] - Major CSPs including Google, Amazon, Meta, Microsoft, Oracle, Tencent, Alibaba, and Baidu are projected to collectively spend over $420 billion by 2025, marking a staggering 61% increase compared to previous years [1][2] - The capital expenditure is primarily directed towards three areas: procurement of NVIDIA GPU solutions, expansion of data center infrastructure, and acceleration of self-developed AI ASIC chips [2] Group 2 - The AI server industry chain in the A-share market is expected to be the most direct beneficiary of the CSP capital expenditure increase, with the global AI computing server market projected to grow from approximately $39.97 billion in 2024 to $113.96 billion by 2031, reflecting a compound annual growth rate (CAGR) of 16.4% [3] - High-performance AI server shipments are forecasted to increase by 21% and 39% for 2025 and 2026, respectively, while inference AI server shipments are expected to rise by 3% and 5% during the same period [3] Group 3 - The semiconductor sector is set to benefit from the CSP capital expenditure growth, focusing on the GPU supply chain and domestic alternatives, with NVIDIA holding an 86% market share in the AI GPU market by 2025 [5][6] - The demand for liquid cooling technology is surging as traditional air cooling fails to meet the thermal requirements of high-power AI servers, with leading liquid cooling suppliers expected to capture 5% and 10% of the global liquid cooling market by 2027 and 2030, respectively [8] Group 4 - ASIC chips are emerging as a critical avenue for CSPs to break NVIDIA's dominance, with global AI ASIC chip sales projected to approach 8 million units by 2027 [9] - The urgency for domestic alternatives in the semiconductor field is increasing due to U.S. export controls on EDA tools, which has created a pressing need for local GPU and AI ASIC production [7] Group 5 - The investment landscape is characterized by a clash between traditional value investors ("old investors") and younger tech-focused investors ("young investors"), with the current capital expenditure trend favoring the latter's preferences for AI and semiconductor sectors [10][12] - The article suggests that future investment opportunities may lie in identifying quality companies that can benefit from the AI wave while maintaining reasonable valuations and solid performance [13]

Summary of Key Points from the Conference Call Industry and Company Overview - **Industry**: AI Supply Chain, specifically focusing on semiconductor manufacturing and AI ASICs - **Key Companies**: TSMC (Taiwan Semiconductor Manufacturing Company), NVIDIA, Meta, Google, AWS, Alchip, MediaTek, GUC (Global Unichip Corp) Core Insights and Arguments 1. **CoWoS Capacity and Demand**: - TSMC's 2026 CoWoS capacity plan is currently under-supplying based on anticipated demand from key customers, particularly NVIDIA [2][10] - NVIDIA's RTX Pro 6000 forecast remains strong, indicating robust inference demand in China [1][2] - TSMC's current offering of 590k CoWoS-L wafers in 2026 is projected to be 20% below NVIDIA's demand [2] 2. **NVIDIA's AI ASIC Developments**: - The Rubin CPX GPU is expected to adopt TSMC's CoWoS-S packaging, with a performance of 30 PFLOPS and 128GB of GDDR7 memory [3][39] - NVIDIA's AI capacity costs are estimated at $50-60 billion per GW, with $35-40 billion allocated to NVIDIA, suggesting potential revenues of $350-400 billion starting in 2H26 [2][11] 3. **AI ASIC Volume Forecast**: - AI ASIC volumes could reach approximately 5.7 million units in 2026 and 8 million units in 2027, with Google and AWS being the primary players [47] - MediaTek's chances of winning the Meta MTIA project are diminishing due to strong competition from Broadcom and Marvell [57][58] 4. **Market Dynamics and Competitor Analysis**: - GUC is expected to benefit from increased demand for Google Axion CPUs, potentially contributing around $250 million to its revenue in 2026 [59] - Alchip is projected to generate approximately $1.8 billion in revenue from Trainium3 in 2026, despite a back-loaded shipment schedule [56] 5. **Capacity Expansion and Future Projections**: - TSMC's clean room space is ready for expansion to 110-120k if demand is confirmed, with a revised CoWoS capacity assumption of 100kwpm by 2026 [2][12] - The overall demand for CoWoS is expected to grow significantly, with NVIDIA's demand forecasted to increase by 61% in 2026 [20] Additional Important Insights - **NVIDIA's Strategic Positioning**: - NVIDIA's Rubin architecture is designed to optimize inference performance, potentially delivering a 30x to 50x return on investment [39] - The demand for NVIDIA's Blackwell GPU, including RTX Pro 6000, is strong among Chinese customers, with expected sales of 1.5 million to 2 million units in 2H25 [14] - **Competitive Landscape**: - SMIC is aggressively expanding its 7nm node capacity to meet domestic GPU and AI ASIC demand in China [14] - The competition in the AI ASIC market is intensifying, with various players vying for market share and technological advancements [57][58] This summary encapsulates the critical insights and projections regarding the AI supply chain, particularly focusing on TSMC and NVIDIA's roles within the semiconductor industry.

V i e w p o i n t | 16 Sep 2025 12:20:59 ET │ 10 pages Nicholas Lai +886-2-8726-9093 nicholas.lai@citi.com See Appendix A-1 for Analyst Certification, Important Disclosures and Research Analyst Affiliations. Taiwan Semiconductors and Electronic Components & Equipment US marketing feedback - Structural AI growth continues, prefer NVIDIA supply chain on smooth product cycle transition CITI'S TAKE We were in the US meeting investors over the past two weeks. We note that US investors appear more positive on AI' ...

Summary of Key Points from the Conference Call Industry Overview - The conference focused on the semiconductor industry, particularly advancements in AI chips, heterogeneous integration, advanced packaging, and optical interconnect technologies, reflecting the growing importance of these areas in the market [2][3][20]. Core Findings 1. **TSMC's Capacity Expansion**: TSMC is expected to expand its CoWoS capacity to 100kwpm by the end of 2026, up from 70kwpm at the end of 2025, driven by robust demand for Cloud AI GPUs and ASICs [3]. 2. **AI Computing Demand**: AI computing requirements have surged by 10x in the past year, necessitating advancements in chip scaling, memory, and interconnect technologies [3]. 3. **3.5D Advanced Packaging**: The event highlighted significant discussions around 3.5D advanced packaging, which is anticipated to become mainstream for high-performance computing, improving cost structures and product design speeds [3]. 4. **Heterogeneous Integration**: The trend towards co-packaged optics (CPO) is gaining traction, with expectations for power consumption to be optimized by 2028, allowing for the replacement of copper in AI server integrations [3]. 5. **Testing Innovations**: The complexity of die and package designs is increasing the need for more rigorous testing at the wafer/die level to identify yield issues early [3]. Stock Recommendations - Top stock picks in the Greater China semiconductor sector include TSMC, ASE, MediaTek, Alchip, and Aspeed, all rated as "Buy" due to their structural AI opportunities [4]. Additional Insights - **Optical Interconnects**: Nvidia's advancements in networking infrastructure, particularly with its Spectrum-X CPO solution, promise significant power savings and improved signal integrity [12]. - **AI Data Center Power Consumption**: The power consumption of AI data centers is projected to rise dramatically, with examples like Meta's Hyperion data center expected to consume 2GW by 2030 [16]. - **Challenges in Advanced Packaging**: The industry faces challenges in transitioning to panel-level packaging and CoWoP technologies, which require overcoming technical hurdles related to system design and materials [30][39]. Emerging Technologies - **Silicon Photonics**: TSMC's COUPE platform aims to enhance integration of optics and electrical signaling, addressing bandwidth bottlenecks in computing performance [12]. - **GaN Technology**: GaN is highlighted for its efficiency and potential in powering AI applications, with Texas Instruments and Infineon leading developments in this area [36][38]. Conclusion - The semiconductor industry is at a pivotal point, driven by AI advancements and the need for innovative packaging and integration solutions. Companies like TSMC, Nvidia, and MediaTek are positioned to capitalize on these trends, while challenges in testing and power consumption remain critical areas for development [3][4][16][20].

Summary of Conference Call Notes Industry Overview - The focus is on the **Taiwan ABF (Ajinomoto Build-up Film) substrate market**, particularly in relation to **ASIC (Application-Specific Integrated Circuit) AI servers** and their suppliers, including **Unimicron**, **NYPCB**, and **Ibiden**. Key Points and Arguments Market Demand and Growth Projections - The **2026/27 AI server ABF Total Addressable Market (TAM)** has been raised by **35%/55%**, reflecting a strong demand outlook for ASIC AI servers, with projected shipments of **3.4 million** and **4.7 million** units in **2026E/27E** respectively, compared to **5.6 million** and **7.1 million** for GPU AI chips [1][8] - The **average selling price (ASP)** for ASIC AI substrates is expected to be **US$180/US$220** in **2026E/27E**, while GPU AI substrates are projected at **US$143/US$183** [1][8] - The combined market share for Taiwan's ASIC AI substrate suppliers (Unimicron and NYPCB) is estimated to be **71%/65%/59%** for **2025E/26E/27E**, significantly higher than the **10%/19%/21%** for GPU AI server market share [9] Supplier Performance and Projections - **Unimicron** and **NYPCB** are expected to benefit the most from the improving demand outlook, with revenue contributions from AI servers projected to reach **16%** and **20%** respectively by **2027E**, up from previous estimates of **13%** and **12%** [1][4] - **Ibiden** is also anticipated to gain market share in the ASIC AI server substrate market due to its technology leadership and plans for **40% capacity expansion** from **2025-27** [2] Pricing and Supply Chain Dynamics - Prolonged lead times for high-end ABF substrates are expected due to **T-glass material shortages**, which could lead to higher pricing despite a low equipment utilization rate (UTR) forecast of **75%** for **2H25-1H26E** [3][39] - Price hikes for ABF substrates are anticipated to occur in **late September/early October**, coinciding with quarterly contract negotiations, but may not significantly impact earnings outlook for suppliers in **2H26** [3][40] - The overall ABF substrate market is projected to grow at an **18% CAGR** from **2025-27E**, driven primarily by AI server applications [38] Earnings Revisions and Target Price Adjustments - **Unimicron**'s earnings estimates for **2025E/26E/27E** have been raised by **1%/3%/3%**, reflecting higher expected demand for ASIC AI server substrates [51] - The target price for **Unimicron** has been increased to **NT$116** from **NT$110** [4][55] - **NYPCB**'s earnings estimates have been revised upwards by **96%/67%/32%** for **2025E/26E/27E**, with a new target price of **NT$125** [57][62] Long-term Outlook and Risks - The long-term outlook for high-end ABF substrate suppliers remains positive, driven by the overall AI server demand uptrend, with an **81% CAGR** expected from **2024-27E** [10] - Key risks include slower-than-expected recovery in PC demand, delays in ABF substrate upgrades, and challenges in qualifying new high-end capacities [66] Additional Important Insights - The report emphasizes the potential conflict of interest due to Goldman Sachs' business relationships with the companies covered in the research [4] - The analysis suggests that while short-term oversupply may weigh on stock performance, the long-term demand growth for high-end substrates will support earnings [44]

Summary of Astera Labs (ALAB US) Conference Call Company Overview - **Astera Labs** is a U.S.-listed company specializing in PCIe retimer and switch chips, with a focus on the upcoming custom Scorpio-X switch chip [1] Key Industry Insights - **Growth Drivers**: Astera Labs' growth is driven by two main products: - Custom **NeuronLink** switch chip for AWS's Trainium series, launching in the second half of the year - Custom **UALink** switch chip for AMD's MI400 series, expected in the second half of next year [2] Technical Comparisons - **UALink vs. NVLink**: - UALink uses SerDes with differential signaling, allowing longer-distance data transmission compared to NVLink's single-ended signaling, which saves chip area but limits distance [3][4] - UALink can connect up to 1,024 nodes, while NVLink is limited to 576 nodes [5] - **UALink Protocol Versions**: - UALink has two versions: 128 Gbps and 200 Gbps, with the latter being suitable for GPU-to-GPU connections only [6][9] - UALink 128G supports mixed connections and is compatible with PCIe Gen7, making it suitable for model inferencing [9] - **Broadcom's SUE**: - SUE is a point-to-point protocol that draws from NVLink's logic but has limitations in heterogeneous expansion compared to UALink [10] Product Development - **AMD's Helios AI Rack**: - The upcoming Helios AI rack will adopt the UALink 200G protocol, with Astera Labs developing a switch chip expected to tape out in Q1 2026 [11][31] - **AWS Trainium Series**: - Astera Labs is developing the Scorpio-X switch chip for AWS's Trainium rack, which will be software-programmable and meet high-performance transmission requirements [13] Financial Projections - **Revenue Estimates**: - For every one million Trainium 2.5 chips deployed, Astera Labs could generate a content dollar value of approximately **$1.75 billion** from both large and small switch chips [22] - For Trainium 3 chips, the estimated content dollar value could reach **$3.3 billion** per million chips [26] - Additional revenue of **$150 million** is projected for every million Trainium 4 chips due to collaboration with Alchip [28] - **AMD's MI400 Series**: - Astera Labs' content dollar value for every one million MI400 GPUs used in the Helios rack is estimated at **$576 million** [32] Conclusion - Astera Labs is positioned to capitalize on the growing demand for advanced interconnect solutions in high-performance computing environments, particularly through its partnerships with AWS and AMD, with significant revenue potential from its innovative switch chip technologies [1][2][22][26][32]

Summary of Key Points from Conference Call Records Industry or Company Involved - **Unity Technologies (U)** [2][3] - **Marvell Technology (MRVL)** [4][5] - **ASML** [7][8][10] - **Amazon (AMZN)** [12] - **Palantir Technologies (PLTR)** [13][14] - **Roblox (RBLX)** [15][16] - **Alphabet (GOOGL)** [17][18][19] - **Twilio (TWLO)** [21][22] - **Nvidia (NVDA)** [24][25] - **Advanced Micro Devices (AMD)** [26] - **Meta Platforms (META)** [27][28][29] - **Uber (UBER)** [30][31] - **Booking Holdings (BKNG) and Expedia (EXPE)** [32] - **IBM** [32] Core Points and Arguments Unity Technologies - Jefferies raised Unity's price target to $35 due to positive feedback on its Vector ad product, with ROAS gains now in the 10-20% range, up from 5-7% last quarter [2][3] - Unity's 2025 ad share intent increased by 119 basis points year-over-year [2] Marvell Technology - Fubon Jefferies stated that Alchip is extending its lead over Marvell, with expectations for Alchip to ramp production to 2 million units compared to Marvell's 300,000 [4][5] ASML - ASML's Q2 revenue was €7.69 billion, up 23% year-over-year, with a gross margin of 53.7% [8] - The company provided a Q3 revenue guidance of €7.4-7.9 billion, which was below market expectations [8][10] - Management expressed uncertainty about 2026 growth due to macroeconomic factors [9] Amazon - Bank of America removed Amazon from its US 1 list, citing weaker-than-expected performance during Prime Day [12] Palantir Technologies - Mizuho upgraded Palantir to Neutral, raising the price target to $135, citing strong momentum in AI-driven segments [13][14] Roblox - JPMorgan raised Roblox's price target to $125, highlighting record engagement and bookings growth [15][16] Alphabet - Needham raised Alphabet's price target to $210, citing strength in digital ads and YouTube [17][18] - Cantor Fitzgerald also raised its price target to $196, anticipating strong Q2 results driven by Search and YouTube [19] Twilio - Mizuho noted Twilio's upcoming price hikes in voice and email services, which could add 1-2 percentage points to growth in 2025/2026 [21][22] Nvidia - Goldman Sachs reiterated a Buy rating on Nvidia, projecting potential EPS upside from the resumption of H20 GPU shipments to China [24][25] Advanced Micro Devices - Wells Fargo raised AMD's price target to $185, citing strong data center performance and inventory signals [26] Meta Platforms - Canaccord raised Meta's price target to $850, highlighting strong performance across ad products and AI tools [27][28] - Cantor Fitzgerald also raised its price target to $828 based on solid Q2 ad trends [29] Uber - TD Cowen expects Uber to report Q2 gross bookings of $46 billion, driven by strong performance in Mobility and Delivery [30][31] Booking Holdings and Expedia - Barclays warned of soft U.S. travel trends, expressing caution about rebound expectations for travel stocks [32] IBM - Stifel raised IBM's price target to $310, citing a strong setup for Q2 and a stable defensive profile [32] Other Important but Possibly Overlooked Content - The overall sentiment in the tech sector appears mixed, with some companies experiencing strong growth while others face challenges due to macroeconomic uncertainties [9][12][32] - The focus on AI and digital transformation continues to drive investment and growth expectations across multiple sectors, particularly for companies like Palantir and Nvidia [13][24] - The competitive landscape in the semiconductor industry is intensifying, with companies like Alchip gaining significant advantages over established players like Marvell [4][5]

Investment Rating - The report maintains a "Buy" rating for MediaTek and a "Neutral" rating for Alchip, while downgrading GUC to "Sell/H" [5][43][52]. Core Insights - The report emphasizes the increasing importance of system integration and interoperability in the semiconductor industry, particularly for AI ASICs and GPUs. It highlights that successful chip design alone is insufficient; cohesive and scalable systems are crucial for performance [1][9][18]. - The transition of AI chips from N5/4 to N3 nodes is expected to significantly increase average selling prices (ASP) and computing capabilities, despite a slowdown in overall AI chip shipment growth [2][16]. - Nvidia continues to dominate the AI server market, holding over 80% of market value, while Broadcom leads in the ASIC market driven by its Google TPUs [23][36]. Summary by Sections AI ASIC Market Dynamics - Broadcom holds the largest market share in the ASIC market, with stable trends expected for TPU v7 and growth anticipated for TPU v8 in late 2026 [3][36]. - The report estimates that AI ASIC shipments from major cloud service providers (CSPs) will reach approximately 4.5 million units in 2026, which is lower than consensus estimates [2][23]. System Integration and Design - The report stresses the need for advanced system design capabilities, including thermal management and network fabric design, to support the growing complexity of AI workloads [18][22]. - It notes that developing an HPC AI ASIC can take 3-4 years, necessitating partnerships with experienced semiconductor makers to expedite the process [20][22]. Company-Specific Insights - MediaTek is highlighted as a preferred choice due to its design expertise and expected revenue contributions from AI ASICs starting in late 2026 [5][43][45]. - Alchip's contribution is expected to begin in 2H26, with a cautious outlook due to potential redesign delays [52][53]. Financial Projections - The report provides revised financial estimates for MediaTek, projecting a revenue of NT$588.85 billion for 2025, with a YoY growth of 11% [46]. - Alchip's financial outlook is less optimistic, with a projected revenue of NT$41.96 billion for 2025, reflecting a decline [52][53].

Group 1: Economic and Fiscal Implications - The "Big and Beautiful" Act signed by Trump includes tax cuts and significant spending measures, projecting a $4 trillion tax reduction and a $1.5 trillion spending cut over the next decade [1] - The Act is expected to increase the U.S. deficit by approximately $3.4 trillion over the next ten years, raising concerns about fiscal deficits and debt [1][2] - The Act is anticipated to create a super loose fiscal stimulus environment, potentially benefiting the economy and stock market liquidity, while negatively impacting U.S. debt and the dollar [1][2] Group 2: Market Reactions and Predictions - The Act is likely to push up U.S. Treasury yields and weaken the dollar, with increased demand for gold as a safe haven [2] - Analysts suggest that the Act may lead to a stronger recovery for the U.S. economy, although the extent remains uncertain [2] - The S&P 500 index is projected to reach 6850 points within 12 months, despite concerns over tariffs and geopolitical tensions [3] Group 3: Sector-Specific Impacts - The Act provides significant tax incentives and funding support for sectors such as chip manufacturers, energy companies, defense contractors, and real estate developers, while cutting subsidies for electric vehicles and renewable energy projects [1] - Traditional energy and military sectors are expected to benefit, while the renewable energy sector may face challenges [2] - The German stock market has shown strong performance, with military procurement plans potentially benefiting key defense manufacturers [5][6] Group 4: Oil Market Dynamics - OPEC+ has agreed to increase production by 548,000 barrels per day, exceeding market expectations, which may lead to a global oversupply risk and downward pressure on oil prices [7][8] - The strategy shift from limiting supply to increasing production aims to reclaim market share lost to U.S. shale oil [8] Group 5: AI Chip Supply Chain Growth - The global ASIC supply chain is entering a growth cycle driven by the widespread adoption of AI ASIC chips by cloud service providers like Google and Amazon [9] - Companies such as MPI, Aspeed, and Alchip are expected to benefit significantly from this trend, with MPI planning to double its production capacity [9][10]

Group 1: ASIC Chip Market Outlook - Bank of America reports strong and sustained growth in the ASIC chip supply chain due to widespread adoption of AI ASICs by CSPs [1] - Despite extended project cycles, the demand for current generation products has created a historical super demand cycle for the global ASIC supply chain [1] - Google and Amazon are expected to maintain annual ASIC chip production above 1 million units, with Meta and Microsoft gradually catching up [1] Group 2: MPl's Growth and Market Position - MPl, a key supplier of probe cards for ASIC chips, is benefiting from strong customer demand, with MEMS probe card capacity expected to expand from 600,000 units per month to nearly 1 million by Q3 2025 [1][2] - The target price for MPl has been raised to NT$1050 based on a projected P/E ratio of 28 times for 2026 [2] Group 3: Aspeed's Business Performance - Aspeed is showing strong performance in its BMC business, with the AST2700 expected to be adopted by the Rubin platform, leading to reduced competitive pressure [2] - Earnings per share estimates for 2026 and 2027 have been raised by 5% and 12% respectively, with the target price increased to NT$6250 [2][3] - Aspeed's revenue is projected to reach NT$2.2 billion in Q3 2025, with a gross margin of 66% [3] Group 4: Alchip's Opportunities and Risks - Alchip is seeing increased collaboration opportunities with key customers as CSP projects advance, with a higher likelihood of mass production for Trainium 3 [4] - The target price for Alchip has been raised to NT$3900 based on a rolling valuation period extending to mid-2027 [4] - Upward risks for Alchip include faster mass production of non-AWS projects and improved profit margins due to relaxed restrictions on China [4]