Core Insights - Nvidia's valuation has reached $5 trillion due to its significant share in the artificial intelligence spending boom [2] - The new benchmark for advanced data centers is measured in gigawatts of computing power, shifting the focus from physical size or server count [2] - The cost of 1 gigawatt (GW) of AI data center capacity is approximately $35 billion, representing a new economic foundation for the AI industry [3] Cost Structure of AI Data Centers - Approximately 39% of total spending in AI data centers is allocated to GPUs, with Nvidia's products dominating this segment [6] - Nvidia captures nearly 30% of the profits from AI data center expenditures due to its 70% gross margin [6] - Each gigawatt of power can support over 1 million GPU chips, generating $1.3 billion in revenue for Nvidia's manufacturing partner, TSMC [6] Networking Equipment - 13% of data center costs are attributed to networking equipment, benefiting companies like Arista Networks, Broadcom, and Marvell [7] - Component manufacturers such as Amphenol and Luxshare Precision will also gain from cables and connectors [7] Power and Cooling Infrastructure - Physical infrastructure, including power distribution, accounts for nearly 10% of the costs of a 1 GW AI data center [9] - Major players in this sector include Eaton, Schneider Electric, ABB, and Vertiv, with Vertiv also having opportunities in thermal management [9] Real Estate and Labor Costs - Land and buildings represent about 10% of upfront costs, while operational costs are relatively low, with annual electricity costs for a 1 GW data center around $1.3 billion [11] - Large data centers typically employ only 8 to 10 staff members, with salaries ranging from $30,000 to $80,000 [11] - The bottleneck is shifting to power supply, with companies like Siemens Energy and GE Vernova reporting increased orders for turbine and grid infrastructure [11]

Core Insights - Nvidia has become the first company globally to reach a market capitalization of $5 trillion, driven by its significant share in AI investment spending [1] - The measurement of data centers has shifted from physical size to computing power measured in gigawatts, with Wall Street now evaluating data centers based on "cost per gigawatt" [1] - The construction cost for a 1 gigawatt AI data center is estimated at $35 billion, representing a new economic foundation for the AI industry [2] Cost Structure of AI Data Centers - The largest single cost in AI data centers is attributed to GPUs, accounting for approximately 39% of total expenditures, with Nvidia's chips being the primary contributors [3] - Each 1 gigawatt of computing power requires over 1 million GPU chips, highlighting the central role of GPUs in the AI industry [3] - Networking equipment follows GPUs in cost, comprising about 13% of data center expenses, benefiting companies like Arista Networks and Broadcom [4] Infrastructure and Operational Costs - Power and cooling infrastructure, including generators and transformers, represent nearly 10% of total costs for a 1 gigawatt AI data center [6] - The operational costs of running a 1 gigawatt AI data center are relatively low, with annual electricity costs estimated at $1.3 billion and minimal personnel requirements [7] - The industry is facing challenges related to power supply, with major companies like Siemens Energy and GE Vernova reporting a surge in orders for turbines and grid infrastructure [7]

Core Viewpoint - NVIDIA's CEO Jensen Huang expressed optimism about the revenue potential of the Blackwell and Rubin AI product lines, projecting significant sales growth compared to the previous Hopper series, although subsequent clarifications adjusted these figures [2][4]. Revenue Projections - The initial claim of $500 billion in revenue from Blackwell and Rubin products over the next five quarters was clarified to represent cumulative shipments from 2025 to 2026, including revenues from NVIDIA's InfiniBand and NVLink products [4]. - It is estimated that 30% of the expected demand has already been shipped, contributing $100 billion in revenue from Blackwell products this month [4]. - The revised revenue expectation for the next five quarters is $307 billion, lower than the initial projection made by Huang [4]. Product Performance - The Blackwell series is noted for its superior performance and energy efficiency, which has made it popular among customers, marking a significant achievement for NVIDIA [4]. - The Rubin product line is anticipated to be crucial for NVIDIA's expansion in computing capabilities, with the introduction of the Vera Rubin superchip, which integrates ARM-based Vera CPU and Rubin chipsets [4]. Market Position and Growth - NVIDIA's market capitalization reached a record $5 trillion, indicating strong momentum in the AI sector [5]. - The company has transitioned from a focus on consumer GPUs to establishing a foundational role in providing necessary computing power for large tech firms, significantly increasing its market share [5]. - NVIDIA's recent announcements at the GTC 2025 conference have contributed to its soaring market value and reinforced its leadership in AI computing [5][6]. Strategic Developments - The GTC 2025 conference showcased NVIDIA's collaborations with notable companies like Nokia and Palantir, highlighting untapped applications of AI technology [6]. - There is potential for NVIDIA to re-enter the Chinese AI market, which could yield substantial additional revenue, as discussions about Blackwell AI chips with Chinese leaders are anticipated [6].

Market Performance - As of October 23, 2025, the Zhongzheng Semiconductor Industry Index decreased by 1.52%, with major stocks like ShenGong Co. leading the decline at 5.10% [2] - The Sci-Tech Innovation Board Chip Index fell by 2.00%, with stocks such as TuoJing Technology rising by 3.06% while ShengKe Tongxin dropped by 4.98% [4] - The Sci-Tech Innovation Board New Materials Index declined by 1.32%, with TianCheng Technology increasing by 5.98% and ShenGong Co. again leading the decline [7] ETF Performance - The Semiconductor Industry ETF (159582) decreased by 1.37%, with a recent price of 2.09 yuan, but showed a 3.87% increase over the past month [2] - The Sci-Tech Chip ETF (588990) also fell by 1.94%, with a recent price of 2.43 yuan, and a 0.98% increase over the past month [4] - The Sci-Tech New Materials ETF (588010) dropped by 1.37%, with a recent price of 0.79 yuan, and a 0.75% increase over the past month [7] Liquidity and Trading Volume - The Semiconductor Industry ETF had a turnover rate of 4.43% with a trading volume of 17.2464 million yuan, averaging 77.5828 million yuan daily over the past month [2] - The Sci-Tech Chip ETF had a turnover rate of 8.66% with a trading volume of 58.0718 million yuan, averaging 161 million yuan daily over the past month [4] - The Sci-Tech New Materials ETF had a turnover rate of 8.03% with a trading volume of 19.9337 million yuan, averaging 32.1769 million yuan daily over the past week [7] Industry Developments - Yangtze Memory Technologies is considering an IPO in mainland China, potentially valued over 40 billion USD, which could be one of the largest listings in the domestic semiconductor sector [8] - Tesla has added a manufacturing engineer position for its Optimus robot, planning to launch several products next year, indicating a growing interest in humanoid robots [8] - NVIDIA's new Rubin series will utilize M9 materials, suggesting an increase in material chain focus and potential technological upgrades [8] Institutional Insights - The expectation of Yangtze Memory's IPO strengthens the logic of domestic storage autonomy, potentially boosting valuations in chip equipment and material chains [9] - The AI and semiconductor sectors are experiencing renewed interest, with a focus on upstream components like sensors and drive systems due to Tesla and other developments [9] - The peak production of global advanced packaging lines is expected between 2025 and 2028, with equipment procurement cycles starting in late 2024, indicating a significant demand for related equipment [9] ETF Overview - The Semiconductor Industry ETF closely tracks the Zhongzheng Semiconductor Industry Index, covering key sectors in semiconductor materials, equipment, and applications [10] - The Sci-Tech Chip ETF tracks the Sci-Tech Innovation Board Chip Index, focusing on semiconductor-related companies with a maximum of 50 constituents [10] - The Sci-Tech New Materials ETF follows the Sci-Tech Innovation Board New Materials Index, concentrating on advanced materials relevant to the semiconductor industry [10] Weighting and Composition - As of September 30, 2025, the top ten stocks in the Zhongzheng Semiconductor Industry Index accounted for 78.18% of the index [11] - The top ten stocks in the Sci-Tech Chip ETF represented 59.69% of the index [11] - The top ten stocks in the Sci-Tech New Materials Index made up 51.58% of the index [12]

Group 1: Nvidia's New Product and Market Impact - Nvidia has confirmed the use of M9 materials in its next-generation product Rubin, with the CPX and midplane PCBs utilizing M9 CCL, amid a shortage of Q cloth [1] - The Rubin CPX is designed for ultra-long context AI inference tasks, featuring a decoupled inference architecture and significant hardware changes, including a wireless cable architecture [1] - The market potential for CPX, midplane, and orthogonal backplane is nearly 100 billion [1] - By 2027, Nvidia's AI PCB market is projected to reach $6.96 billion, a 142% increase from 2026 [1] Group 2: PCB Industry Growth Driven by AI - The global PCB market is expected to grow from $62 billion in 2020 to $75 billion by 2024, with a compound annual growth rate of 4.9% [2] - The value of PCBs in AI servers is significantly higher than in traditional servers, leading to a substantial increase in demand for high-performance PCBs [3] - Low dielectric constant and low loss factor materials are deemed most suitable for high-performance applications, with Q cloth outperforming second-generation cloth [3] Group 3: Recommendations for M9 and Related PCB Manufacturers - Strong recommendations have been made for M9 upstream and related PCB manufacturers, with companies like 建滔积层板 reporting a revenue increase of 11% year-on-year [4] - 建滔集团 is expanding its production capabilities for AI-related products, with a projected investment of approximately 800 to 1,000 million RMB for a new production line [4] - The demand for copper-clad laminates and printed circuit boards is expected to rise significantly due to the rapid development of AI technologies [4]

Group 1: Nvidia's New Product and Market Impact - Nvidia is set to launch its new product series, Rubin, in the second half of next year, utilizing M9 materials for its CPX and midplane PCBs due to a shortage of quartz fabric [1] - The Rubin CPX is designed specifically for long-context AI inference tasks, featuring a decoupled inference architecture and significant hardware innovations, which are expected to expand its market size [1] - According to CICC, the AI PCB market for Nvidia is projected to reach $6.96 billion by 2027, representing a 142% increase from 2026 [1] Group 2: PCB Industry Growth Driven by AI - The global PCB market is expected to grow from $62 billion in 2020 to $75 billion by 2024, with a compound annual growth rate (CAGR) of 4.9% [2] - The demand for high-performance PCBs is anticipated to rise significantly due to the higher value of AI server PCBs compared to traditional servers, with low dielectric constant materials being preferred [3] Group 3: Recommendations for M9 and Related PCB Manufacturers - Strong recommendations have been made for M9 upstream and related PCB manufacturers, with companies like Kintor achieving a revenue of HKD 9.588 billion in the first half of the year, a year-on-year increase of 11% [4] - Kintor Group is actively developing high-frequency and high-speed products for AI server GPU motherboards and plans to establish an AI PCB production line in Guangdong with an investment of approximately RMB 800 million to 1 billion [4]

Investment Rating - The investment rating for the company is maintained as "Buy" with a target price of JPY 31,880, slightly down from the previous JPY 32,000 [1][8]. Core Views - The company has adjusted its full-year revenue and profit guidance for FY2026 down by 9.6% to JPY 2.35 trillion and by 21.6% to JPY 570 billion, respectively, primarily due to changes in customer capital expenditure plans [1][2]. - Despite the downward revision in revenue and profit forecasts, the company maintains its capital expenditure guidance for FY2026 at JPY 240 billion, reflecting ongoing investments in next-generation technologies driven by AI demand [2][3]. - The report highlights that the demand for AI chips remains strong, particularly for the Rubin series, which is expected to positively impact the company's performance in the semiconductor equipment sector [1][8]. Summary by Sections Financial Performance - For FY26Q1, the company reported revenue of JPY 549.5 billion, a year-on-year decrease of 1.0% and a quarter-on-quarter decrease of 16.1%, which was below Bloomberg consensus expectations [1]. - Operating profit for the same period was JPY 144.6 billion, down 12.7% year-on-year and 21.3% quarter-on-quarter, also missing consensus estimates [1]. - Net profit attributable to shareholders was JPY 117.8 billion, reflecting a year-on-year decline of 6.6% and a quarter-on-quarter decline of 17.6%, again below expectations [1]. Revenue and Profit Forecasts - The company has revised its revenue forecasts for FY26, FY27, and FY28 down by 11.7%, 11.9%, and 12.0%, respectively, and net profit forecasts down by 17.1%, 15.8%, and 16.1% [3][7]. - The adjustments are attributed to several factors, including the correction of capital expenditure plans by leading logic customers and a cautious approach to NAND investments [1][3]. Market Outlook - The global wafer fab equipment (WFE) market size forecast for CY2025 has been raised from USD 110 billion to USD 115 billion, reflecting currency fluctuations [2]. - The report anticipates significant increases in transistor counts and storage capacities for high-end AI server modules from CY2025 to CY2027, indicating robust growth potential in the AI-driven equipment market [2]. Valuation Metrics - The company’s estimated PE ratios for FY26 and FY27 are 25.47 and 22.63, respectively, with a target price based on a PE of approximately 26.4 times FY27E earnings [8][21]. - The report also provides a comparison of valuation metrics with peer companies, indicating a competitive position in the market [21].

Investment Rating - The report maintains a "Buy" rating for the company with a target price of JPY 31,880, slightly down from the previous target of JPY 32,000 [6][4]. Core Views - The company has adjusted its FY26 revenue and operating profit guidance downwards by 9.6% to JPY 2.35 trillion and by 21.6% to JPY 570 billion, respectively, primarily due to changes in customer capital expenditure plans [2][4]. - Despite the downward revision in revenue and profit forecasts, the company remains optimistic about the demand for AI-related chips and the potential positive impact of the Rubin series on global semiconductor equipment performance [2][4]. - The company expects the global wafer fab equipment (WFE) market size for CY2025 to increase from USD 110 billion to USD 115 billion, reflecting the impact of currency fluctuations and the anticipated growth in AI server module capabilities [3][4]. Summary by Sections Financial Performance - For FY26Q1, the company's revenue was JPY 549.5 billion, a year-on-year decrease of 1.0% and a quarter-on-quarter decrease of 16.1%, which was below Bloomberg consensus expectations by 5.4% [1]. - Operating profit for the same period was JPY 144.6 billion, down 12.7% year-on-year and 21.3% quarter-on-quarter, also missing consensus expectations by 7% [1]. - Net profit attributable to shareholders was JPY 117.8 billion, a decline of 6.6% year-on-year and 17.6% quarter-on-quarter, falling short of consensus expectations by 11% [1]. Revenue and Profit Forecasts - The company has revised its FY26 revenue forecast down by 9.6% to JPY 2.35 trillion and operating profit down by 21.6% to JPY 570 billion, citing several factors including adjustments in capital expenditure by leading logic customers and reduced investments by emerging Chinese chip manufacturers [2][4]. - The report anticipates a continued strong demand for AI chips, particularly from companies like NVIDIA, which may drive performance in semiconductor equipment [2][4]. Capital Expenditure and R&D - The company maintains its capital expenditure guidance for FY26 at JPY 240 billion, significantly up from JPY 162.1 billion in FY25, with R&D expenses set at JPY 295 billion, reflecting ongoing investments in next-generation etching, deposition, and bonding equipment [3][4]. - The report highlights that the demand for advanced logic foundry services is expected to grow, particularly as TSMC continues to increase its capital expenditures while Intel and Samsung face challenges [2][4]. Valuation Metrics - The report projects a decline in FY26/27/28 operating revenue by 11.7%/11.9%/12.0% and a decrease in net profit attributable to shareholders by 17.1%/15.8%/16.1%, with diluted EPS expected to be JPY 1,073/1,208/1,342 [4][10]. - The company is valued at approximately 26.4 times FY27E PE, based on an average PE of 22.4 times for comparable companies [4][10].

Core Viewpoint - The article discusses the advancements in AI technology and infrastructure by NVIDIA, highlighting the launch of new architectures and partnerships aimed at enhancing AI capabilities and performance in various sectors [1][2][3][4][5][6]. Group 1: AI Architecture Developments - NVIDIA is transitioning from Generative AI to Agentic AI, with future developments leading to Physical AI, indicating a significant evolution in AI capabilities [1]. - The Blackwell architecture has been fully launched, showcasing the Grace Blackwell NVLink 72 chip, which integrates 72 Blackwell GPUs and achieves 1.4 EFLOPS performance [2]. - The Blackwell Ultra NVL72 platform is set to double the bandwidth and increase memory speed by 1.5 times compared to its predecessor, paving the way for advanced AI inference [3]. Group 2: Market Demand and Procurement - The top four U.S. cloud service providers have purchased 1.3 million Hopper chips in 2024 and are expected to acquire 3.6 million Blackwell chips in 2025, indicating a strong demand for AI computing infrastructure [2]. - By 2028, capital expenditures for intelligent computing centers are projected to exceed $1 trillion, reflecting the growing investment in AI technologies [2]. Group 3: Future Product Launches - The Vera Rubin platform is anticipated to start shipping in the second half of 2026, featuring NVLink 144 technology and achieving performance levels 3.3 times greater than the GB300 NVL72 [4]. - The next-generation Rubin Ultra NVL576 is expected to launch in the second half of 2027, with performance projected to be 14 times that of the GB300 NVL72 [4]. Group 4: Strategic Partnerships and Collaborations - NVIDIA is expanding its collaboration with General Motors to develop autonomous vehicles and enhance AI model training [5]. - Partnerships with Cisco and T-Mobile aim to explore AI-native networks for next-generation 6G wireless technology [5]. - The introduction of the GR00T N1 robot model and collaboration with Google DeepMind and Disney for a physics engine indicates NVIDIA's commitment to advancing robotics [5]. Group 5: Cost Efficiency and Market Impact - The efficiency of computing power is improving at a rate of three times every eight months, leading to significant cost reductions, which benefits global cloud providers and applications [6].