Core Insights - The report emphasizes long-term opportunities with the strategy "蓄力新高" suggesting that the Shanghai Composite Index briefly broke 4000, while the 2026 strategy "奔马资产, 策马逐牛" focuses on embracing "奔马资产" (globally competitive leaders) leading to a revaluation of value [3][10] - The mid-term analysis indicates a potential for market fluctuations towards the end of the year, with a strong market rally observed in the first week of January, confirming previous predictions [3][10] Industry and Sector Analysis - Leading sectors such as telecommunications, electronics, and non-ferrous metals remain core themes, with internal shifts observed, such as a transition from rare earths and precious metals to industrial metals and lithium-cobalt-nickel within non-ferrous metals, and from consumer electronics to storage and semiconductor equipment in electronics [4][14] - The report identifies three key investment directions: 1. Core growth assets, particularly in the Hang Seng Internet sector, benefiting from platform economy support and potential AI catalysts, alongside improvements in US-China relations and passive foreign capital inflow due to RMB appreciation [5][13] 2. Globally competitive assets (奔马 50), which are expected to benefit from global economic recovery, strong policy support, and institutional capital inflow, with a high cost-performance ratio due to trends in AI, high-end manufacturing, and resource supply-side adjustments [5][13] 3. Emerging growth sectors, particularly those related to the "Musk chain," focusing on AI applications and underground transportation, with a bottom-up investment approach in areas like computing power and humanoid robots [5][13] Market Dynamics - The report notes that despite recent volatility, the fundamentals of a long-term bull market remain intact, with market sentiment high and financing balances nearing a ten-year high, indicating a healthy market environment [7][11] - Historical patterns suggest that after a major rally, the market may enter a consolidation phase, but the underlying growth logic remains strong, particularly in technology and cyclical sectors [12][14]

周五（1月16日），创业板人工智能下跌2%回踩五日线，资金大举涌入！AI应用概念股全线回调，汉得 信息、蓝色光标、昆仑万维等多股跌超10%。但光模块等算力赛道仍保持活跃，联特科技、太辰光、致 尚科技等多股涨超5%，长芯博创、天孚通信、光库科技、锐捷网络等多股涨超1%。 热门ETF方面，双线布局"算力+AI应用"的创业板人工智能ETF（159363）场内收跌1.81%回踩五日线， 成交额超12亿元。资金加速涌入，单日净申购4.36亿份（按当日场内平均价预估金额约4.78亿元），叠 加前四日增持12亿元，单周累计金额约16.78亿元！ 1、创业板人工智能持续领跑！周度复盘看，前半周受AI应用爆发驱动，随后市场回切光模块等算力方 向，创业板人工智能ETF（159363）周线强势八连阳，标的指数八周累计涨幅为34.66%，显著跑赢科创 AI、CS人工智、人工智能等同类AI主题指数，验证创业板人工智能板块弹性优势突出。 | 2025/11/28-2026/01/16 (8周) | | --- | | .. | | инание как компания к как компании компании и и | | . ...

- The report introduces a **market timing system** that uses the distance between the 20-day moving average and the 120-day moving average of the WIND All A Index to determine market trends. The system identifies an uptrend when the short-term moving average is above the long-term moving average, with a significant distance threshold of 3%[2][6][11] - The **industry trend allocation model** is highlighted, which signals opportunities in specific sectors. For the medium term, the "distressed reversal expectation model" suggests focusing on innovative healthcare. The "TWO BETA model" continues to recommend the technology sector, particularly AI applications and commercial aerospace after adjustments. In the short term, the "earnings trend model" points to opportunities in computing power (e.g., Sci-Tech Chip ETF, code 588200) and energy storage batteries (e.g., Energy Storage Battery ETF, code 159566)[2][5][7] - The **position management model** is used to determine stock allocation levels. Based on the WIND All A Index's valuation and trend, the model recommends an 80% stock allocation for absolute return products[5][7] - The **valuation indicators** for the WIND All A Index are also discussed. The PE ratio is at the 90th percentile, indicating a relatively high valuation, while the PB ratio is at the 50th percentile, representing a medium level[5][7][11]

Core Viewpoint - The launch of Colossus 2, the world's first 1GW supercomputing cluster, marks a significant advancement in AI infrastructure, with plans to upgrade to 1.5GW by April and potentially reach 2GW, which could match the power consumption of major U.S. cities [2][12]. Group 1: Colossus 2 Overview - Colossus 2 is equipped with approximately 200,000 NVIDIA H100/H200 GPUs and around 30,000 NVIDIA GB200 NVL72 GPUs, significantly enhancing its computational power compared to its predecessor, Colossus 1, which was built in just 122 days [9][10]. - The cluster's 1GW capacity can power about 750,000 households, equivalent to the peak power demand of San Francisco [11]. - Once fully operational, Colossus 2 will house 555,000 GPUs, surpassing the GPU counts of Meta, Microsoft, and Google [13][14]. Group 2: Implications for AI Development - The advancements in Colossus 2 are expected to facilitate the development of Grok 5, which is projected to have parameters around 6 trillion, more than double that of Grok 4 [15][18]. - With the recent $20 billion funding round for xAI, the scaling capabilities for Grok 5 are increasing, leading to larger model parameters and faster training and deployment speeds [18][19]. - The rapid development of AI models is seen as a competitive advantage in the industry, emphasizing that speed is a crucial factor in the AI era [20]. Group 3: Energy Supply Concerns - The construction of large data centers like Colossus 2 is contributing to a projected annual electricity demand growth of 4.8% over the next decade, which is unprecedented for the U.S. energy system [27]. - The imbalance between rapidly increasing demand and slow supply growth is causing concerns about the stability of the power grid, leading to potential rolling blackouts for 67 million residents in 13 states during extreme weather [5][22][23]. - PJM, the regional transmission organization, is struggling to maintain supply-demand balance and has proposed measures to reduce peak demand from data centers, which have faced opposition from major tech companies [32][34].

Investment Rating - The investment rating for TSMC is "Buy" [7] Core Insights - TSMC confirms the strength and sustainability of AI demand, leading to an upward revision of long-term guidance and boosting the AI narrative [3] - Capital expenditures have been raised beyond expectations, with wafer manufacturing remaining a bottleneck, indicating a positive outlook for computing power [3] Financial Summary - Total revenue projections for TSMC are adjusted to NT$ 4,880,557 million for FY2026, NT$ 6,151,836 million for FY2027, and NT$ 7,532,792 million for FY2028, reflecting growth rates of 28%, 26%, and 22% respectively [5] - GAAP net profit is projected to be NT$ 2,362,106 million for FY2026, NT$ 2,976,546 million for FY2027, and NT$ 3,650,595 million for FY2028, with growth rates of 38%, 26%, and 23% respectively [5] - The gross profit margin is expected to increase to 63.5% in FY2026 and 64.0% in FY2028 [5] Earnings Forecast and Investment Recommendations - The earnings forecast for TSMC has been adjusted to NT$ 4,880.6 billion for FY2026, NT$ 6,151.8 billion for FY2027, and NT$ 7,532.8 billion for FY2028, with GAAP net profit projected at NT$ 23,621 million, NT$ 29,765 million, and NT$ 36,506 million respectively [11] - The target price for TSMC's stock is set at $407 for FY2027, based on a PE ratio of 22x [11]

Core Insights - China is projected to have more electricity than any other country and potentially more chips, leading to a significant advantage in AI computing power by 2026, with an estimated electricity generation three times that of the US [1][6][9] Group 1: Importance of Energy in AI Development - The rapid growth of AI is shifting energy supply from a traditional support role to a critical factor influencing industry competitiveness, with data centers' electricity demand increasing significantly [3][9] - A robust and continuously upgraded power system is essential for leading AI computing capabilities, as AI activities are highly energy-intensive [3][9] - The relationship between energy, chips, and algorithms forms a "computing power triangle," where energy supports scaling, chips are the output carriers, and algorithms enhance efficiency [4][9] Group 2: Collaborative Optimization Pathways - The industry is developing a collaborative optimization path among energy, chips, and algorithms, utilizing strategies like "cluster breakthroughs" to enhance performance in specific scenarios [4][10] - Distributed training of large models requires substantial chip clusters, which depend on stable energy supply for optimal performance [10][12] Group 3: Advantages from Energy Supply and Policy - The large-scale development of renewable energy sources like wind and solar enhances electricity supply stability and reduces costs, creating favorable conditions for AI computing expansion [6][12] - Forward-looking policies are guiding the construction of computing hubs and data center clusters, aligning energy supply with computing demand and improving energy efficiency [6][12] Group 4: Innovation and Ecosystem Collaboration - Companies are leveraging technological innovations to enhance computing efficiency and create a comprehensive AI industry ecosystem, driving rapid growth in the digital economy [7][13] - The modernization of the energy system is seen as a key competitive factor in the AI era, with China's focus on energy transformation supporting its economic and carbon neutrality goals [7][13]

Core Viewpoint - The AI sector in the ChiNext market is experiencing a significant influx of capital, despite a recent 2% decline in the ChiNext AI index, with notable fluctuations in AI application stocks and continued strength in computing power-related stocks [1][3]. Group 1: Market Performance - The ChiNext AI ETF (159363) has shown strong performance, achieving an eight-week consecutive increase with a cumulative gain of 34.66%, outperforming similar AI-themed indices [1][3]. - The ETF recorded a daily net subscription of 436 million units, with a total weekly inflow of approximately 1.678 billion yuan, indicating robust investor interest [1][3]. Group 2: Sector Analysis - The light module industry is in a high prosperity cycle, driven by the explosive demand for AI computing power, with supply becoming a core issue [3]. - Major manufacturers in the light module sector are accelerating capacity expansion, with expectations of a significant release of production capacity in Q1 2026, which could drive a new performance growth phase [3]. Group 3: Fund Characteristics - The ChiNext AI ETF (159363) has reached a record size of 5.527 billion yuan, with an average daily trading volume of nearly 800 million yuan over the past six months, leading among eight ETFs tracking the ChiNext AI index [3]. - The ETF has been included in the Stock Connect program, effective January 19, which is expected to enhance its trading activity and attract northbound capital [3]. Group 4: Investment Strategy - The current trend in AI development is shifting from computing power construction to application implementation, making the ChiNext AI ETF a direct beneficiary of the commercialization of AI technology [5]. - The ETF's portfolio consists of approximately 60% in computing power (primarily light modules) and 40% in AI applications, representing both core computing and true AI application sectors [5].

Core Viewpoint - TSMC's recent earnings report exceeded market expectations, igniting a new wave of AI-themed market enthusiasm, with a record capital expenditure plan of up to $56 billion by 2026, boosting confidence across the semiconductor industry [1] Group 1: Impressive Earnings - TSMC's Q4 2025 revenue reached approximately $33.7 billion, with a net profit of $16 billion, showing significant year-on-year growth [2] - The gross margin increased to 62.3%, indicating strong pricing power and technological advantages in advanced processes [2] - High-performance computing revenue, including CPU, GPU, and AI acceleration chips, grew significantly, accounting for 55% of total revenue, surpassing smartphone business for the first time [2] - The 3nm and 5nm processes contributed over 60% of revenue, demonstrating the financial returns from technological leadership [2] Group 2: Bold Future Investments - TSMC provided an optimistic revenue growth forecast of nearly 30% for 2026 and raised its capital expenditure plan to $52 billion to $56 billion, a nearly 30% increase from the previous year [3] - This investment decision was made after extensive discussions with major clients like NVIDIA, Apple, and AMD, aimed at addressing confirmed long-term structural demand driven by AI [3] - A significant portion of the capital expenditure (10%-20%) will be allocated to advanced packaging technology to enhance AI chip production capacity [3] Group 3: Industry Chain Revaluation - TSMC's positive outlook has triggered a value reassessment across the semiconductor supply chain, particularly benefiting semiconductor equipment manufacturers [4] - ASML, the sole supplier of extreme ultraviolet lithography machines, saw its market value soar past $500 billion following TSMC's report, becoming one of Europe's highest-valued tech companies [4] - The optimistic sentiment spread to the broader chip sector, with stocks of companies like NVIDIA, AMD, and Micron Technologies rising [4] Group 4: Evolving AI Demand - The underlying logic supporting TSMC and the industry's confidence is the qualitative shift in AI demand, moving from model training to inference needs driven by widespread application [5][6] - As AI transitions from experimental phases to various industries, the demand for computing power is expected to grow exponentially, indicating a more sustainable demand base [6] Group 5: Challenges Ahead - Despite the optimism, there are cautious voices regarding the sustainability of high profitability and capital returns amid geopolitical pressures and cyclical risks [7] - TSMC's $56 billion capital expenditure requires sustained high customer orders over several years to avoid potential overcapacity risks [7] - The high costs of building fabs in regions like the U.S., Japan, and Europe compared to Taiwan pose long-term profitability challenges [7]

Core Insights - The AI computing sector is experiencing a pullback, with the cloud computing ETF Huatai (159273) dropping over 1% and recording a trading volume exceeding 300 million yuan, indicating a second consecutive day of capital inflow [1][3] Group 1: AI Demand and Financial Performance - Strong demand for artificial intelligence is evident, with TSMC's Q4 2025 earnings significantly exceeding expectations, achieving a net profit increase of 35% year-on-year to approximately 16 billion USD, marking a historical high with a gross margin surpassing 60% [3] - TSMC has raised its 2026 capital expenditure guidance to 52-56 billion USD, nearly 40% higher than previous expectations [3] Group 2: Market Movements and Stock Performance - The majority of the index-weighted stocks in the cloud computing ETF Huatai (159273) are showing declines, with notable drops including Hengsheng Electronics down over 2% and Tencent Holdings slightly down [3] - Conversely, stocks such as Zhongji Xuchuang and Alibaba-W have seen increases of over 1% [3] Group 3: Future Projections and Investment Opportunities - According to CITIC Securities, the demand for computing power is shifting from a single "training-driven" model to a "training + inference" dual-driven model, which is expected to maintain rapid growth [5] - The report predicts that by 2030, global token consumption could increase to 100-340 times that of 2025, with inference computing power potentially growing to 65-220 times [6] - The domestic cloud service providers (CSPs) are expected to accelerate their investments in computing power, driven by advancements in domestic computing chips and the introduction of super nodes [7]

Core Viewpoint - The AI sector is experiencing a correction, with significant declines in AI application stocks, while computing power stocks remain active and show gains, indicating a mixed market sentiment in the AI industry [1][3]. Group 1: AI Application Sector - The AI application concept stocks have seen notable declines, with Chinese Online dropping over 10% and several others like Hand Information and BlueFocus falling more than 8% [1]. - Despite the downturn in AI application stocks, the overall AI industry remains dynamic, with new financing rounds for overseas AI companies and the introduction of domestic policies promoting "AI + manufacturing" [1][3]. Group 2: Computing Power Sector - The computing power sector continues to show strength, with stocks like LianTe Technology and ZhiShang Technology leading gains of over 5% [1]. - The light module industry is in a high prosperity cycle, driven by the explosive demand for AI computing power, with major manufacturers accelerating capacity expansion [3]. Group 3: ETF Performance - The Huabao ChiNext AI ETF (159363) has seen a significant inflow of funds, with a net subscription exceeding 200 million yuan and a total increase of 1.2 billion yuan over the past four days [1][4]. - The Huabao ChiNext AI ETF has reached a record size of 5.527 billion yuan, leading in both size and trading volume among ETFs tracking the ChiNext AI index [4]. Group 4: Market Outlook - Analysts suggest that the current short-term correction does not alter the strong performance expectations for the A-share market in the first half of the year, with light modules expected to be a favorable sector during this period [3]. - The AI development is transitioning from computing power construction to application implementation, with the Huabao ChiNext AI ETF positioned to benefit from the commercialization of AI technology [3].