Report Industry Investment Rating The provided content does not mention the report industry investment rating. Core View As blockchain technology integrates deeply with the traditional financial system, tokenization of Real-World Assets (RWA) has become a key path to connect "on-chain value" and "offline entities." RWA bonds, with "real - asset credit anchoring" and "blockchain technology empowerment," are reshaping the operation logic of the traditional debt - financing market. The report comprehensively analyzes RWA bonds, including their definition, operation mechanism, comparison with traditional bonds and ABS, global policy and regulatory frameworks, typical cases, and investor participation paths [12]. Summary by Related Catalogs RWA Bond Core Definition - Concept: RWA bonds tokenize the creditor's rights of real - world assets with stable cash flows into standardized debt - financing tools using blockchain technology. They have "dual - anchoring" features, anchoring both the underlying asset's cash flow and the smart - contract - defined distribution rules, and are classified as "security tokens" subject to securities regulations [13][14]. - Underlying Asset Types: They include financial assets (e.g., accounts receivable, corporate loans, and national debts), physical assets (e.g., real estate, infrastructure, charging piles, and photovoltaic power stations), and equity assets (e.g., carbon emission rights, intellectual property rights, and art revenue rights) [15][16]. RWA Bond Operation Mechanism - Asset Confirmation and Pooling: Non - standardized assets use a consortium chain for multi - node verification to generate "on - chain ownership certificates" and then package them into an asset pool. Standardized financial assets are confirmed through a licensed custodian and directly mapped on - chain [19][21]. - Chain Mapping and Token Generation: Non - standardized assets need to disassemble and map rights and interests, and then issue tokens according to a special standard. Standardized financial assets are directly mapped and use a general - purpose token standard [22][25]. - Trading, Circulation, and Settlement: Non - standardized assets can be traded on compliant centralized or decentralized exchanges, and the settlement is completed through smart contracts. Standardized financial assets are traded on high - frequency platforms, and the settlement is also fast, with a T + 0 arrival [26][30]. - Income Distribution and Asset Monitoring: For non - standardized assets, IoT and oracles are used to collect data and distribute income automatically. For standardized financial assets, the income distribution is more simplified, and the asset monitoring focuses on price and ownership risks [31][34]. Comparison with Traditional Bonds and ABS - Compared with ABS: RWA bonds are more efficient and intelligent, with an automated operation mode, 7×24 - hour second - level settlement, full - chain real - time transparency, and a lower investment threshold [2][38]. - Asset Feature Comparison: RWA bonds have a wider range of underlying assets, a more advanced technological foundation, and higher information transparency than traditional bonds and ABS. However, traditional bonds have a more mature regulatory system, and ABS has a more established approval process [44][45]. Global RWA Bond Policy and Regulatory Framework - US: It has evolved from technology exploration to institutional dominance. The CLARITY Act provides a dynamic regulatory framework, and technological upgrades and institutional capital inflows have driven market growth [48][50]. - Europe: It has moved from infrastructure pilots to a unified MiCA framework. MiCA clarifies the legal attributes of RWA tokens and provides a unified license, while technological improvements meet institutional requirements [51][54]. - China: It features a dual - track approach of cautious pilot projects in regions excluding Hong Kong, Macao, and Taiwan, and international linkage in Hong Kong. Through rule recognition, asset interconnection, and technological interconnection, a unique RWA ecosystem is being built [55][59]. RWA Typical Cases - Shenzhen Futian RWA Digital Bond: Issued in 2025 with a scale of 500 million yuan and a coupon rate of 2.62%, it is the world's first public - offering RWA bond, listed on both the Macau Exchange and the Shenzhen Stock Exchange, with underlying assets of charging piles and office building revenue rights [3][60]. - Langxin Group Charging Pile RWA: In 2024, it was a Hong Kong sandbox project with a scale of 100 million yuan, anchoring the revenue rights of over 9000 charging piles, and was the first domestic new - energy RWA [3][63]. - GCL New Energy Photovoltaic Power Station RWA: In 2024, it issued 200 million digital tokens corresponding to the revenue rights of an 82MW photovoltaic power station, bundling carbon - reduction benefits, and using a "two - chain and one - bridge" architecture [3][64]. RWA Bond Investor Participation Path and Threshold - Primary Market Subscription: Institutions in regions excluding Hong Kong, Macao, and Taiwan can directly connect with overseas licensed underwriters or indirectly subscribe through domestic QDII products [68]. - Secondary Market Trading: They can trade through the Hong Kong MOX, the Shenzhen Stock Exchange Cross - border Connect, or licensed digital exchanges, with different risk - level bond trading requirements [69][70].

Report Industry Investment Rating - Not provided in the given content Core Viewpoints - The report presents data on government bond net financing, including the net financing amounts of government bonds, national bonds, local bonds, etc. in the 38th week (9/15 - 9/21) and the 39th week (9/22 - 9/28), as well as their cumulative amounts and progress compared to the previous year [1][6]. - As of September 24, the planned issuance of local bonds in the fourth quarter is 730 billion yuan, including 40 billion yuan of new general bonds and 390 billion yuan of new special bonds. The remaining quotas for new general bonds and special bonds are 140 billion yuan and 740 billion yuan respectively [2][9]. - The issuance progress of special refinancing bonds is 99%, and the net financing of urban investment bonds is expected to be -7 billion yuan in the 39th week, with a balance of 10.2 trillion yuan as of this week [3][27]. Summary by Related Catalogs Government Bond Net Financing - In the 38th week (9/15 - 9/21), the net financing of government bonds was 317.9 billion yuan, and in the 39th week (9/22 - 9/28), it was -22 billion yuan. As of the 38th week, the cumulative amount was 11.5 trillion yuan, exceeding the same period last year by 5.1 trillion yuan [1][6]. - The net financing of national bonds + new local bond issuance in the 38th week was 405.6 billion yuan, and in the 39th week, it was 10.8 billion yuan. As of the 38th week, the cumulative generalized deficit was 9.7 trillion yuan, with a progress of 81.9% [1][6]. National Bonds - The net financing of national bonds in the 38th week was 287.1 billion yuan, and in the 39th week, it was -144.4 billion yuan. The annual net financing of national bonds is 6.66 trillion yuan. As of the 38th week, the cumulative amount was 5.5 trillion yuan, with a progress of 83.2% [1][7]. Local Bonds - The net financing of local bonds in the 38th week was 30.9 billion yuan, and in the 39th week, it was 122.5 billion yuan. As of the 38th week, the cumulative amount was 6.0 trillion yuan, exceeding the same period last year by 2.7 trillion yuan [1][9]. - As of September 24, the planned issuance of local bonds in the fourth quarter is 730 billion yuan, including 40 billion yuan of new general bonds and 390 billion yuan of new special bonds. The remaining quotas for new general bonds and special bonds are 140 billion yuan and 740 billion yuan respectively [2][9]. New General Bonds - In the 38th week, the issuance of new general bonds was 20.7 billion yuan, and in the 39th week, it was 5.6 billion yuan. As of the 38th week, the cumulative amount was 656.2 billion yuan, with a progress of 82.0%, exceeding the same period last year [2][11]. New Special Bonds - In the 38th week, the issuance of new special bonds was 97.8 billion yuan, and in the 39th week, it was 149.6 billion yuan. The planned issuance of new special bonds in 2025 is 4.4 trillion yuan. As of the 38th week, the cumulative amount was 3.5 trillion yuan, with a progress of 79.8%, exceeding the same period last year [14]. - The special new special bonds issued reached 1201.2 billion yuan, of which 233.3 billion yuan was issued since September. The land reserve special bonds issued were 346.6 billion yuan. As of September 23, 27 provincial - level administrative regions had announced relevant projects, covering 4992 parcels of land, with a capital scale of 553.7 billion yuan [2][14]. Special Refinancing Bonds - In the 38th week, the issuance of special refinancing bonds was 21.4 billion yuan, and in the 39th week, it was 11.4 billion yuan. As of the 38th week, the cumulative amount was 1.98 trillion yuan, with a issuance progress of 99% [3][27]. Urban Investment Bonds - The net financing of urban investment bonds in the 38th week was 15.5 billion yuan, and it is expected to be -7 billion yuan in the 39th week. As of this week, the balance of urban investment bonds was 10.2 trillion yuan [3][27].

Investment Rating - The investment rating for the company is "Outperform the Market" [6][20][23] Core Insights - The company reported a revenue of 683 million yuan for the first half of 2025, representing a year-on-year growth of 30.68%, and a net profit of 146 million yuan, which is a 28.80% increase year-on-year [1][7] - The growth in performance is primarily driven by the implementation of industrial AI technology in the 3C industry and the recovery in the lithium battery sector, with expectations for continued revenue growth as demand in the semiconductor and automotive industries increases [1][2] - The company has established a dual-track strategy focusing on "industrial automation + intelligent robotics," aiming to become a core supplier of visual components for robotic perception [3][2] Summary by Sections Financial Performance - In Q2 2025, the company achieved a revenue of 363 million yuan, a 32.15% increase year-on-year, and a net profit of 73 million yuan, up 29.45% year-on-year [1] - The gross margin and net margin for the first half of 2025 were 65.47% and 21.75%, respectively, showing stability in profit margins [1][15] - The company’s expenses for sales, management, R&D, and financial costs were 19.61%, 3.39%, 18.93%, and -0.95%, with slight year-on-year changes [1] Product Development - The company has enhanced its core product layout in machine vision and expanded into industrial sensors and intelligent robotics [2] - Revenue from the 3C industry was 439 million yuan, a 23.82% increase, while the lithium battery sector saw a revenue of 167 million yuan, growing by 49.35% [2] - The company has launched upgraded visual software products and introduced advanced technologies in its product lines, solidifying its technological advantages [2] Strategic Initiatives - The establishment of a robotics division in the first half of 2025 aims to cover key visual components required for robotic perception, with plans for a new production facility in East China [3] - The company targets a revenue growth rate of no less than 20% for 2025, based on its 2024 revenue [3] - The company maintains a strong focus on R&D investment, which is expected to support long-term growth [15][17]

Investment Rating - The investment rating for the company is "Outperform the Market" [6][20]. Core Viewpoints - The company experienced a decline in revenue and profit in the first half of 2025, with revenue down 9.71% year-on-year to 263 million yuan and net profit down 27.82% to 43 million yuan, primarily due to weak downstream demand and a decline in consumables sales [2][3]. - Despite the challenges, the company has made significant progress in semiconductor equipment research and development, completing several new projects aimed at high-end markets [3][4]. - The company is expanding its production capacity and enhancing its overseas presence, with 19.87% of its revenue coming from international sales [4]. Financial Performance Summary - In the first half of 2025, the company's gross margin was 31.07%, and net margin was 16.21%, both showing a decline compared to the previous year [2]. - The company’s revenue from vacuum coating equipment reached 234 million yuan, a year-on-year increase of 9.52%, while revenue from the consumer electronics sector fell significantly [3]. - The company has a robust pipeline of new products and has established a strong customer base, including major clients like Apple and BYD [4]. Earnings Forecast and Financial Metrics - The company is projected to achieve net profits of 103 million yuan, 161 million yuan, and 286 million yuan for the years 2025, 2026, and 2027, respectively, with corresponding PE ratios of 111, 71, and 40 [5][21]. - Revenue is expected to grow significantly in the coming years, with a forecasted increase of 22.4% in 2025 and 44.1% in 2026 [5][21]. - The company maintains a high level of R&D investment, which is crucial for its long-term growth and competitiveness in the high-end equipment market [4][21].

Core Insights - The report emphasizes the need for a multi-dimensional, AI-driven framework to effectively predict and manage risks associated with short-term market surges, particularly in the context of the A-share market [2][3] - It introduces a comprehensive multi-factor system based on four core dimensions: trend, momentum, capital flow, and valuation, which collectively enhance market state characterization [2][4] - The AI-enhanced multi-factor timing strategy is expected to provide investors with an objective risk warning tool, reducing losses from blind chasing of high prices [3][4] Trend Analysis - The report illustrates that traditional technical indicators often fail to provide timely warnings for rapid market fluctuations driven by emotions rather than fundamentals [2][6] - It highlights the "9·24" market surge as a case study, where the index rose over 21% in a short period, demonstrating the risks of emotion-driven trading [5][6] - The analysis of moving averages (MA5, MA10, MA20) indicates that a bullish trend was confirmed before the surge, while subsequent signals suggested a weakening momentum [6][8] Momentum Indicators - The report discusses the use of KDJ and RSI indicators, which reached extreme levels during the "9·24" surge, signaling potential overbought conditions [8][9] - It notes that these momentum extremes often occur at the end of price waves, serving as critical signals for potential market tops [9][10] Capital Flow Insights - The report emphasizes the correlation between trading volume and price movements, indicating that significant increases in trading volume often precede price surges [11][12] - It also points out that a decline in trading volume following a price peak can signal weakening momentum and potential market corrections [12] Valuation Metrics - The report highlights the rapid increase in the price-to-earnings (PE) ratio during the "9·24" surge, indicating a shift from undervaluation to overvaluation, which raises risk concerns [15][16] - It suggests that high PE ratios, especially when combined with momentum indicators showing overbought conditions, serve as strong signals for potential market corrections [15][16] AI-Driven Quantitative Strategy - The report outlines a comprehensive AI-driven quantitative strategy that integrates various data sources and employs machine learning algorithms to enhance decision-making [19][20] - It emphasizes the importance of feature engineering and factor processing to ensure the robustness and interpretability of the model [20][33] - The strategy's backtesting results indicate a significant annualized return of 36.41% with a Sharpe ratio of 2.30, outperforming the market benchmark [41][42] Performance Evaluation - The strategy demonstrated strong performance during market uptrends while effectively managing drawdowns during downturns, showcasing its risk management capabilities [42][45] - The report notes that the model's predictive accuracy, while modest, indicates its ability to identify market trends better than random guessing [51][56]

Core Insights - The report emphasizes the need for a multi-dimensional, AI-driven framework to effectively predict and manage risks associated with short-term market surges, particularly in the context of the A-share market [2][3] - It introduces a comprehensive multi-factor system based on four core dimensions: trend, momentum, capital flow, and valuation, which collectively enhance market state characterization [2][4] - The AI-enhanced multi-factor timing strategy is expected to provide investors with an objective risk warning tool, reducing losses from blind chasing of high prices [3][4] Trend Analysis - The report illustrates that traditional indicators often fail to provide timely warnings for rapid market fluctuations driven by emotions rather than fundamentals [2][6] - The analysis of the "9·24" market surge shows that moving averages indicated a bullish trend before the surge, while subsequent signals indicated a weakening momentum [5][6][8] Momentum Indicators - The report highlights that extreme values in momentum indicators like KDJ and RSI often signal the end of a price surge, as seen during the "9·24" event where both indicators reached overbought levels [8][9] - The KDJ and RSI thresholds serve as critical points for identifying market cycles, aiding investors in timing their trades effectively [9] Capital Flow Insights - The report notes a strong correlation between trading volume and price movements during the "9·24" surge, indicating that volume often precedes price increases [11][12] - A decline in trading volume following price peaks serves as a warning signal for potential market corrections, as evidenced in the analysis [12] Valuation Metrics - The report discusses how valuation metrics, such as PE ratios, can indicate market risk accumulation, particularly when they exceed historical high thresholds [15][16] - The combination of high valuation levels and overbought momentum indicators has historically signaled market tops and subsequent corrections [15] AI-Driven Quantitative Strategy - The report outlines a comprehensive AI-driven quantitative strategy that automates the process of factor selection, modeling, and execution, enhancing the robustness of trading signals [19][20] - The strategy employs a closed-loop system that continuously optimizes itself based on real-time performance feedback, ensuring adaptability to changing market conditions [19][20] Factor Processing and Model Selection - The report emphasizes the importance of factor processing, including standardization and ranking, to ensure comparability and robustness of the indicators used in the model [30][33] - The HistGradientBoosting model is selected for its ability to capture non-linear relationships among factors, providing a more accurate timing signal for trades [39][40] Performance Evaluation - Backtesting results indicate that the AI-driven strategy significantly outperforms the market benchmark, achieving an annualized return of approximately 36.41% with a Sharpe ratio of 2.30 [41][42] - The strategy demonstrates strong risk management capabilities, maintaining a maximum drawdown of -19.51%, which is notably lower than the benchmark during volatile periods [45][46]

Investment Rating - The investment rating for the industry is "Outperform the Market" (maintained) [1] Core Viewpoints - The rapid development of efficient cooling technologies is driven by the increasing scale and power of data centers, with liquid cooling addressing high energy consumption and heat generation issues. By 2024, China's computing centers are expected to consume 166 billion kWh, accounting for 1.68% of the total electricity consumption in society [2] - The main liquid cooling technologies are cold plate and immersion cooling, with cold plate cooling being the most widely used due to its mature technology. Liquid cooling offers multiple advantages such as high heat exchange efficiency, energy savings, and reduced noise [2] - The market for cooling liquids is projected to grow significantly, with an estimated demand of approximately 89,000 tons of cooling liquid driven by the addition of AI data centers by 2028 [2] Summary by Sections 1. Main Cooling Pathways - Liquid cooling technology is categorized into indirect cooling (mainly cold plate) and direct cooling (including immersion and spray cooling). Cold plate cooling is the most widely used, while immersion cooling offers higher heat dissipation capabilities but at a higher operational cost [2][23] 2. Water-based Cooling Liquids - Water-based cooling liquids are primarily used in single-phase cold plate cooling, offering good thermal conductivity and low cost. However, they may pose risks of algae and bacteria growth [48][50] 3. Oil-based Cooling Liquids - Oil-based cooling liquids, including mineral oil and synthetic oil, are mainly used in single-phase immersion cooling. They are cost-effective but have drawbacks such as high viscosity and cleaning difficulties [53][56] 4. Fluorinated Cooling Liquids - Fluorinated cooling liquids are used in both single-phase and multi-phase immersion cooling. They offer excellent thermal properties and low toxicity but are relatively expensive [2][56] 5. Market Space Estimation - The liquid cooling market in China is expected to exceed 100 billion yuan by 2027, with a compound annual growth rate of 51.4% from 2019 to 2027 [17][18] 6. Commercialization of Liquid Cooling Technology - The commercialization of liquid cooling technology is increasing, with major IT companies like Alibaba, Baidu, and Tencent implementing successful liquid cooling applications [27][28]

Group 1: Market Dynamics - The core driver of the current rise in equity assets is not due to macro liquidity excess but rather a recovery in risk appetite since the "anti-involution" policy was implemented[5] - The market is primarily driven by internal fund reallocations and leverage rather than large-scale inflows from external funds[5] - The correlation between stocks and bonds has shifted to a "see-saw" effect, indicating that growth factor changes are now dominant, contrasting with the liquidity-driven environment of 2015[13] Group 2: Price Stabilization and PPI Insights - Price stabilization is expected to continue into Q4, supported by significant differentiation in pricing between domestic and external demand[5] - The Producer Price Index (PPI) is influenced by overseas inflation, with a notable divergence between Chinese and U.S. PPI trends[25] - The PPI gap between different industries, such as non-ferrous and ferrous metals, has reached 20%, a historically unprecedented level[25] Group 3: Fund Flows and Market Sentiment - As of September 14, 2025, new equity fund sales reached 42.85 billion units, a significant increase from less than 10 billion units at the beginning of 2024, although still below the peak levels seen in 2015 and 2021[18] - The margin ratio for internal funds reached 294.17% on August 24, 2025, nearing historical peak levels, indicating high leverage in the market[18] Group 4: Future Projections - If capacity utilization rises above 75%, a 1.35% increase is expected, with a corresponding price increase of approximately 1.5% due to the price elasticity of capacity utilization[40] - The stock market's upward trend since September 2024 is compared to the 1999 market rally, suggesting a potential further increase of around 30% if the current trajectory continues[55][58]

证券研究报告 | 2025年09月25日 AI 赋能资产配置（十七） AI 盯盘： "9·24"行情案例 策略研究·策略解读 | 证券分析师： | 王开 | 021-60933132 | wangkai8@guosen.com.cn | 执证编码：S0980521030001 | | --- | --- | --- | --- | --- | | 证券分析师： | 陈凯畅 | 021-60375429 | chenkaichang@guosen.com.cn | 执证编码：S0980523090002 | 事项： 金融市场中的短期内快速上涨行情往往因情绪驱动而非基本面改善，容易导致阶段性追高。传统技术分析 （如 KDJ、RSI、MACD、均线体系、成交量、换手率及估值水平等单一指标）虽能提供部分洞察，但其 信号纷杂、滞后性强且受主观经验影响较大，难以有效预警此类脉冲式行情的风险。 为系统性地解决这一问题，本研究旨在构建一个多维度、量化、由人工智能驱动的综合研判框架。研究首 先从趋势、动量、资金流向、估值四个核心维度出发，构建了十二个关键原始指标，形成一个全面刻画市 场状态的多因子体系，并初步判断和市场趋势的关 ...

Investment Rating - The report maintains an "Outperform" rating for the industry [1] Core Insights - The rapid growth of data centers and the increasing power demands necessitate efficient cooling technologies, with liquid cooling addressing high energy consumption and heat generation issues. By 2024, China's computing centers are expected to consume 166 billion kWh, accounting for 1.68% of the total electricity consumption [2][2] - The main liquid cooling technologies include cold plate and immersion cooling, with cold plate cooling being the most widely used due to its mature technology. Liquid cooling offers multiple advantages such as high heat transfer efficiency, energy savings, and reduced noise [2][2] - The market for cooling liquids is projected to grow significantly, with an estimated demand of approximately 89,000 tons for AI data centers by 2028, alongside 19,000 tons from traditional server installations [2][2] Summary by Sections 1. Main Cooling Paths - Liquid cooling technology is categorized into indirect cooling (mainly cold plate) and direct cooling (including immersion and spray cooling). Cold plate cooling is the most prevalent, while immersion cooling offers higher heat dissipation capabilities but comes with higher operational costs [2][2][2] 2. Water-Based Cooling Liquids - Water-based cooling liquids are primarily used in single-phase cold plate cooling, offering good thermal conductivity and low cost. However, they may pose risks of microbial growth [50][50] 3. Oil-Based Cooling Liquids - Oil-based cooling liquids, including mineral oil and synthetic oil, are mainly applied in single-phase immersion cooling. They are cost-effective and environmentally friendly but have lower thermal conductivity compared to synthetic alternatives [56][56] 4. Market Space Estimation - The liquid cooling market in China is expected to exceed 100 billion yuan by 2027, with a compound annual growth rate of 51.4% from 2019 to 2027 [17][17] 5. Commercialization of Liquid Cooling Technology - The commercialization of liquid cooling technology is increasing, with major IT companies like Alibaba, Baidu, and Tencent implementing successful liquid cooling applications [27][27]