1858 年的夏天，一根铜芯电缆横穿大西洋海底，把伦敦和纽约连在了一起。 这件事的意义从来不在于传输速度，而在于权力结构，谁铺设了海底电缆，谁就能在信息流动中抽水。大英帝国靠着这张全球电报网，把殖民地的情报、 棉花的价格、战争的消息都攥在手里。 帝国的强大不仅是有舰队，还有那根电缆。 一百六十多年后，这个逻辑正在以一种意想不到的方式重演。 AI 模型 大迁徙 2026 年 2 月 24 日，OpenRouter 发布了一份周度数据：平台前十模型的总 Token 消耗约 8.7 万亿，中国模型独占 5.3 万亿，占比 61%。MiniMax M2.5 以 2.45 万亿 Token 空降榜首，Kimi K2.5、智谱 GLM-5 紧随其后，前三名清一色来自中国。 | | & LLM Leaderboard | | | | | | This Week <> | | --- | --- | --- | --- | --- | --- | --- | --- | | | | Compare the most popular models on OpenRouter (1) | | | | | | | 1 | ® | ...

证券研究报告 | 2026年02月27日 2026年02月28日 香港交易所（00388.HK） 优于大市 香港市场景气延续，盈利持续高增 2025 年业绩概览：市场景气驱动高增长，盈利质量突出。公司 2025 年实现 营业收入 291.61 亿港元，同比增长 30.3%；实现归母净利润 177.54 亿港元， 同比增长 36.0%。业绩增长的核心驱动力来自于港股市场交易活跃度的大幅 提升。2025 年现货市场日均成交额（ADT）达到 2,315 亿港元，同比大幅增 长 92.9%。公司 EBITDA 利润率攀升至 78.2%的历史新高，年化净资产收益率 （ROE）达到 31.7%，超越 2015 年高点，凸显其轻资产运营模式下的强大规 模效应与高盈利质量。 | 盈利预测和财务指标 | 2024 | 2025 | 2026E | 2027E | 2028E | | --- | --- | --- | --- | --- | --- | | 营业收入(百万港元) | 22,374 | 29,161 | 31,886 | 34,302 | 37,604 | | (+/-%) | 9.1% | 30.3% | 9. ...

证券研究报告——煤炭行业周报 供需格局优化，动力煤价稳中有升 【2026.2.9-2026.2.15】 行业评级：看好 核心观点 发布日期：2026.2.25 煤炭行情走势图 分析师：景剑文 执业证书编号：S0770523090001 邮箱：jingjw@dtsbc.com.cn 风险提示 研究助理：温慧 执业证书编号：S0770125070012 邮箱：wenh@dtsbc.com.cn 地址：山西太原长治路 111 号山 西世贸中心 A 座 F12、F13 网址:http://www.dtsbc.com.cn 请务必阅读最后一页免责声明 1 终端需求大幅回落，板块轮动加快，板块突发利空。 动力煤方面，本周动力煤各环节价格呈现震荡上行态势。港口 端、产地端与国际端价格均上涨。短期来看，节后上下游复工 复产推进，叠加库存处于低位，煤价有望延续偏强运行，中长 期在供给约束与需求韧性支撑下，价格将向合理区间修复，维 持窄幅震荡。 数据来源：Wind 大同证券 炼焦煤方面，本周炼焦煤各环节价格呈现弱稳分化态势。港口 端价格周环比持平，产地端价格稳中有降，国际端澳洲峰景矿 硬焦煤 FOB 价周环比小幅下降。整体来看， ...

2月7日，南海区域海—气双向耦合智能大模型"飞鱼—1.0"在广州正式发布。该大模型由中国科学 院南海海洋研究所与中国石油大学（华东）联合研发，为全球首个面向南海的海—气双向耦合智能大模 型。 据介绍，"飞鱼—1.0"大模型实现了3项核心技术创新，包括核心数据自主可控、海—气双向智能耦 合、"即插即用"式低成本学习与灵活可扩展。该大模型训练所用的核心数据采用中国科学院南海海洋研 究所自主研制的高分辨率南海再分析数据集，打破了以往国内海洋大模型高度依赖欧美再分析数据的局 面，并使大模型具备了精准刻画海洋内波、锋面等小尺度过程的能力。 前期AI大模型大多是单纯的大气大模型或海洋大模型，大气与海洋之间是相对独立的。"飞鱼— 1.0"大模型将物理机理与AI深度融合，通过"快慢"双通道学习，智能化模拟海气之间动量、热量等的双 向交互影响，显著提升了海气关键要素预报精度。 "飞鱼—1.0"大模型首创面向海气要素预报的多专家系统，能根据不同预报任务智能调用最合适的 计算模块，大幅降低了学习能耗，同时可针对新海域或新任务实现模块化功能扩展。该大模型不仅在温 盐等核心要素预报上显著优于国际主流再分析产品，还具备精准刻画大尺度环流 ...

证券分析师 022-28451618 SAC NO：S1150511010016 cuijian@bhzq.com 晨会纪要（2026/02/06） 市场处于震荡整固阶段，结构性行情将是重点——A 股市场 2026 年 2 月投资 策略报告 行业研究 关注年报业绩催化与创新药械产业链——医药生物行业 2 月投资策略展望 崔健 渤海证券研究所晨会纪要（2026.02.06） 宏观及策略研究 晨会纪要（2026/02/06） 宏观及策略研究 市场处于震荡整固阶段，结构性行情将是重点——A 股市场 2026 年 2 月投资策略报告 请务必阅读正文之后的声明 渤海证券股份有限公司具备证券投资咨询业务资格 1 of 6 要 宋亦威（证券分析师，SAC NO：S1150514080001） 云计算提价印证高景气，AI 应用迎来多点催化——计算机行业 2 月投资策略 展望 证 券 研 究 报 告 晨 会 纪 严佩佩（证券分析师，SAC NO：S1150520110001） 靳沛芃（研究助理，SAC NO：S1150124030005） 1、宏观方面 2026 年年初，出口端在结构优化、产品竞争优势以及经贸环境改善的支撑下将 ...

项目简介：当前中国东方资产管理股份有限公司对于智能问答、知识检索、业务流程自动化等智能化应用需求持续增长。现有硬件资源无法支持大模型运 行，亦缺乏统一的智能应用平台。将通过本次招标建设基础算力平台和智能应用平台，并构建试点应用，为后续智能化建设奠定底座。 中标候选人公示 2026 年 2 月 2 日发布中标候选人公示，讯飞智元信息科技有限公司 3510000 元（中）。 中标候选人： | 排名 | 中标候选人名称及投标报价 | | --- | --- | | 第一中标候选人 | 讯飞智元信息科技有限公司, 3510000.00 元 | | 第二中标候选人 | 软通动力信息技术(集团)股份有限公司, | | | 2980000.00 元 | | 第三中标候选人 | 南京云问网络技术有限公司, 3920000.00 元 | 项目简介：当前中国东方资产管理股份有限公司对于智能问答、知识检索、业务流程自动化等智能化应用需求持续增长。现有硬件资源无法支持大模型运 行，亦缺乏统一的智能应用平台。将通过本次招标建设基础算力平台和智能应用平台，并构建试点应用，为后续智能化建设奠定底座。 中标候选人公示 2026 年 2 月 ...

Summary of Weimeng's GU Product Conference Call Industry and Company Overview - The conference call discusses Weimeng's GU product, which is designed to optimize advertising effectiveness through improved click-through and conversion rates, rather than direct competition with traditional advertising methods [2][6]. - Weimeng serves a wide range of industries, including private hospitals and high-precision instrument manufacturing, indicating a broad market potential [2][7]. Core Insights and Arguments - **Product Pricing and Structure**: - The GU product is charged based on the number of scenarios and questions, with a basic version priced at approximately 10,000 yuan per quarter or 20,000 yuan annually, and an advanced version at about 30,000 yuan per quarter or 60,000 yuan annually [2][3][4]. - The advanced version includes additional features such as public opinion monitoring, which helps clients manage negative sentiment [3][5]. - **Market Potential**: - Weimeng has a client base of around 60,000 to 70,000 advertisers and has accumulated 200,000 advertisers since 2016. If each client purchases the standard version at 20,000 yuan, the market size could reach 4 billion yuan; if they opt for the advanced version at 60,000 yuan, the market size could expand to 12 billion yuan [4][9]. - The gross margin for Weimeng is maintained at over 60%, indicating a strong profitability potential [4][10]. - **Effectiveness and ROI**: - The GU product operates on a performance-based payment model, where clients are charged only when their brand appears in the top three positions on specified platforms, ensuring they pay for actual results [8]. - For instance, a well-known private hospital spends several hundred million yuan annually on marketing. A mere 0.2% to 0.3% increase in ROI could justify the 1 million yuan annual fee for the GU product, showcasing its sustainability and effectiveness [8]. Additional Important Insights - **Competitive Advantage**: - Weimeng's core advantages include a deep understanding of large models and industry needs, extensive industry experience, broad media platform integration, and comprehensive management capabilities [10]. - The company plans to focus on existing clients while gradually expanding to new clients, leveraging trust and familiarity in the advertising space [10]. - **Industry Applications**: - The GU product is applicable across various sectors, with order amounts varying significantly based on the complexity of scenarios, ranging from 100,000 to 1 million yuan annually [7]. - The product is not seen as a competitor to traditional advertising but rather as a tool to enhance advertising effectiveness, thereby increasing the value for businesses that invest heavily in advertising [6]. This summary encapsulates the key points discussed in the conference call regarding Weimeng's GU product, its market potential, pricing structure, effectiveness, and competitive advantages.

Investment Rating - The report maintains a "Neutral" rating for the computer industry and an "Overweight" rating for Hongsoft Technology (688088) [41] Core Insights - The "Artificial Intelligence + Manufacturing" policy has been jointly issued by eight departments, marking a significant step towards the integration of AI and manufacturing [7] - Nvidia has announced the full production of its Rubin chip platform, which shows significant performance improvements over the previous Blackwell architecture [10][11] - The software industry in China has shown robust growth, with business revenue reaching 139,777 billion yuan, a year-on-year increase of 13.3% [16] Industry News - The "Artificial Intelligence + Manufacturing" initiative aims to achieve reliable supply of core AI technologies and maintain a leading position in industrial scale and empowerment by 2027 [7] - Super Fusion, a leading computing power company, has initiated its listing guidance [7] - Nvidia's Rubin platform integrates six chips and boasts a peak computing power of 50 Petaflops, with significant improvements in energy efficiency [10] Industry Data - In November 2025, the Producer Price Index (PPI) for the computer industry decreased by 0.2% month-on-month and 0.5% year-on-year [12] - From January to November 2025, the cumulative output of electronic computers in China was 31,690.5 million units, a decrease of 1.1% year-on-year [12] - The software industry reported a total profit of 16,954 billion yuan, reflecting a year-on-year growth of 6.6% [16] Company Announcements - Haohan Deep announced an investment and acquisition agreement to gain control of 54% of Cloud Edge Cloud Technology [28] - Nengke Technology plans to raise up to 1 billion yuan through a private placement to fund various AI projects [29][30] Market Review - From December 1 to December 31, the Shenwan Computer Industry index fell by 0.25%, with mixed performance across sub-sectors [31] - As of December 31, 2025, the price-to-earnings ratio for the Shenwan Computer Industry was 208.15 times, with a premium of 1447.27% compared to the CSI 300 [32] Monthly Strategy - The report emphasizes the potential of AI computing power and applications, particularly with the advancements from Nvidia and the new policy initiatives [40] - It suggests focusing on leading companies that have strong capabilities in AI technology implementation and scene adaptation [40]

Summary of Conference Call on Optical Module Industry Industry Overview - The optical module industry is benefiting from data center upgrades and the demand for AI large models, with significant growth expected from 2025 to 2028 due to the increasing need for 800G optical modules driven primarily by AI applications [1][2] Key Points and Arguments - **Growth Projections**: The demand for optical modules, particularly 800G and 1.6T, is expected to experience explosive growth, with 800G module shipments projected to double from 2025 to 2026 and 1.6T modules reaching a shipping peak in 2026 [2] - **Market Drivers**: Approximately 20%-30% of the growth in 800G optical modules is attributed to traditional data center upgrades, while around 70% is driven by AI model training and inference [4] - **Technological Adoption**: Major companies like Google and NVIDIA are adopting Skyop designs, which will significantly increase the demand for high-speed optical modules as they replace traditional copper connections [5] - **Silicon Photonics Growth**: Silicon photonics is expected to capture 20%-30% of the single-mode optical module market by 2025, doubling to 40%-50% by 2026 due to the maturity of 800G silicon photonics products and a shortage of 100G EML chips [8] - **Custom Solutions**: New customized optical modules such as LPO, LRO, and TRO are being developed to reduce power consumption and improve efficiency in specific applications [9] Additional Important Insights - **Supply Chain Challenges**: The optical module industry faces supply chain issues, particularly with the shortage of 100G EML chips, which is expected to persist until 2027-2028 [3][12] - **NPU and CPO Technologies**: NVIDIA and Broadcom are pushing CPO solutions, but the complexity and high costs of co-packaged structures limit widespread adoption. NPU technology aims to integrate optical modules within switches to reduce signal transmission distances [10] - **OCS Technology**: Google is leading the application of OCS technology for large-scale interconnects, with other companies like Meta and AWS following suit. The adoption of OCS is expected to outpace CPO in the coming years [11] - **PCB Material Supply Issues**: The new PCB materials used in 1.6T projects are experiencing supply shortages, necessitating earlier procurement compared to previous materials [17] Conclusion The optical module industry is poised for significant growth driven by advancements in AI and data center technologies, although it faces challenges related to supply chain constraints and the adoption of new technologies. The shift towards silicon photonics and customized solutions will play a crucial role in meeting the increasing demand.

Investment Rating - The report assigns a "Buy" rating for the company, marking its first coverage [1]. Core Insights - The company is a leading provider of integrated smart cockpit solutions, having successfully transitioned from a focus on vehicle networking systems to smart cockpit solutions since 2018, solidifying its position in the industry [8][13]. - The penetration rate of smart cockpits in domestic and global passenger vehicles is steadily increasing, with projections indicating a rise in China's market penetration from 35.3% in 2019 to 75.9% by 2025, and globally from 38.4% to 59.4% in the same period [8][42]. - The company's revenue growth is primarily driven by the increasing orders for high-end domain controllers, with expectations of a turnaround in net profit by 2025 [8][49]. Summary by Sections Company Overview - The company was established in 2009 and initially focused on vehicle networking systems, later shifting its focus to integrated software, hardware, and cloud services for smart cockpit solutions [8][13]. - It has established six R&D centers and three manufacturing bases, serving over 50 automotive brands, with Li Auto being a key customer driving revenue growth [8][22]. Market Position and Growth - The domestic and global market for smart cockpit solutions is rapidly expanding, with the Chinese market expected to grow from 129 billion RMB in 2024 to 299.5 billion RMB by 2029, reflecting a compound annual growth rate (CAGR) of 18.4% [42][48]. - The company has formed deep partnerships with Qualcomm and Huawei, enhancing its competitive edge in high-end domain controller products [8][56]. Financial Projections - Revenue forecasts indicate significant growth, with total revenue expected to reach 34.58 billion RMB in 2025, 54.52 billion RMB in 2026, and 78.39 billion RMB in 2027, alongside a projected net profit turnaround in 2025 [1][8]. - The company has shown a consistent increase in revenue since 2021, with a compound annual growth rate of 43.5% from 2021 to 2024 [26][30]. Product Development and Strategy - The company emphasizes a "software + hardware + cloud" self-research system, developing core technologies that support the scalability and customization of its smart cockpit solutions [61]. - The high-end domain controller product ratio is increasing, with shipments rising from 1.5% in 2023 to 51.1% in early 2025, indicating a shift towards higher-value products [64].