天风通信 | 润泽科技：AIDC&算力中心规模交付推动发展，国产算力空间广阔AIDC龙头紧抓机遇 润泽科技今日大涨+14%，公司AIDC+IDC+液冷领先，智算中心持续交付规模提升，同时算力业务持续布局推 进，受益国内大客户需求放量业绩高增动能强。 1、AIDC推动持续增长，算力业务有望带来增量。 1、AIDC推动持续增长，算力业务有望带来增量。 公司25年归母净利润高速增长，主要系发行公募REITs产生的非经常性 天风通信 | 润泽科技：AIDC&算力中心规模交付推动发展，国产算力空间广阔AIDC龙头紧抓机遇 润泽科技今日大涨+14%，公司AIDC+IDC+液冷领先，智算中心持续交付规模提升，同时算力业务持续布局推 进，受益国内大客户需求放量业绩高增动能强。 公司25年归母净利润高速增长，主要系发行公募REITs产生的非经常性损益所致；业务上，公司AIDC业务推动发 展，25年新增算力中心交付约220MW，且顺利交付行业领先的单体100MW智算中心，并推动液冷技术应用。 公司①储备在手能耗资源充沛行业领先；②单体AIDC规模行业领先；③体内算力业务或带来显著增量。 随着公司全国化布局的持续落地，终端客户结构 ...

盈利端虽受战略性重组一次性费用影响，净利润同比有所下滑，但剔除非经营性项目后，调整后归母净 利润达 5.9 亿美元，同比增长 36%，调整后净利润率提升至 2.7%，核心盈利表现亮眼。成本端，公司 费用管控成效显著，费用 / 收入比同比大幅下降 1.3 个百分点至 10.8%；研发端，在持续加大 AI 等核心 领域投入的同时，研发费用投放效率提升，研发费用同比增长 3% 至 6.4 亿美元，为长期发展筑牢技术 根基。 对于联想集团未来业绩，国投证券国际给出乐观预期，预计 2025/26 财年和 2026/27 财年的经调整净利 润分别为 18.2 亿美元(yoy+26.2%)和 21.0 亿美元(yoy+21.0%)。从长期财务预测来看，2026-2028 财年公 司收入将持续稳步增长，毛利率逐步回升，ROE 维持在 20% 以上的较高水平，盈利能力与运营效率持 续优化，存货、应收账款周转天数不断改善，整体财务状况向好。 2026 年 2 月 13 日，国投证券国际发布联想集团(992.HK)最新研报，公司 25/26 财年第三季度业绩交出 超亮眼答卷，收入突破财季历史最高纪录，所有业务集团均实现双位数增长， ...

格隆汇2月24日丨冰山冷热(000530.SZ)在互动平台表示，公司有液冷方面技术储备，其中数据中心已纳 入公司重点深耕的细分市场范围，2026年将强力拓展。 ...

Core Viewpoint - The recent performance of VRT's Q4 FY2025 financial results and future guidance is noteworthy, indicating strong growth and demand in AI infrastructure [1] Financial Performance - For Q4 FY2025, the company reported a 252% year-over-year increase in organic orders, with an order shipment ratio rising to 2.9 times and backlog orders growing to $15 billion, a 109% increase year-over-year [2] - Adjusted operating profit increased by 33% to $668 million, while adjusted free cash flow surged by 151% to $910 million, leading to a 22.55% stock price increase on the reporting day, reaching a historical high [2] Strategic Outlook - The company raised its full-year outlook, projecting FY2026 net sales between $12.25 billion and $13.75 billion, representing an approximate 28% year-over-year growth, and adjusted earnings per share of $5.97 to $6.07, a 43% increase [3] - The Q1 guidance indicates net sales of $2.5 billion to $2.7 billion and adjusted earnings per share of $0.95 to $1.01, reinforcing market expectations for sustained growth in AI computing infrastructure demand [3] Industry Position - VRT is recognized as a leading player in global data center infrastructure, closely tied to major clients like Nvidia and Google, with significant barriers in liquid cooling and high-voltage power technologies attracting institutional attention [4] - Analysts, such as those from Evercore, maintain a "buy" rating, reflecting high order visibility, improved profitability, and structural trends driven by AI demand [4]

Summary of Key Points from Conference Call Records Industry Overview - The conference call discusses the rapid growth of capital expenditure in the AI sector, particularly among major tech companies like Amazon and Meta, with projected capital expenditures reaching $180 billion and $125 billion respectively by 2026, indicating an overall growth rate of nearly 70% from $376.1 billion last year to $650 billion this year [1][2]. Core Insights and Arguments - **AI Investment Trends**: AI-related investments are expected to accelerate, with a projected growth of 55% in 2024, 65% in 2025, and potentially 70% in 2026. Investments include cloud computing and logistics, with over 70% of purchases directly related to AI [3]. - **Importance of Optical Modules**: Optical modules are becoming increasingly significant in AI infrastructure, with their share of AI capital expenditure rising from 3% to 6% and potentially reaching 20% in the future. The demand for higher bandwidth optical modules is expected to surge due to network bandwidth becoming a bottleneck for computing clusters [4]. - **Leading Companies in Optical Communication**: Companies like Xuchuang and NewEase are highlighted for their high cost-performance ratio and comprehensive industry chain layout, which positions them favorably for valuation reassessment as semiconductor content increases [5][6]. - **Transformer Exports**: The power shortage in North America is driving an increase in transformer exports from China, with significant orders being fulfilled. The delivery cycles for large transformers are long, making the certainty of orders from North America high [7]. - **Market Dynamics for Large Transformers**: The U.S. market for large transformers is experiencing supply tightness, with a shift in procurement needs from power systems to data centers, which may lower entry barriers for new suppliers [8]. - **AI Power Market Outlook**: The year 2026 is anticipated to be pivotal for the AI power market, with significant orders expected for both external high-voltage direct current (HVDC) and internal systems [9]. - **Data Center Impact on Energy Storage**: The development of data centers is expected to significantly increase the demand for energy storage solutions, with recommendations for companies like Sungrow, which have competitive advantages in the U.S. market [10]. - **Commercialization of AI**: 2026 is projected to be a landmark year for AI commercialization, driven by advancements in language and visual models, as well as increased competition among domestic internet giants [11]. - **CPU Demand Surge**: The AI era is expected to increase demand for CPUs due to the unpredictable resource consumption associated with sandboxed execution tasks [12]. Additional Important Insights - **Changes in the Gas Turbine Supply Chain**: The gas turbine industry is experiencing significant order growth, with lead times extending to 2030 for some manufacturers, indicating a tightening supply chain [13][14]. - **Opportunities in the Media Sector**: The media industry is identified as a key area for AI commercialization, with major companies expected to ramp up capital expenditures. Specific opportunities are noted in AI advertising, AI comics, and AI gaming [19]. - **Recommendations in the Electronics Sector**: Key areas for investment in the electronics sector include storage chips, PCB technology, liquid cooling technology, and CPU-related segments, with specific companies recommended for their growth potential [20]. This summary encapsulates the critical insights and trends discussed in the conference call, providing a comprehensive overview of the current state and future outlook of the AI and related industries.

Market Overview - In January 2026, global stock markets experienced a broad rally, with emerging markets outperforming developed markets. Chinese stocks performed moderately, while US stocks lagged behind [5][11] - The cyclical sector led the market, driven by rising commodity prices, particularly in energy and materials, while the financial sector underperformed [5][11] Core Themes - February marks a traditional bullish window for A-shares, with historical data showing a 76% probability of gains and an average increase of 3.4% [17] - Regulatory measures are limiting speculative trading, directing funds towards sectors with solid fundamentals. The focus is on core sectors with improving profitability, particularly in cyclical and technology manufacturing [21][22] A-share Market and Structural Analysis - The A-share market is expected to transition into a "slow bull" market, driven by profit improvement rather than valuation expansion. The liquidity environment remains relatively loose, supporting market growth [28][31] - The basic fundamentals are expected to strengthen, with corporate profits likely to improve due to policies aimed at reducing competition and stabilizing prices [28][30] Sector Focus - The main sector themes are price increases and technology manufacturing, with a focus on performance realization rather than speculative trends. The technology manufacturing sector is benefiting from the integration of AI and overseas expansion [36][38] - The cyclical sector, particularly non-ferrous metals and basic chemicals, is expected to see significant profit recovery due to rising commodity prices [38] Growth and Value Dynamics - The growth-oriented ChiNext index is projected to outperform the value-oriented CSI 300 index, although the margin of outperformance is narrowing [47] - The proportion of companies with positive earnings forecasts is slightly improving, indicating a potential recovery in corporate profitability [22] Market Style and Trends - Small-cap stocks are expected to perform strongly in February, benefiting from a favorable environment due to limited IPOs and a focus on earnings reports [57][63] - The Hong Kong market is anticipated to continue its slow bull trend, supported by liquidity and the ongoing AI industry wave [64][65] US Market Outlook - The US market is consolidating its fundamentals, preparing for the next upward movement, with corporate earnings showing strong growth across various sectors [70] - A balanced investment strategy is recommended, maintaining core positions in technology while diversifying into cyclical sectors to capture potential excess returns [70]

Core Viewpoint - The central air conditioning industry is undergoing an adjustment period due to declining traditional engineering demand, prompting companies like Gree Electric and Midea Group to invest more resources in AI data center cooling and high-performance infrastructure to seek structural growth opportunities [5][6]. Group 1: Market Dynamics - Air conditioning still accounts for about 70% of Gree Electric's revenue, with central air conditioning being a core segment. However, the industry faces pressure from a downturn in real estate, weak traditional engineering demand, and intensified price competition [7]. - In 2025, the sales scale of China's central air conditioning industry is projected to be 138.68 billion yuan, with a 7.4% year-on-year decline in the domestic market, while the export market is expected to grow by 12.7%. The growth in exports is significantly driven by ongoing investments in overseas data centers and computing infrastructure [7]. Group 2: Technological Trends - As AI computing power increases, the power density of server cabinets is rising rapidly, revealing limitations in traditional air cooling solutions in terms of efficiency and heat dissipation. Liquid cooling is gradually becoming an important technology route for high-performance data centers [8]. - Liquid cooling is not merely an equipment upgrade but involves comprehensive changes in system architecture, core components, and operational modes. It is seen as a supplementary solution in high-performance scenarios, often coexisting with air cooling solutions, requiring higher system integration and project experience [8]. Group 3: Competitive Landscape - The data center cooling market is primarily dominated by two types of companies: specialized manufacturers with extensive experience in data center temperature control, such as Vertiv, Ingersoll Rand, and others, and foreign brands like Johnson Controls and Carrier, alongside domestic central air conditioning manufacturers like Gree and Midea [10]. - Domestic central air conditioning companies face significant challenges in the data center market, including the maturity of technology in high-performance scenarios, the ability to deliver complex projects, and the long verification cycles required by leading clients [10]. Group 4: Future Outlook - Overall, AI data center cooling presents a rare structural growth opportunity for the central air conditioning industry, but its scale, pace, and profit model differ significantly from traditional engineering markets. In the short term, data center business serves more as a buffer against the decline of traditional business rather than a decisive replacement [11]. - The ability of companies like Gree and Midea to establish stable engineering capabilities and long-term customer relationships in high-barrier technology paths like liquid cooling will be crucial in determining their position in the computing infrastructure cycle [11].

● 本报记者 刘英杰 新型电力系统建设纵深推进 随着风电、光伏等新能源装机占比持续快速提升，我国新型电力系统建设已进入纵深攻坚阶段。曾韬表 示，当前我国面对的最大系统性瓶颈在于电力系统的"承载力"与"灵活性"不足。新能源发电固有的随机 性、波动性与用电负荷的相对稳定性之间存在显著矛盾，导致电网调峰调频压力剧增，部分地区甚至出 现"弃风弃光"与"用电紧张"并存的结构性问题。 破解这一难题，离不开电网的智能化与柔性化改造。曾韬表示，未来重点关注的方向包括建设具备海量 接入能力并能智能控制分布式资源的"有源配电网"，推广应用柔性直流输电等技术，以及大力发展可聚 合分布式电源、储能和可调负荷的"虚拟电厂"。当下，以云计算、大数据、人工智能和物联网技术为代 表的数字化技术正广泛应用于电网状态监测、故障预警、负荷预测与调度优化。 在此背景下，储能的作用已发生根本性转变。"储能已从'可选项'转变为新型电力系统的'必选项'，是解 决新能源消纳、保障电网安全、提升能效的关键。"曾韬表示，国内储能的商业模式正从早期单一的配 套模式，转向由电能量市场收益、容量补偿、辅助服务收益和专项补贴等构成的多元化市场收益体系。 在政策方面，国 ...

证券日报网讯 2月2日，宝信软件在互动平台回答投资者提问时表示，公司多年来始终高度关注人工智 能相关技术发展趋势，在算力基础设施领域做了大量技术及解决方案研究。液冷技术作为其中非常重要 的组成内容，公司已有一定技术储备及实践经验。 （编辑 丛可心） ...

液冷材料专家小范围 20260127 摘要 液冷技术通过缩小液冷板尺寸和仿生设计提升热交换效率，但材料是主 要瓶颈。铜基材料虽散热优异但成本高，金属基复合材料尚处商业化过 渡期。 芯片与冷板间的界面材料（如导热硅脂、凝胶等）选择至关重要，不同 材料表现差异大，需根据需求优化。膨胀系数匹配影响传热效率，材料 需具备良好导热性和稳定性。 市场上的液冷方案主要有铝合金和铜基冷板，以及各类界面填隙材料。 铝合金成本低但导热性有限，铜基适用于高功率散热，填隙材料各有优 缺点，需综合考虑。 传统金属材料在高功率芯片冷却中，因不融化导致孔洞影响传热。液态 金属虽贴合度好，但需防泄漏，且注入过程可能产生气孔，降低传热效 率。 一体化设计将热界面材料与冷板结合，消除界面热阻，通过化学修饰连 接材料，显著提升传输效率，优于传统导热硅脂等。 远东集团的一体化设计将液态金属固定在结构内，通过压缩赶走气孔， 化学修饰结合液态金属与冷板，消除界面热阻，关注界面材料本身。 Q&A 液冷技术目前的发展情况如何？有哪些发展瓶颈和亟待解决的问题？ 液冷技术已经发展多年，特别是在结构上已经非常成熟。目前液冷材料主要是 铝合金和铜，这些材料在结构设 ...