此前，广州豪特打造的绿色节能型大数据中心项目，凭借冷热通道封闭等核心技术实现制冷效率提升 15%以上，PUE值低于1.35，机柜密度最高提升100%。 11月24日，证券时报记者从汽车热管理细分龙头开特股份(920978)获悉，公司宣布与广州豪特节能环保 科技股份有限公司(简称"广州豪特")签署增资入股协议。开特股份拟投资1000万元，正式切入数据中心 热管理赛道，开启"汽车+算力中心"双场景热管理布局。 作为国家级专精特新"小巨人"企业，开特股份构建了传感器、控制器、执行器"三器联动"的技术闭环， 拥有348项自主专利，是国内率先实现车用温度传感器国产替代的企业，同时凭借六维力传感器等技术 切入人形机器人领域。 开特股份表示，此次投资正是基于其在热管理核心部件研发、精密制造领域的积累，向高增长的数据中 心热管理赛道延伸的关键布局。 公开资料显示，广州豪特是中国第五大中立运营商数据中心能源管理解决方案提供商，公司专注数据中 心全生命周期服务，覆盖AI算力数据中心、边缘数据中心等核心场景，其液冷技术处于行业领先水 平，可提供适配高性能数据中心的先进冷却解决方案，保障设备稳定运行。 据开特股份介绍，双方将以此次千 ...

格隆汇11月24日丨紫光股份(000938.SZ)在互动平台表示，公司全面布局液冷技术，可提供冷板式、浸 没式、喷淋式等多种技术路线，致力于打造技术先进、性能卓越的全栈液冷解决方案，并推出面向AI 计算和智能科学计算领域的液冷整机柜产品系列。作为国内极少数具备液冷整机柜设计制造全流程和大 规模交付能力的厂商，公司持续完善液冷产业布局，在现有产能基础上积极筹建更大规模生产线以应对 市场需求的增长。 ...

11月18日，南方泵业召开"以泵之心，驭冷却之道"数据中心液冷系统解决方案发布会，呈现为数据 中心场景专业定制的液冷系统整体解决方案，并带来重量级的新品NF-CDU系列冷量分配单元，成为国 内第一家主动拥抱数据中心变革的国有上市泵企。 南方泵业致力于提供数据中心一站式制冷解决方案，从一次侧到二次侧，从 5kW 边缘计算到 600kW 超算中心，为每个"沸腾"的芯片提供定制化冷静方案，构建永不间断的"冷却生命线" 。 数据中心风冷系统用泵方案 南方泵业CDU冷量分配单元、液冷泵、预制管路，为液冷数据中心二次侧介质循环提供支持，助 力液冷数据中心新建与改造，为PUE限值达成提供高效解决方案。 数据中心二次侧液冷高效用泵方案 从生产到售后的全过程均体现高标准服务能力，包括通过多项三方测试验证CDU系列在性能、安 全与能效方面的合规性；采用ISO2852卡盘接口实现快速维保，支持不停机带压插拔，仅需15分钟即可 完成维护；提供覆盖所有核心部件的超长质保及主动式保养服务，可提前排查潜在故障并支持定制化运 维；依托覆盖全国的"124服务工程"体系，可提供远程技术支持并在紧急情况下协调就近资源，确保故 障处理周期不超过8小 ...

Core Viewpoint - Tianfeng Securities has given a "buy" rating for Far East Holdings, highlighting the company's sustainable growth driven by its focus on "electricity + computing power + AI" [1] Group 1: Financial Performance - In the first half of 2025, Far East Holdings achieved a revenue of 12.976 billion, representing a year-on-year growth of 14.38%, marking a record high for the same period [1] Group 2: Business Strategy - The company is actively expanding into emerging markets related to artificial intelligence, computing power, and robotics, integrating "computing power + AI" into its development strategy [1] - Far East Holdings is enhancing its equipment side by accelerating the development of liquid cooling technology, high-speed connectors, high-speed copper cables, multimode optical fibers, AI servers, and backup power supplies for data centers [1] Group 3: Market Position and Partnerships - The company has supported over 200 large data centers, including those for China Telecom and Alibaba Cloud, establishing a comprehensive support system [2] - Far East Holdings has obtained vendor codes from leading global AI chip companies, enabling bulk supply of power lines and automotive lines [2] Group 4: Technological Development - The company is focusing on the research and development of liquid cooling technology, particularly in biomimetic manifold microchannel and material innovation, to meet the increasing performance requirements of cooling solutions [2] - Far East Holdings has developed a new generation of liquid cooling technology and is conducting sample testing, leveraging its proprietary liquid cooling heat dissipation technology [2]

Core Viewpoint - The company has advanced manufacturing capabilities for plate-fin heat exchangers and cryogenic process design, which supports technology transfer to other thermal management fields, including liquid cooling [1] Group 1 - The company has completed internal technical discussions and is advancing research and development projects in the liquid cooling sector [1] - The company is in the process of team formation to support its initiatives in liquid cooling [1] - The company aims to actively expand its market layout, leveraging its established technical, manufacturing, and sales frameworks [1]

（文章来源：每日经济新闻） 每经AI快讯，有投资者在投资者互动平台提问：公司的UniPoD S80000超节点产品已采用整机柜液冷技 术。在下一代千卡及以上规模的超节点产品研发中，液冷技术将面临哪些新的挑战？公司在相变冷板等 新兴技术上的研发进展如何？ 紫光股份（000938.SZ）11月21日在投资者互动平台表示，在下一代超节点产品研发中，面临更高散热 需求，液冷技术需要进一步升级，包括微通道冷板、相变冷板、整机柜全液冷等技术。公司已在相变冷 板、微通道冷板等课题开展预研，建立新液冷的技术储备，在未来生态成熟后会逐步应用到产品上。 ...

证 券 研 究 报 告 润泽科技（300442）2025 年三季报点评 推荐（维持） IDC 规模稳步扩张，业绩环比持续增长 目标价：62.4 元 [ReportFinancialIndex] 主要财务指标 | | 2024A | 2025E | 2026E | 2027E | | --- | --- | --- | --- | --- | | 营业总收入(百万) | 4,365 | 5,886 | 8,255 | 9,717 | | 同比增速(%) | 0.3% | 34.9% | 40.2% | 17.7% | | 归母净利润(百万) | 1,790 | 5,217 | 3,171 | 4,327 | | 同比增速(%) | 1.6% | 191.4% | -39.2% | 36.4% | | 每股盈利(元) | 1.10 | 3.19 | 1.94 | 2.65 | | 市盈率(倍) | 42 | 15 | 24 | 18 | | 市净率(倍) | 7.6 | 5.4 | 5.0 | 4.2 | 资料来源：公司公告，华创证券预测 注：股价为 2025 年 11 月 20 日收盘价 公司研究 云基础设施服 ...

Core Viewpoint - The company emphasizes the importance of continuous development of new products such as liquid cooling and aims to enhance its economic efficiency and market competitiveness through technological innovation and open ecosystem collaboration [1] Financial Performance - The company advises stakeholders to refer to its periodic reports for specific financial performance details [1]

Group 1: AI and Energy Demand - The AI competition has shifted from chip supply to power supply, with significant implications for data center investments, projected to reach $3 trillion globally [1] - AI's energy consumption is rapidly increasing, with the energy demand for AI inference workloads expected to reach petawatt-hours by 2026, potentially overwhelming power grids [2][3] - The cost of electricity for data centers is substantial, with a 1 GW data center potentially incurring annual electricity costs of $700 million to $876 million [3] Group 2: Energy Strategies of Tech Giants - Tech companies are increasingly viewing energy as a survival necessity, with energy efficiency improvements translating to significant cost savings [4] - Major tech firms are adopting aggressive strategies, such as building their own power plants to ensure energy supply for their data centers [5][6] - The "Stargate" project, involving OpenAI, SoftBank, and Oracle, aims to invest $500 billion in constructing multiple giant data centers globally [6] Group 3: China's Energy Strategy - China is leveraging its clean energy advantages and policy guidance to meet AI computing power demands, with a massive $1.2 trillion power project expected to fulfill 75% of the national AI computing needs [7] Group 4: Cooling Technologies - AI data centers face a cooling crisis, with cooling energy consumption accounting for approximately 40% of total power demand [9] - Liquid cooling technology is becoming essential for AI data centers, with its ability to handle higher heat densities compared to traditional air cooling methods [11] - The global data center cooling market is projected to grow from $18.78 billion in 2025 to $42.48 billion by 2032, with a CAGR of 12.4% [11] Group 5: Lenovo's Position in Liquid Cooling - Lenovo has established itself as a key player in the liquid cooling market, with a 154% year-on-year revenue growth in its liquid cooling technology [13] - The company’s liquid cooling solutions have demonstrated significant energy efficiency, achieving a PUE value below 1.1 [13] - Lenovo's liquid cooling systems have been deployed in over 80,000 units globally, showcasing its competitive edge and market recognition [14] Group 6: Future Outlook - The global AI boom is transforming energy infrastructure into a high-growth asset class, with liquid cooling technology becoming a critical solution for energy efficiency [15] - Companies like Lenovo are positioned to benefit from the increasing demand for efficient cooling solutions in the AI sector [15]

"我们有海量闲置的英伟达GPU，但他们只能躺在仓库，因为没有足够的电力去点亮他们。" 不久前在一档播客节目中，微软CEO纳德拉向市场揭示了这一残酷的真相：AI竞赛的瓶颈，已经从芯片供应转向了电力。 在过去的3个多月，英伟达完成了从4万亿美元到5万亿美元市值的跨越。万亿美元级别的估值增长，正是建立在持续不断、指数级增长的AI基础设施需求 之上。市场预计，全球用于建设数据中心的投资未来将达到惊人的3万亿美元。 在持续数年的AI热潮中，人们热衷于讨论芯片制程的纳米之争，惊叹于大模型万亿级的参数规模，却鲜少关注支撑这一切的 "隐形基石"—— 电力。 如今，这场看似由算法与芯片主导的 AI 争霸赛，已经悄然演变为一场围绕电力资源、散热技术和能源战略的全球较量。从美国田纳西州正在建设的巨型 数据中心，到中国雅砻江流域的水电算力枢纽，电力的供给能力与利用效率，正成为决定各国和巨头们AI竞争力的核心变量。 算力狂潮下的能源危机 AI为什么越来越费电？ 过去，AI 的能源消耗主要集中在模型训练阶段 —— 一次GPT-3级别的模型训练，耗电量相当于300户家庭一年的总用电量。但随着大型语言模型 （LLMs）的普及，能源消耗已从 ...