（文章来源：央视网） 人民财讯11月29日电，随着AI训练与推理对电力的需求飙升，全球多个地区电网已出现吃紧迹象。英 伟达、谷歌等巨头正集体将下一代算力战场转向地球轨道。虽然许多科技巨头看好太空数据中心市场， 并纷纷开始布局，但也有分析指出其中存在一系列问题。例如虽然支持者认为将数据中心放置在太空可 以免受自然灾害破坏，但太空本身也存在风险，包括辐射、太空碎片等，太阳耀斑也会干扰通信。并且 将更多人造物体送入太空却没有清理方案将可能带来严重后果，一旦发生碰撞，许多支撑我们日常生活 的技术都将受到影响。 ...

Core Insights - The development of artificial intelligence (AI) is increasingly dependent on the scale of data centers as the performance improvement of computing chips slows down [1] - The competition between China and the US in AI is shifting towards energy consumption, particularly electricity [1] Data Center Efficiency - The Power Usage Effectiveness (PUE) of ground data centers is typically around 1.4, while space data centers can theoretically reduce PUE to nearly 1 due to simpler heat dissipation [1] - This significant reduction in power consumption is crucial for operational cost efficiency in data centers [1] Renewable Energy Integration - Space solar power stations can achieve over 8000 hours of annual generation through optimal deployment, which substantially lowers electricity costs [1] - The combined deployment of space data centers and space solar power stations, utilizing high-energy lasers for power transmission, is seen as a promising solution to address energy consumption issues in computing [1] - This approach is expected to unlock the growth ceiling for computing power, marking a new trend in the future development of artificial intelligence [1]

Core Insights - The cooling system failure at CyrusOne's data center in Aurora, Illinois, led to significant disruptions in trading activities at the Chicago Mercantile Exchange (CME), highlighting vulnerabilities in critical financial market infrastructure [1][2] - The CME had previously outsourced the operation of this data center to CyrusOne after selling it in 2016, which raises concerns about the risks associated with outsourcing essential operations [3][4] - The incident has drawn attention to the role of private equity firms in the data center sector, particularly following the acquisition of CyrusOne for approximately $11.4 billion by KKR & Co. and Global Infrastructure Partners [5][6] Group 1: Data Center Operations - The Aurora data center, established by CME in 2009, has become a crucial hub for electronic trading, processing a nominal trading volume of at least $25 trillion daily [1][2] - High-frequency traders have invested heavily in proximity to this data center to minimize trading delays, indicating its strategic importance in the financial ecosystem [2] Group 2: Outsourcing and Risks - CME's decision to sell the data center and lease it back for 15 years reflects a broader trend of outsourcing critical infrastructure, which can introduce operational risks [3][4] - CyrusOne's business model focuses on securing large clients like CME, emphasizing its reputation as a significant player in the data center industry [4] Group 3: Design and Operational Concerns - The cooling system failure has raised questions about the design and redundancy measures in place at the data center, as it was expected to have additional cooling units to prevent such failures [6][7] - Despite having a disaster recovery plan that included a backup data center in New York, CME opted to restart operations in Aurora based on the belief that the cooling issue would be resolved quickly [7] Group 4: Market Dynamics - The surge in demand for data centers, driven by trends like artificial intelligence, has attracted private equity interest, as seen in the acquisition of CyrusOne [5][6] - The incident underscores the critical nature of data centers in global trading and the potential implications of operational failures on market stability [1][2]

Core Insights - The recent trading disruption at CME Group highlights the critical issue of cooling systems in data centers, which can significantly impact global markets [1] Group 1: Incident Overview - On November 27, CME Group's trading platform experienced a multi-hour outage affecting trillions of dollars in contracts across various asset classes due to a cooling system failure at its Aurora, Illinois data center [1] - The cooling system failure was attributed to a malfunction in a chiller unit, which impacted multiple cooling units, leading to market turmoil [1] - CyrusOne, the operator of the data center, stated that the incident was caused by a "simple" physical failure, emphasizing the vulnerability of data centers to such issues [1] Group 2: Data Center Cooling Challenges - Data centers consume significantly more energy than typical office buildings, with energy consumption per square foot being 50 times higher, leading to substantial waste heat generation [5] - Traditional cooling methods using air are being increasingly replaced by liquid cooling systems due to the higher heat output from AI-related workloads, although these systems are more complex and costly [6] - The reliance on cooling systems raises concerns about water resource consumption, particularly in water-scarce regions, as data centers require substantial water for cooling [6] Group 3: Consequences of Overheating - Overheating in data centers can result in data loss, damage to expensive server components, and service interruptions, similar to recent outages experienced by other digital infrastructure providers [8] - Despite investments in redundancy measures, such as backup generators and additional cooling units, the complexity of systems can still lead to unavoidable disruptions [8] Group 4: CME Incident Analysis - The CME trading platform is located in a CyrusOne-operated facility in Aurora, which reportedly has advanced cooling technology and additional cooling units to mitigate failures [9] - Following the incident, CyrusOne deployed temporary cooling equipment while working to restore full cooling capacity [9] - The effectiveness of the redundant cooling systems during this incident remains unclear, raising questions about their reliability [9]

近日，北京市科委、中关村管委会等单位召开"智绘星空胜算在天——太空数据中心建设工作推进会"， 宣布北京拟在700公里至800公里晨昏轨道建设运营超过千兆瓦(GW)功率的集中式大型数据中心系统， 以实现将大规模AI算力搬上太空。 建设拟分为三个阶段：2025年至2027年，突破太空数据中心能源与散热等关键技术，迭代研制试验星， 建设一期算力星座，实现"天数天算"应用目标；2028年至2030年，突破太空数据中心在轨组装建造等关 键技术，降低建设与运营成本，建设二期算力星座，实现"地数天算"应用目标；2031年至2035年，卫星 大规模批量生产并组网发射，在轨对接建成大规模太空数据中心，支持未来"天基主算"。 业内普遍认为，"算力上天"是大势所趋。从行业发展来看，算力正在加速渗透各行业各领域，规模随之 持续提升。据《2025年中国人工智能计算力发展评估报告》，2025年中国智能算力规模将达 1037.3EFLOPS，同比增长43%；通用算力规模预计达85.8EFLOPS，同比增长20%。总体来看，2023年 至2028年中国智能算力规模和通用算力规模的复合增长率预计达46.2%和18.8%。 算力需求的激增直接推 ...

Core Viewpoint - Data Port (603881.SH) announced that Ningbo Ruixin plans to reduce its holdings by up to 21,551,308 shares, representing 3.00% of the company's total share capital [1] Summary by Categories Share Reduction Plan - Ningbo Ruixin intends to reduce its stake through centralized bidding and block trading methods, with a maximum of 7,183,769 shares (1.00% of total share capital) to be sold via centralized bidding and 14,367,539 shares (2.00% of total share capital) through block trading [1] - The reduction period will commence fifteen trading days after the announcement of the reduction plan and will last for three months [1] Trading Restrictions - The total number of shares reduced through centralized bidding within any consecutive 90-day period shall not exceed 1.00% of the company's total shares [1] - Similarly, the total number of shares reduced through block trading within any consecutive 90-day period shall not exceed 2.00% of the company's total shares [1] Adjustments for Corporate Actions - If there are any corporate actions such as dividends, stock splits, capital reserve transfers, or rights issues during the reduction period, the number of shares to be reduced will be adjusted accordingly [1] - The price for the shares to be reduced will be determined based on market prices [1]

Core Viewpoint - The company has announced its participation in the Guotai Haitong Industrial Fund to expand its investment in various sectors including optical communication, data centers, automotive electronics, and robotics [1] Investment Details - The company will invest 35.4 million RMB as a limited partner in the Guotai Haitong Zhongji Xuchuang Technology Equity Investment Fund (Kunshan) [1] - The investment is made in collaboration with Guotai Junan Innovation Investment Co., Ltd., Jiangsu Qianrong Capital Management Co., Ltd., and other limited partners [1] Fund Registration - The fund has completed the registration and filing procedures as per relevant laws and regulations, with the fund name officially registered as Guotai Haitong Zhongji Xuchuang Technology Equity Investment Fund (Kunshan) [1] - The fund's registration number is SBKE21 [1]

分析师指出，随着人工智能算力需求大幅增长，各国数据中心均有向太空延伸的趋势：谷歌的"捕日者 项目"旨在将机器训练部署至轨道，业界称近地轨道近乎无限的太阳能是优势；分析人士预计，到2030 年，基于太空的计算有望实现盈亏相抵，中国也在推进轨道上的"空间计算中心"；马斯克亦提出将轨道 人工智能卫星与SpaceX和特斯拉关联。 ...

科华数据夯实自身优势，通过五大能力的系统整合，积极融合多年来在电源行业、数据中心行业的场景 服务经验，与用户一同接受时代的检验，形成了从算力基础设施产品到垂直行业应用场景的"全生命周 期完整服务链条"，助力各行业的智能化转型升级。 论坛期间，科华数据全球首款200kW高密UPS模块产品正式亮相，并获泰尔认证中心颁发的全球首份 1.2MW UPS（200kW模块）认证证书。这一产品不仅体现了科华数据在数据中心供电架构技术上的领 先突破，还构建了面向AI算力场景的高可靠供电解决方案，为AI算力集群的持续演进奠定了坚实基 础。 随后，科华数据中心产品总监吴庆彬在《与AI共进|算力基础设施革新与供电创新技术突破》主题分享 中系统阐述了AI时代下算力基础设施面临的五大核心挑战，分享了公司在供电技术方面的创新突破， 深入解读了此次亮相的全新产品。他表示，当前国内外AI芯片发展存在差异化，国外高端芯片的技术 演进及国产AI芯片的加速崛起，正塑造着全球智算中心基础设施的技术路径，供电架构正处于"双轨 制"共同发展阶段。 基于此，此次亮相产品以三大特点引领供电系统进入"兆瓦级"时代：首先，该系列产品直面传统供电密 度低、占地大 ...

Group 1 - The company Keda Data (SZ 002335) announced the convening of its first temporary board meeting of the tenth session on November 28, 2025, in Xiamen, Fujian Province, to discuss the election of members for various specialized committees [1] - For the first half of 2025, Keda Data's revenue composition was as follows: 49.62% from the renewable energy sector, 37.44% from the data center sector, 11.77% from the smart electric power sector, and 1.17% from other businesses [1] - As of the report date, Keda Data's market capitalization was 27.4 billion yuan [1]