格隆汇2月8日｜在2月6日刚结束的一场深度访谈中，马斯克称："五年后，我预计我们每年向太空发射 并运营的AI算力，将超过地球上所有AI算力的总和。" 这一计划需要每年向轨道发射约100吉瓦(GW)的 太阳能与算力载荷，相当于1万次星舰发射任务。这也意味着SpaceX将不仅是火箭公司，更将成为AI时 代的"超级服务商"。 ...

近日，SpaceX向美国联邦通信委员会提交申请，计划在近地轨道部署100万颗卫星，并构建全球首个在 轨太空数据中心。 SpaceX此番狮子大开口，本质上是对近地轨道资源的"抢先卡位"，暗藏着美国掠夺近地轨道资源、推行 太空霸权的野心。 缔造"太空商业帝国"。 SpaceX提交的申请文件显示，此次计划的核心使命是组建具备"空前计算能力"的轨道数据中心系统。 今年1月，马斯克在达沃斯论坛上公开表示，当前AI芯片的生产正呈指数级增长，但电力供应增长缓 慢，严重阻碍了数据训练和大模型部署的效率。 据统计，2024年全球数据中心耗电量占全球总用电量的1.5%，冷却系统更是消耗了数亿吨水资源，经 济成本与环境限制成了AI算力增长的"天花板"。 SpaceX的在轨数据中心计划，试图将AI算力从地面的物理限制中解放出来，打造一个不受资源约束的 太空算力底座。 太空不仅拥有取之不尽的太阳能发电资源，更有着接近绝对零度的绝佳散热环境，能大幅降低能耗与维 护成本。 在马斯克看来，摆脱地面部署的限制后，AI计算有望在几年内大幅降低成本，从而推动AI模型"以前所 未有的速度和规模"迭代升级。 在通信层面，SpaceX将凭借星链已有网络 ...

Core Viewpoint - The surge in AI infrastructure capital expenditure in the tech industry is deemed reasonable, appropriate, and sustainable, driven by extremely high demand for computing power, as stated by Jensen Huang [1][5]. Group 1: AI Infrastructure Investment - Huang indicated that the capital expenditure from key clients like Meta, Amazon, Google, and Microsoft is projected to reach approximately $650 billion by 2026, representing a 60% increase from 2025 [3][7]. - This spending level significantly exceeds the GDP of many medium-sized economies, with most funds allocated for purchasing NVIDIA chips [3][7]. - The anticipated capital expenditures for these companies will be close to or exceed their total spending over the past three years, setting records for individual company annual capital expenditures [7]. Group 2: Market Reactions and Concerns - Recent financial reports and guidance have led to severe sell-offs in tech stocks, with a cumulative market value loss of about $1.35 trillion among major tech firms [3][10]. - Despite some companies like Meta and Alphabet seeing stock price increases, others like Amazon and Microsoft faced significant declines, resulting in a total market value drop exceeding $950 billion since the latest earnings reports [10]. - Concerns about investment efficiency and potential overcapacity have created a negative sentiment in the market, with analysts highlighting the structural issues surrounding the massive capital expenditures required for AI development [11][12]. Group 3: Profitability and Future Outlook - Huang emphasized that as long as companies continue to pay for AI, they will generate profits, leading to exponential growth in revenue [6][5]. - AI companies are already becoming profitable, with NVIDIA's clients leveraging AI to enhance their operations, such as Meta transitioning its recommendation systems to generative AI [6][5]. - The ongoing demand for AI computing power is reflected in the rental of all previously sold GPUs, indicating a robust and sustained need for AI infrastructure [6][5].

Investment Rating - The report maintains a "Positive" investment rating for the electronic industry, indicating an expectation of returns stronger than the market benchmark by over 5% [5]. Core Insights - Strong demand for computing power driven by AI applications is expected to continue, with significant investments from major cloud providers [8]. - The hardware supply-demand imbalance is spreading across various sectors, leading to price increases [8]. - New technologies such as CPO (Co-Packaged Optics) are anticipated to create additional demand [8]. Summary by Sections Investment Recommendations and Targets - Key investment targets include: - Semiconductor manufacturing: SMIC (688981, Buy), Hua Hong Semiconductor (01347, Buy) - Testing and packaging: Changdian Technology (600584, Buy), Tongfu Microelectronics (002156, Buy), and others [9]. - Server storage: Lianqi Technology (688008, Buy) - CPUs: Haiguang Information (688041, Buy), Longxin Technology (688047, Not Rated), and others [9]. - Passive components: Sanhua Group (300408, Buy), Fenghua Advanced Technology (000636, Not Rated) [9]. - Server manufacturing: Industrial Fulian (601138, Buy), Huaqin Technology (603296, Buy) [9]. - Analog and power chips: Naxin Micro (688052, Buy), Sierui Technology (688536, Not Rated), and others [9]. - Semiconductor equipment: Zhongwei Company (688012, Buy), Northern Huachuang (002371, Buy), and others [9]. - Optical devices/chips: Zhishang Technology (301486, Not Rated), Tianfu Communication (300394, Not Rated), and others [9]. AI Applications and Edge Computing - Key targets in edge AI applications include: - AI main control chips: Amlogic (688099, Buy), Hengxuan Technology (688608, Buy) - Edge storage: Zhaoyi Innovation (603986, Buy), Bawei Storage (688525, Buy) [10]. - Terminal manufacturers: Hikvision (002415, Buy), Luxshare Precision (002475, Buy), BYD Electronics (00285, Not Rated), and others [10]. - Core components for AI edge: Huanxu Electronics (601231, Buy), Sunny Optical Technology (02382, Buy), and others [10].

而葛卫东另一位关联人王萍持股则保持稳定，截至1月26日，王萍仍持有1148.92万股麦格米特，与去年 底持股数量相同。仅因定增股本稀释，持股比例被动降至1.98%，变为麦格米特第五大股东。 从投资路径来看，葛卫东家族近年来尤为青睐AI产业标的。目前葛卫东不仅是兆易创新的第三大股 东，还是沐曦股份的最大外部股东，与混沌投资合计持有其2693.79万股。 值得注意的是，在葛卫东申购失利的同时，其关联人葛贵莲也退出了麦格米特前十大股东名单。 据了解，葛贵莲是在2025年三季度末新进成为麦格米特的第九大股东，持有麦格米特606.76万股，持股 比例为1.1%，直到2025年底持股数量未发生变化。而此次定增完成后（截止1月26日），麦格米特第十 大股东最新持股数量为529.98万股。以此参照，2026开年以来葛贵莲至少减持麦格米特76万股。 2月5日，麦格米特发布了股票发行上市公告以及最新股东名单。麦格米特此次定增以85.01元/股的价格 向10名对象发行约3132.59万股，合计募集资金26.63亿元。 申购名单显示，葛卫东也参与了麦格米特此次定增，申购总金额高达24亿元，分别出价83.7元、79.98 元、75.6 ...

Market Overview - The U.S. stock market experienced a significant rebound, with the Dow Jones Industrial Average rising by 2.47%, surpassing the 50,000 mark for the first time, while the S&P 500 and Nasdaq Composite increased by 1.97% and 2.18%, respectively, driven primarily by large technology stocks [1] - Semiconductor stocks led the rally, with notable gains from ARM and AMD exceeding 11%, and Nvidia and Broadcom rising nearly 8% [1] - Smaller tech stocks in niche sectors such as quantum computing, drones, and nuclear power also showed remarkable performance, indicating a shift in market sentiment towards high-growth technology areas [1][3] Geopolitical Factors - A breakthrough in U.S.-Iran nuclear negotiations has significantly improved global risk appetite, contributing to the stock market rebound [2] - The negotiations, which took place in Oman, resulted in a "good start" according to Iranian officials, and have altered the pricing logic for commodities and risk assets, leading to a decline in oil prices and a return of funds to technology stocks [2] Technology Sector Insights - Despite the overall rebound, there are concerns regarding the divergence within the technology sector, particularly with major players like Amazon and Google facing stock price declines due to high capital expenditure plans [4][5] - Amazon's announcement of a $200 billion capital expenditure plan for 2026, significantly above market expectations, led to an over 11% drop in its stock price [4] - The total AI-related spending by major tech companies, including Amazon, Google, Microsoft, and Meta, has reached $660 billion, raising concerns about the sustainability of such investments relative to revenue growth [5] Emerging Sectors Solar Energy - Tesla's ambitious solar energy production goal of 100 gigawatts within three years is set to reshape the global solar industry, with plans for factory expansions in multiple U.S. states [6] - The integration of AI computing with renewable energy solutions is expected to drive growth in both sectors, presenting opportunities for domestic solar companies to increase market share [6] AI Computing - The Chinese government has initiated a new policy to enhance the efficiency of computing infrastructure, marking a transition to a 2.0 era in computing capabilities [7] - The policy aims to create a unified system for computing resource allocation, which is expected to benefit sectors such as optical communication and computing hardware [8] Chemical and Oil & Gas - The global chemical industry is experiencing price increases, with BASF raising TDI prices by 11% in the Asia-Pacific region, leading to a ripple effect in domestic chemical sectors [9] - The geopolitical situation in the Middle East and severe weather conditions in Europe have contributed to rising oil and gas prices, benefiting domestic oil and gas sectors [10]

来源：中信建投证券 一、往期观点回顾 在上期策略中我们提到，下周A股料将震荡走稳，结构分化将进一步加剧。春节效应下，尽管节前资金心态 偏谨慎，但充裕的流动性仍为市场提供支撑，成长风格有望逐步回暖，不过板块轮动仍会较快。当前市场核 心逻辑围绕政策托底、业绩分化与流动性充裕展开，资金正向低估值蓝筹、业绩预增股及政策主线聚拢。操 作上，可保留核心仓位把握结构机会，同时警惕业绩暴雷、解禁减持等带来的回调风险，控制仓位逢低布 局。配置可聚焦三大主线：资源周期板块、科技成长赛道及消费基建领域，各主线均有明确的利好支撑与细 分布局方向。 二、一周市场回顾及走势分析 （一）市场整体表现 本周指数出现防御类风格企稳叠加成长类调整，且前期热点方向回落较明显，创业板指调整幅度较大。（见 图1） 图1： 资料来源：中信建投通达信 （二）板块表现 从行业板块表现来看，本周食品饮料、家庭用品和银行涨幅靠前。（见图2） 图2： 资料来源：wind资讯 从概念板块涨幅来看，本周市场白酒、饮料制造、银行等表现靠前。（见图3） 图3： 融资融券余额方面，最近两融余额小幅回落迹象。（见图4） 图4： 资料来源：wind资讯 （三）宏观经济数据 1 ...

Core Insights - Tesla's Vice President Tao Lin outlined the company's strategic planning and business layout for 2026, with capital expenditures exceeding $20 billion [1] Group 1: Capital Expenditure Allocation - The capital expenditure will focus on six main areas, including the advancement of Cybercab mass production, with core production line construction in the U.S. nearly completed [1] - Significant investment will be directed towards the AI computing center in Texas, which has already received over $10 billion, with plans for substantial additional investment in 2026 [1] - Upgrades to the robotics factory are underway, with a goal to achieve mass production capabilities for the Optimus robot by the end of 2026 [1] Group 2: Business Expansion and Upgrades - Tesla will continue to expand its energy storage business, increasing manufacturing investments to meet the rapidly growing global energy demand [1] - The global manufacturing system will be upgraded to enhance hardware automation and software capabilities, making the entire manufacturing system more efficient and scalable [2] - The charging network will be continuously expanded and opened to more automotive companies, increasing coverage and accessibility [2]

Group 1: Transformer Industry - The transformer production capacity is fully utilized, with orders increasing significantly since June 2022, driven by a global "transformer shortage" [1][2] - China is the world's largest transformer producer, accounting for approximately 60% of global capacity, which supports domestic companies in securing international orders [2] - Major companies like Igor are expanding production capacity in multiple countries, with forecasts indicating a sustained industry recovery for 3-5 years [2] Group 2: Large Bore Engine Industry - Large bore engines are essential for ensuring continuous operation of AI data centers, with significant growth opportunities in the power supply sector [3][4] - Weichai Power has broken the monopoly of international giants in the large bore engine market, with a projected 259% increase in sales of data center power generation equipment by 2025 [4] - The company is expanding its product offerings to meet the stringent power requirements of data centers, including customized solutions [3][4] Group 3: Core Components Industry - The demand for core components is surging, particularly due to the explosion of self-supply power needs in North American data centers [6][7] - Tianrun Industrial is enhancing production capabilities with advanced automated lines, aiming to significantly increase output in the coming years [7] - The global market for data center generator sets is expected to double from $6 billion in 2023 to $12 billion by 2030, indicating strong growth potential [7]

Core Insights - The demand for AI computing power is driving rapid growth in the liquid cooling sector, with companies actively pursuing mergers and acquisitions to capture market opportunities [1][2] - Traditional air cooling solutions are becoming inadequate for high-density computing environments, leading to a shift towards liquid cooling technologies due to their energy efficiency and superior heat dissipation capabilities [1] Group 1: Mergers and Acquisitions - Jinfu Technology announced plans to acquire 51% stakes in two companies specializing in liquid cooling components for a total consideration of up to 714 million yuan, aiming to enter the high-growth liquid cooling market [1] - Lingyi Technology completed the acquisition of Dongguan Liminda Electronics, enhancing its capabilities in AI server liquid cooling and establishing a foundation for international market expansion [2] - Cheng Tian Wei Ye plans to raise up to 800 million yuan for projects related to liquid cooling system industrialization and R&D, with significant investments allocated for advanced production equipment [2][3] Group 2: Commercialization and Technological Advancements - UCloud has launched an immersion liquid cooling cabinet at the Ulanqab Intelligent Computing Center, achieving a 15°C reduction in GPU temperature, thus ensuring stable hardware operation [2] - Xin'an Co. has commercialized its immersion liquid cooling project using self-developed silicon-based cooling liquid, achieving an energy usage efficiency (PUE) of below 1.1, which is over 35% more energy-efficient than traditional air cooling [3] - Kangsheng Co. is set to complete its intelligent computing center by September 2025, marking the large-scale commercial application of its immersion liquid cooling technology [3] Group 3: Industry Trends and Strategic Focus - The immersion liquid cooling solution is seen as a breakthrough for high-density computing clusters, offering reduced maintenance costs and improved equipment longevity [4] - Companies are increasingly positioning liquid cooling as a strategic growth area, with significant investments in R&D and market development [4][5] - Ningbo Jingda is focusing on liquid cooling as a new growth engine, with plans to enhance R&D efforts and collaborate with clients on technological challenges [5]