过去两个月，AI这轮投资的"增量"开始从芯片、服务器，外溢到更难补的环节：电力。超大规模云厂商 上修资本开支与研发预算，训练与推理两端的算力部署更激进，直接推高了数据中心用电的远期斜率。 与此同时，市场的担心也在变：不是"需不需要电"，而是"供给链能不能按时把电送到机房"。 高盛全球投资研究部分析师 Brian Singer 在23日的报告中写道："我们将2030年全球数据中心用电需求 相对2023年的增幅从175%上调到220%。"这次上调的重点落在美国：新增用电中约六成来自美国，数 据中心容量预测也被明显抬高。 更麻烦的是，需求上修并不等于路径清晰。电网侧的并网、输配电和设备交付周期都在拉长，于是"表 后电"（behind-the-meter）这种更像过渡方案的选择被推到前台——以天然气为主，先让机房跑起来， 后续再接回电网。高盛也把美国电力需求增速预期抬到2030年前年化3.2%，其中数据中心贡献了2个百 分点。 在投资上，高盛的态度并不保守：尽管数据中心电力供应链相关股票已明显跑赢，报告仍然维持看多， 背后抓的是一个更大的叙事——为了避免电力、水、网络、供应链的"可靠性事故"，基础设施进入一轮 更长的 ...

Core Insights - The investment in AI is shifting from chips and servers to electricity, with major cloud providers increasing capital expenditures and R&D budgets, leading to a significant rise in data center electricity demand [1][2] - Goldman Sachs has revised its forecast for global data center electricity demand by 220% by 2030 compared to 2023, with approximately 60% of this demand expected to come from the U.S. [3][5] - The report highlights a growing concern about the supply chain's ability to deliver electricity on time, as the integration and delivery cycles for power infrastructure are lengthening [1][6] Electricity Demand Forecast - Goldman Sachs estimates that the global data center electricity demand will increase by 905 TWh by 2030, up from a previous estimate of 175% growth to 220% [3] - The U.S. is projected to account for about 60% of this increase, with data center capacity expected to rise to 95 GW by 2030 [3][5] - The annualized growth rate for U.S. electricity demand has been raised to 3.2%, with data centers contributing 2 percentage points to this growth [5][6] Investment Trends - Goldman Sachs maintains a bullish outlook on data center electricity supply chain stocks, indicating a longer investment cycle in infrastructure to avoid reliability issues [2][12] - The report notes a significant increase in capital expenditures and R&D budgets for hyperscalers, with an expected doubling by 2029 compared to 2025 [4] - The reinvestment rate for hyperscalers is projected to reach 87% in 2026 and 83% in 2027, indicating a squeeze on free cash flow for shareholders [4] Efficiency and Power Consumption - While new server generations are more efficient, the overall demand for computing power is growing faster, leading to higher energy consumption per server [7] - The report identifies 2026 as a critical year for observing whether inference servers will maintain lower power consumption or increase due to higher power server ratios [7] Pricing and Policy Implications - The concept of a "Green Reliability Premium" has emerged, with clean energy supply costs for data centers exceeding baseline prices by approximately $40 to $48 per MWh [8][9] - Goldman Sachs anticipates that the costs associated with reliability and infrastructure will not be passed on to other electricity customers, leading to more contract designs to isolate these impacts [9] Labor Market Constraints - The report emphasizes that labor shortages, particularly in transmission and distribution roles, pose a significant constraint on meeting electricity demand growth [10] - An estimated 510,000 new jobs will be needed in the U.S. and Europe to meet electricity demand from 2023 to 2030, with a focus on training and recruitment [10] Broader Market Dynamics - The report suggests that the reliability theme in infrastructure investment is expected to continue, with annual capital expenditure growth exceeding $80 billion [12] - The performance of data center electricity supply chain stocks has outpaced broader market indices, indicating a shift in investor focus towards reliability and infrastructure [12]

智通财经APP获悉，鸿腾精密1 月份营收增长 35.5%，这表明在全球人工智能开发的浪潮下，对英伟达 (NVDA.US)服务器的需求依然保持强劲。 鸿海负责制造数据中心存放芯片的服务器，是围绕英伟达的人工智能硬件行业的关键参与者之一。尽管 由于该技术变现路径仍存在不确定性，市场出现了产能过剩的警告，但从 Meta 到亚马逊(AMZN,US)在 内的美国公司仍在投入数十亿美元部署用于训练和运行人工智能所需的设备，鸿海也因此从中受益。 鸿腾精密报告上月营收为 7300 亿新台币(约合 230 亿美元)。然而，去年同期的对比基数可能受到了农 历新年假期的时间影响，2025 年的农历新年落在了 1 月份。据预测，该公司在截至 3 月份的三个月内 销售额将增长 28%。 ...

来源：格隆汇APP 格隆汇2月5日｜鸿海1月份营收增长35.5%，表明在全球人工智能发展浪潮中，市场对英伟达服务器的 需求依然强劲。这家也被称为富士康的公司上月报告营收为7300亿新台币（约合230亿美元）。不过， 与去年同期的对比可能受到了农历新年假期的影响，2025年的春节假期落在1月份。预计该公司在截至3 月的三个月内销售额将增长28%。鸿海负责制造数据中心存放芯片的服务器，是围绕英伟达的人工智能 硬件行业的关键参与者之一。尽管由于技术变现仍存在不确定性，市场出现了产能过剩的警告，但从 Meta到亚马逊在内的美国公司仍在投入数十亿美元部署训练和运行人工智能所需的设备，鸿海因此从 中受益。 ...

目前的"延误"是一个涵盖极广的概念，其成因复杂多样。从公用事业公司无法兑现电力承诺、关键设备 到货滞后，到现场劳动力短缺导致的施工进度延误，各类情况层出不穷。 对于旨在建设大规模AI基础设施的建筑商而言，交付速度与规模的复杂性正在显现：发电机、变压器 和液冷组件的等待时间不断延长，且由于英伟达最先进的硬件必须采用液冷，这一尚处实验阶段的技术 仍有许多难题待解。 这种行业性的普遍滞后正在引发资本市场的连锁反应。从近期案例来看，并非所有的延误影响都相同， 真正令市场担忧的是涉及政治阻力、融资失败以及具有约束力的合同告吹。这些因素不仅直接冲击了相 关公司的股价，也迫使部分资本方重新审视项目的经济可行性。 与此同时，行业巨头正试图从源头解决这一瓶颈。英伟达首席执行官黄仁勋上周在加州总部秘密召集了 一次峰会，旨在推动"表后"发电模式，即数据中心自行建设发电设施以绕过电网互连的漫长等待。 对于行业而言，能否快速解决电力供应和工程延误问题，将决定谁能在这场AI竞赛中生存。 政治阻力与融资难题 部分项目的延误并非源于技术问题，而是遭遇了地方政府的政治阻力，尤其是当数据中心规模从几百兆 瓦跃升至吉瓦（GW）级别时。 随着人工智 ...

曾因押注美国次贷危机而名声大噪的《大空头》原型人物、投资人迈克尔·伯里(Michael Burry)，近日再度掀起波澜，将火力对准了美股最 炙手可热的人工智能(AI)赛道。 他指责美国几家最大的科技公司通过激进的会计操作，在AI热潮中"虚增"盈利。 与此同时，华尔街大型投 行高管及知名做空者频频发出警告，认为美股可能出现回调。 伯里当地时间周一在社交平台X上发文称，那些所谓的"超大规模算力服务商"，正在通过延长芯片折旧年限、压低折旧费用，人为放大AI热 潮带来的账面利润。 周一，英伟达股价在上周下跌7%后反弹近6%；Palantir则在上周大跌11%后回升约9%。不过，英伟达和Palantir周二股价再度走低。 伯里最后还表示，他将在11月25日公布更多细节，并让关注者拭目以待。 与此同时，日本软银集团作出了一项令人意外的举动：清仓了"AI芯片霸主"英伟达股份，套现58亿美元，加剧了外界对AI热潮或已触顶的担 忧。 他写道： "通过延长资产的使用年限来低估折旧支出，从而虚增利润，是现代财报中最常见的把戏之一。在以英伟达芯片和服务器为代表、 产品周期仅两三年的情况下，大规模增加资本开支并不应成为企业延长计算设备 ...

Core Viewpoint - The article discusses the current trends and forecasts in the memory market, particularly focusing on NAND and DRAM, highlighting the demand dynamics and inventory levels of major players like Samsung and SK Hynix [1][2][3]. Summary by Sections Market Overview - The article references UBS research, indicating that Samsung and SK Hynix expect a modest increase in DRAM shipments for Q2 2025, with Samsung projecting less than 10% growth and SK Hynix over 20% for NAND [2]. - UBS has adjusted its NAND price forecast for Q2 from +5% to +3%, citing resistance from customers against rising SSD prices [2]. Demand and Inventory - As of Q1 2025, DRAM inventory levels are decreasing faster than expected, with smartphone customers holding about 10 weeks of inventory and PC manufacturers around 12 weeks [3]. - For NAND, smartphone manufacturers have approximately 9 weeks of inventory, while SSD inventory stands at 11 weeks [3]. HBM Demand Forecast - UBS has revised its HBM demand forecast for 2025 down from 203 billion Gb to 189 billion Gb, reflecting a year-on-year growth of 105%, and for 2026 from 303 billion Gb to 291 billion Gb, with a growth rate of 54% [3][24]. - The demand for ASICs is expected to outpace GPUs, with ASICs accounting for 54% of total HBM demand by 2026, up from 41% in 2025 [3]. Supplier Insights - Samsung's HBM bit shipments are projected to increase from 5.1 billion Gb in 2024 to 11.2 billion Gb by 2026, representing 8.1% of its total DRAM bits [9]. - SK Hynix is expected to grow its HBM shipments from 6.8 billion Gb in 2024 to 17.4 billion Gb by 2026, which will account for 17% of its total DRAM bits [9]. Production Capacity - Longxin's wafer production capacity is expected to reach 170,000 wafers per month by the end of 2024, with plans to increase to 230,000 wafers per month by the end of 2025 [24]. - Yangtze Memory Technologies Company (YMTC) is also expanding its production capacity, aiming for 160-layer technology despite facing challenges due to U.S. restrictions [25]. Competitive Landscape - The article highlights the competitive dynamics among major players, with Samsung and SK Hynix leading in HBM production, while Micron is also increasing its HBM output significantly [9][24]. - The market share of DRAM suppliers is detailed, showing Samsung at 39.5%, SK Hynix at 28.9%, and Micron at 22.9% for 2025 [27]. Future Outlook - The article concludes with a cautious outlook for the memory market, emphasizing the uncertainty due to tariff issues and potential specification downgrades in NAND demand [2][24].