| 89,700 KRW | | | | | | | | | | + Follow | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | +3,700 (4.30%) ↑ today | | | | | | | | | | | | Oct 2, 9:19AM GMT+9 · Disclaimer | | | | | | | | | | | | 1D | 5D | 1M | 6M | YTD | 1Y | 5Y | Max | | | | | 90,000 | | | | | | | | | | | | 89,000 | | | | | | | | | | | | 88,000 | | | | | | | | | | | | 87,000 | | | | | | | | | | Previous | | | | | | | | | | | | close | | 86,000- | | | | | | | | | | 86,000 | | 10:00AM | | | | | | 1:00 PM | | | 3:00 PM ...

Core Viewpoint - Super X AI Technology has launched the SuperX Modular AI Factory, a data center-level solution aimed at addressing the key pain points of traditional AI data center construction, including long build cycles, high costs, significant energy consumption, and scalability challenges [1][2]. Group 1: Product Features - The SuperX Modular AI Factory offers a full-stack solution that integrates computing power, cooling, and power systems, reducing the on-site delivery and deployment time to under 6 months, compared to the traditional 18-24 months [2][5]. - The solution features ultra-high density computing, with a single module capable of supporting up to 20MW and housing 6 SuperX NeuroBlock core computing units, which can support up to 144 NVIDIA GB300 NVL72 GPUs [3][5]. - The modular architecture allows for flexible expansion, enabling customers to scale from a single module to multiple units as business needs grow [3][6]. - The integration of high-voltage direct current (HVDC) technology enhances power efficiency, achieving over 98.5% end-to-end power efficiency and reducing overall energy consumption by 23% compared to traditional cooling systems [5][6]. Group 2: Strategic Implications - The launch of the SuperX Modular AI Factory signifies a strategic upgrade for SuperX, transitioning AI infrastructure from an engineering project to a standardized product, thereby enhancing deployment speed and energy efficiency [7]. - This innovation positions SuperX as a potential standard setter in the next generation of AI factories, aiming to drive the transformation of traditional data centers into AI-centric facilities [7].

【导读】券商突然上攻，新能源产业链全线爆发 9月29日早盘，新能源产业链全线爆发，在券商板块的掩护下，上证指数翻红。截至午盘，上证指数涨0.13%，报3832.9点；深证成指涨1.11%；创业板指 数涨1.77%，重回3200点上方。沪深京三市半日成交额1.29万亿元，全市场超3000只个股上涨。 盘面上热点轮番活跃，新能源板块集体爆发，其中绿色甲醇概念股快速拉升，嘉泽新能（601619）3连板，东华科技（002140）2连板；储能概念股反 弹，通润装备（002150）4天2板。固态电池概念股表现强势，天际股份（002759）2连板，万润新能、多氟多（002407）等多股涨停。此外，证券板块盘 中走强，国盛金控（002670）涨停创新高。 | 代码 | 名称 | 5分钟涨跌幅 | 涨跌幅 √ | | --- | --- | --- | --- | | 886011 | | -0.04% | 3.02% | | 886054 | | -0.20% | 2.92% | | 886006 精细化工指数 | | -0.08% | 2.80% | | 886068 工程机械指数 | | -0.07% | 2.37% | ...

其次，技术层面以差异化战略构建多元生态。 2025年，全球科技巨头在AI基础设施领域的投资竞赛进入白热化阶段。 在刚刚结束的2025云栖大会上，阿里巴巴宣布正积极推进3800亿元的AI基础设施建设计划，并明确表 示将持续追加更多投入。这是全球科技巨头加码AI基建的又一重要动向。前不久，英伟达宣布将向 OpenAI投资高达1000亿美元，作为交易的一部分，OpenAI将利用英伟达系统建设并部署至少10吉瓦的 AI数据中心，用于训练和运行下一代模型。此外，微软、谷歌等企业也持续投入扩建算力网络。 可以看到，科技巨头的投资方向高度一致，全部指向AI基础设施建设。这场AI基建竞赛，已超越单纯 的技术迭代，演变为产业生态与全球价值链重构的角力场，更将重塑全球科技产业格局。 首先，竞争焦点从模型创新转向算力竞赛。 曾几何时，科技巨头在模型创新的赛道上你追我赶，OpenAI的GPT系列、百度的文心一言、阿里的通 义千问等大模型相继涌现，这些产品以强大的语言理解与生成能力，为用户带来了前所未有的交互体 验，在全球范围内掀起了大模型竞赛热潮。但如今，竞争的焦点正从模型创新转向算力竞赛，背后有着 深刻的技术与市场逻辑。 从技术发 ...

Group 1 - Alibaba is actively advancing a three-year plan to invest 380 billion in AI infrastructure, with intentions for further investment [1] - By 2032, the energy consumption of Alibaba Cloud's global data centers is projected to increase tenfold compared to 2022 [1] - AI-related product revenue has achieved triple-digit year-on-year growth for eight consecutive quarters, with AI revenue accounting for over 20% of Alibaba Cloud's external commercialization revenue [3] Group 2 - Wu Yongming predicts that large models will serve as the next generation operating system, linking tools and user demands across various industries [3] - AI Cloud is expected to replace energy as the most important commodity, driving daily operations across multiple sectors [3] - The majority of AI capabilities will be generated and delivered in the form of tokens within cloud computing networks, with tokens being likened to future electricity [3]

这是阿里云投资最直接受益的环节，主要为AI计算提供硬件和环境支撑。 数据中心的高效运行离不开以下关键技术和部件。 三、行业应用与生态伙伴 这些公司将阿里云的云计算和AI能力与具体行业场景结合，推动技术落地。 在2025云栖大会上，阿里宣布正推进3800亿元AI基础设施建设并计划追加投入，目标2032年阿里云全球数 据中心能耗规模较2022年提升10倍，其产业链上的核心合作伙伴备受关注。 数据港：被认为是阿里云最核心的IDC（互联网数据中心）供应商，深度绑定阿里云业务。 浪潮信息：作为国内AI服务器的领军企业，是阿里云服务器的核心供应商，尤其在AI服务器领域占 比很高。 杭钢股份：与阿里巴巴集团有重磅合作，共同投资158亿元建设浙江云计算数据中心项目。 润建股份：与阿里云合作共建"中国-东盟智算云"项目，负责提供算力服务器。 二、关键部件与技术供应商 光模块：光迅科技是阿里云数据中心光模块的核心供应商。中际旭创作为全球光模块龙头，也被视为 阿里云800G等高速光模块的主力供应商。 网络设备：锐捷网络是阿里云数据中心交换机的核心供应商，市场份额较高。 液冷温控：随着算力密度提升，散热至关重要。英维克是数据中心精密 ...

行业周报 | 电力设备 抖音发布 200MWh IDC 配储招标，阿里启动 3800 亿 AI 基建计划 抖音集团发布 200MWh 大规模 IDC 配储招标，国内大型数据中心配储有望持续放量。 字节跳动发布抖音集团数据中心风光储微网合同能源管理寻源公告，拟在中国大陆寻找 具备园区风光储微网投资、建设、运维全流程能力的合作方。采购范围：200MWh 以上 大规模锂电储能，投标人需具备大陆范围内 30MWh 以上大规模锂电储能系统投资、建 设、运维项目经验。 储能行业推荐：宁德时代、亿纬锂能、阳光电源、德业股份、华宝 新能，建议关注海博思创、中创新航。 中汽新能全固态电池计划 2026 年装车一汽示范车型，刚果（金）再次延长钴出口禁令至 10 月 15 日。近日，中汽新能规划到 2026 年固态电池实现 400Wh/kg，2030 年前后将 推进富锂锰、金属锂及无负极技术。电动车板块推荐欣旺达、豪鹏科技、尚太科技，建 议关注万润新能、天际股份。PCB 板块建议关注德福科技、铜冠铜箔、普利特。固态电 池板块推荐三祥新材，建议关注当升科技、先导智能、厦钨新能、北京利尔、纳科诺尔、 英联股份、德龙激光、冠盛股份；锂电 ...

Core Viewpoint - The IPO process for domestic GPU chip companies is accelerating, with Moore Threads and Muxi Co., Ltd. making significant progress despite both companies currently not being profitable. The expected revenue growth in 2024 and the increasing market recognition of their AI computing products are key factors supporting their market entry [1][2][3]. Company Performance - Moore Threads is projected to achieve a significant revenue increase in 2024, with expected revenue of 438 million yuan, while net losses are expected to decrease to 1.492 billion yuan. By the first half of 2025, revenue is anticipated to reach 702 million yuan, a 238.71% year-on-year increase, with net losses improving to 271 million yuan, a 56.02% decrease year-on-year [4][5]. - Muxi Co., Ltd. is expected to see a staggering revenue growth of 1354.9% in 2024, reaching 742 million yuan, although net losses are projected to widen to 1.409 billion yuan [9]. Product Development and Market Strategy - Moore Threads has been iterating its GPU architecture annually, with the latest "Pinghu" chip expected to enhance performance significantly. The company’s revenue structure is shifting towards AI computing products, which are projected to account for nearly 80% of revenue by the first half of 2025 [6][7][8]. - Muxi Co., Ltd. is focusing on expanding its customer base among major internet companies and telecom operators, with significant progress in its AI computing business. The company is actively engaging with key clients and expects to finalize orders in 2025 [15][16]. Industry Trends - The domestic AI chip industry is witnessing a trend towards building an ecosystem, with companies like Moore Threads and Muxi Co., Ltd. contributing significantly to the development of AI computing capabilities. The focus on internet and telecom markets is crucial for the growth of domestic AI chip manufacturers [3][14]. - The establishment of collaborative initiatives, such as the OISA ecosystem, indicates a concerted effort to enhance the interoperability and performance of AI chips, which is essential for the advancement of AI infrastructure in China [19][21].

9月24日的2025云栖大会上，阿里巴巴集团CEO吴泳铭表示，目前阿里正积极推进3800亿的AI基础设施建设，并计划追加更大的投入。对比2022年这个 GenAI的元年，到2032年阿里云全球数据中心的能耗规模将提升10倍。 这无疑是一个大超预期的消息，意味着在甲骨文等海外云厂商宣布加码AI资本开支后，国内云业务巨头也开始迅速跟进。 静态视角：阿里巴巴能拿出多少钱？ 分析阿里巴巴静态可动用的资金潜力，可分为三个层次（不考虑外部融资安排，仅作量级估算）： 第一个层次是短期可动用资金，流动性较强。使用账面现金、短期投资、短期股权证券及其他投资、应收款项等短期资产的总和，衡量公司短期动能调动 的最大资金量。 第二个层次是中长期潜在资金，需变现金融类资产。在第一层的基础上，再考虑公司长期投资（如理财、债券、股权投资等）的变现，衡量公司在长周期 可动用的最大潜在资金量。 第三个层次是偿债后的稳健资金。在第二层的基础上，再考虑到财务稳健性，假定在偿还各类刚性债务后，公司仍可动用的最大资金量。刚性债务主要包 括：短期负债（应付款项、短期借款、短期债券、应付税费）、长期负债（长期借款、长期债券、长期票据）。 那么，阿里巴巴当 ...

"我们也将在完善AIDC产业生态、打造循环经济示范区、攻关行业大规模存储难题等方面展开深度合 作。这次合作也进一步巩固了公司在国内算力产业发展中的地位。"上述工作人员说道。 此外，紫光股份有限公司（以下简称"紫光股份"）也在不断加码智算中心建设。 当前，智算中心建设呈现出快速发展态势。"众多国家和地区都在积极推动相关项目的落地实施，以满 足日益增长的AI研究和应用需求。"盘古智库（北京）信息咨询有限公司高级研究员余丰慧对《证券日 报》记者表示。 据甬兴证券研报，在多种因素共同驱动下，预计2028年中国智算中心市场投资规模有望达2886亿元。 面对持续攀升的市场需求，智算中心加速建设，相关上市公司积极布局。 北京秦淮数据有限公司（以下简称"秦淮数据"）工作人员对《证券日报》记者表示，秦淮数据在今年8 月份与宁夏回族自治区中卫市人民政府在北京签署战略合作框架协议。根据协议，秦淮数据计划投资 240亿元，在中卫市建设3座零碳超大规模智算园区，项目建成后的园区总容量可达1.2GW。 9月24日至26日，2025云栖大会在杭州云栖小镇举行。期间，阿里巴巴集团董事兼首席执行官吴泳铭表 示，公司正积极推进3800亿元的AI ...