Core Insights - The AI industry is experiencing significant changes, with a shift from single-modal to multi-modal models and a transition from general to vertical application scenarios [5][6] - The rise of large models has initiated the integration of AI with various industries, highlighting the importance of high-quality data to address the "last mile" problem in algorithm implementation [6][7] Group 1: AI Model Development - AI large models are evolving towards multi-modal capabilities, enhancing their application in specific verticals [5] - The introduction of Chain of Thought (CoT) technology allows models to improve their accuracy and reliability by shifting from "fast thinking" to "slow thinking" [5] Group 2: Data Demand and Market Dynamics - The demand for training data in the AI sector is changing, driven by the need for high-quality data to solve practical implementation challenges [6] - The domestic AI data market in China represents only a small portion of the global market, with significant opportunities abroad [7] Group 3: Company Profile - Haitai Ruisheng, established in 2005, is one of the earliest providers of AI training data solutions in China and is currently the only publicly listed company in this sector [7] - The company has seen substantial growth in its global business, with nearly half of its revenue coming from overseas in the previous year [7]

证券研究报告 公司研究 [Table_ReportType] 公司点评报告 [Table_StockAndRank] 工业富联(601138) | 投资评级 | 买入 | | --- | --- | | 上次评级 | 买入 | [Table_Author] 莫文宇 电子行业首席分析师 执业编号：S1500522090001 邮 箱：mowenyu@cindasc.com 相关研究 [Table_OtherReport] 工业富联 2024 年报及 2025 年第一季 度报告点评：全面受益 AI 浪潮推进， 多元引擎共促成长 工业富联：AI 相关收入成长显著，公 司正处发展黄金期 工业富联三季报点评：AI 产业链核心 卡位，静待 GB200 花开 信达证券股份有限公司 CINDA SECURITIES CO.,LTD 北京市西城区宣武门西大街甲127号金隅大厦 B座 邮编：100031 [Table_Title] 工业富联 25H1 业绩预告点评：Q2 归母净利 润同比增长约 50%，AI 业务持续释放业绩 [Table_ReportDate] 2025 年 07 月 08 日 [Table_S 事件： 7um ...

Core Viewpoint - The article highlights significant positive developments in the semiconductor sector, particularly focusing on companies like 瑞芯微 (Rockchip), 工业富联 (Industrial Fulian), and 长鑫存储 (Changxin Storage), indicating strong growth potential and upcoming IPOs that could enhance market sentiment and investment opportunities in the tech industry [2][16]. Group 1: Company Performance - 瑞芯微 (Rockchip) expects a net profit increase of 185-195%, with a median net profit of 530 million, reflecting a market capitalization of 641 billion [2][4]. - 乐鑫科技 (Espressif) anticipates a net profit growth of 65.0%-78.0%, with a projected net profit of 260 million, corresponding to a market cap of 214 billion [3][4]. - 工业富联 (Industrial Fulian) forecasts a net profit increase of 36.84-39.12%, estimating around 12 billion in net profit, driven by a 60% year-on-year growth in AI server demand [8][9]. Group 2: Market Trends and Expectations - The semiconductor sector is experiencing a robust growth phase, particularly in the端侧 SoC (System on Chip) market, driven by the increasing demand for AI applications [6][16]. - The overall market sentiment is expected to improve, with a potential upward adjustment in valuations across the semiconductor supply chain due to strong performance indicators from key players [10][11][16]. Group 3: IPO Developments - 长鑫存储 (Changxin Storage) has initiated IPO counseling, with expectations to complete the IPO within a year, potentially raising around 26 billion, which could significantly impact the semiconductor equipment and materials sectors [12][14]. - The anticipated production of HBM (High Bandwidth Memory) by mid-next year is expected to create substantial market excitement and growth opportunities [14][15].

一、停复牌 2、航天工程：预中标23.92亿元项目，主要建设560万吨/年煤制甲醇、148万吨/年甲醇制烯烃及下游深 加工装置，生产聚乙烯、聚丙烯、EVA、SAP等8类产品合计277万吨/年。 1、长鸿高科：拟购买广西长科100%股权，股票停牌。广西长科是一家专注于特种合成树脂高分子材料 研发、生产和销售的高新技术企业，产品包括透明ABS等特种树脂产品。 2、蓝黛科技：公司实控权可能发生变更，股票停牌。 二、并购重组 杭叉集团：子公司拟以增资扩股方式收购国自机器人99.23%股份。 三、投资合作、经营状况 1、华电科工：签约约55.67亿元印尼项目。 9、川金诺：预计2025年上半年净利润为1.68亿元至1.88亿元，同比增长152.25%至182.28%。业绩变动 主要由于公司把握市场机遇，优化生产计划和产品结构，提升高毛利产品占比。 10、利尔化学：预计上半年净利润净利润为2.65亿元-2.75亿元，同比增长185.24%-196.00%。受益于公 司部分产品销量和价格同比上涨以及降本增效取得成效所致，公司业绩出现同比上涨。 4、乐鑫科技：预计上半年净利润为2.5亿元到2.7亿元，同比增加65%到78%； ...

★ 宏观动态 ★ 据南方+，近日，天河区在环五山创新策源区项目中，在天河区委、区政府和广州市规划和自然资源局的指导 下，市规划和自然资源局天河区分局（天河区土地开发中心）成功开出面额高达1亿元的房票。这不仅是天河 区自主实施储备项目发出的首宗房票，而且刷新了广州市单张房票面额的最高纪录，标志着天河区落实市委、 市政府"拼经济"核心工作部署取得重要进展。目前，被收储方已成功运用亿元房票，签订物业购买合同。 深交所：定于7月21日发布深证专精特新指数、创业板专精特新指数 7月7日，深交所发布深证专精特新等2条指数的公告。深圳证券交易所和深圳证券信息有限公司定于2025年7月 21日发布深证专精特新指数（简称深小巨人，代码399268）、创业板专精特新指数（简称专精特新，代码 399267）。上述2条指数均采用自由流通市值加权，于每年6月、12月的第二个星期五的下一个交易日进行样本 股定期调整，并通过深交所行情系统计算发布。 中证专精特新100指数将于7月21日正式发布 国家外汇管理局：受汇率折算和资产价格变化等因素综合作用，6月外汇储备规模上升 国家外汇管理局统计数据显示，截至2025年6月末，我国外汇储备规模为 ...

据公告，公司云计算业务在二季度实现高速增长，整体板块营业收入较去年同期增长超过50%。产品结 构持续优化，产品量产加速爬坡，客户对AI数据中心Hyper-scale AI机柜产品需求不断增加。其中，AI 服务器营业收入较去年同期增长超过60%，云服务商服务器的营业收入较去年同期增长超过1.5倍。公 司在大客户的核心产品份额继续保持领先优势，先进AI算力的GPU模块及GPU算力板等产品的出货， 在第二季实现显著增长。 通讯及移动网络设备业务方面，公司锚定智能制造与AI融合的发展机遇，依托前沿技术研发实力，持 续加码核心产品的技术研发投入，推动该业务实现稳健增长。 公告称，公司云计算业务在二季度实现高速增长，整体板块营业收入较去年同期增长超过50%。 7月7日，工业富联发布2025年半年度业绩预告，预计上半年实现归母净利润119.58亿元—121.58亿元， 同比上升36.84%到39.12%；预计第二季度实现归母净利润67.27亿元到69.27亿元，同比上升47.72%— 52.11%，均创下同期历史新高。 高速交换机方面，在AI需求持续提升的驱动下，公司产品结构持续改善，二季度800G交换机的营业收 入达20 ...

Sekai团队 投稿 量子位 | 公众号 QbitAI LeCun、李飞飞力挺的世界模型，想要实现，高质量数据是关键，也是难点。 现在，国内研究机构就从数据基石的角度出发，拿出了还原真实动态世界的新进展： 上海人工智能实验室、北京理工大学、上海创智学院、东京大学等机构 聚焦世界生成的第一步——世界探索 ，联合推出一个 持续迭代的高 质量视频数据集项目——Sekai （日语意为"世界"） ，服务于交互式视频生成、视觉导航、视频理解等任务，旨在利用图像、文本或视频构 建一个动态且真实的世界，可供用户不受限制进行交互探索。 它汇聚了 来自全球101个国家和地区、750多座城市的超过5000小时第一人称行走与无人机视角真实世界或游戏视频， 配有精细化的标签， 涵盖文本描述、地点、天气、时间、人群密度、场景类型与相机轨迹等重要信息。总的来说，具有视频质量高、视频时间长、视角多样、地域 丰富及多维度标签等特点。 团队还利用Sekai部分数据，训练了一个初步的交互式视频世界探索模型—— Yume （日语意为"梦"）。Yume在输入图片的基础上，通过 交 互式键鼠操作 （移动、视角转动）自回归形式地控制生成视频。 构建Sek ...

以下文章来源于睿兽Pro ，作者Bestla 睿兽Pro . 创业邦旗下横跨一二级市场的科创数据平台。实时投资数据、追踪产业创新。找数据、做分析、链资源，就上睿兽分析。 行业定义 AI数据服务 （ AI Data Services ） 是指围绕人工智能系统开发所需的数据，提供从采集、清洗、标注，到增强、质量控制、隐私合规与交付等全流程的 数据支持服务。 该 服务体系不仅涵盖传统的数据加工任务，更延伸至面向具体应用场景的定制化数据解决方案。 AI开发范式正从专注模型优化转向提升数据质量，通过减少数据与模型的割裂来抑制幻觉、改善输出，释放企业AI潜力 。 无论是大语言模型训练、自动驾 驶系统开发，还是金融风控、医疗图像识别等领域， AI数据服务都为模型提供了高质量、结构化且符合业务语境的数据输入，是推动AI算法从实验走向商 业应用的关键推动力。 来源： Snorkel AI 早期阶段， AI数据服务主要依赖人工采集与标注，通过众包平台完成大规模图像、文本、语音等任务的数据准备。这一阶段技术核心在于构建数据处理流 程、质量审核机制和人力管理体系。 目前 AI数据服务正在向智能化与平台化方向跃升。自动标注、弱监督学习 ...

Core Insights - Meta Platforms is seeking to raise $29 billion for financing AI data centers [1] Company Summary - The company is focusing on expanding its infrastructure to support artificial intelligence initiatives [1] Industry Summary - The move reflects a broader trend in the tech industry towards investing heavily in AI capabilities and infrastructure [1]

Core Insights - Global infrastructure investment, particularly in digital infrastructure, is gaining attention due to its long-term nature, resilience, and high returns [1] - Major overseas investment institutions and tech companies are allocating significant funds towards AI-related infrastructure to secure a competitive edge and stable long-term returns [2] Investment Trends - Large tech and financial institutions are investing heavily in AI infrastructure, with investments reaching billions of dollars [2][3] - Microsoft announced a $400 million investment in AI infrastructure and cloud computing in Switzerland [4] - Amazon plans to invest over $13 billion in expanding AI and cloud infrastructure in Australia [4] - Meta is also planning to build an AI data center, while SoftBank, OpenAI, and Oracle announced a $500 billion investment in AI infrastructure in the U.S. [5] Financial Institutions' Involvement - Financial institutions are actively participating in AI infrastructure investments, with JPMorgan providing over $7 billion in loans for an AI data center in Texas [6] - The investment cycle for AI infrastructure is typically long, ranging from 25 to 50 years, attracting interest from alternative asset investment firms [6] Market Dynamics - The focus of infrastructure investment is shifting towards three key areas: Decarbonization, Digitalization, and Deglobalization [8] - AI is driving demand for computing power, which in turn accelerates the construction of data centers and increases electricity demand [8] Regulatory and Market Developments - China's goal is to have 80% of new data centers powered by green electricity by 2025 [9] - The global capacity of large data centers is expected to double by the end of 2024, driven by AI [9]