海天瑞声(688787.SH)发布2025年度业绩快报，公司2025年度实现营业收入37,697.20万元，较上年同期 增长59.00%;归属于母公司所有者的净利润为1,407.45万元，较上年同期增加24.16%。 报告期内，全球人工智能产业从"技术突破"迈向"规模化落地"，公司训练数据业务迎来强劲增长，其中 视觉业务收入增长超过200%，文本业务收入增长超过100%，语音业务收入亦保持合理水平。驱动这一 高增态势的核心动能来自四重引擎：一是国内外科技巨头加速全球纵深布局，带动各类模态的多语种数 据爆发增长;二是视觉大模型持续突破，推动高精度、场景化视觉数据投入持续加码;三是交互体验迈向 拟人化，多情感、多角色、多音色的语音数据需求持续释放;四是垂直行业渗透加速，文旅、医疗、教 育等领域垂类大模型落地提速，专业化、场景化数据需求日益凸显。 ...

智通财经APP讯，海天瑞声(688787.SH)发布2025年度业绩快报，公司2025年度实现营业收入37,697.20 万元，较上年同期增长59.00%;归属于母公司所有者的净利润为1,407.45万元，较上年同期增加24.16%。 报告期内，全球人工智能产业从"技术突破"迈向"规模化落地"，公司训练数据业务迎来强劲增长，其中 视觉业务收入增长超过200%，文本业务收入增长超过100%，语音业务收入亦保持合理水平。驱动这一 高增态势的核心动能来自四重引擎：一是国内外科技巨头加速全球纵深布局，带动各类模态的多语种数 据爆发增长;二是视觉大模型持续突破，推动高精度、场景化视觉数据投入持续加码;三是交互体验迈向 拟人化，多情感、多角色、多音色的语音数据需求持续释放;四是垂直行业渗透加速，文旅、医疗、教 育等领域垂类大模型落地提速，专业化、场景化数据需求日益凸显。 ...

报告期内，全球人工智能产业从"技术突破"迈向"规模化落地"，公司训练数据业务迎来强劲增长，其中 视觉业务收入增长超过200%，文本业务收入增长超过100%，语音业务收入亦保持合理水平。驱动这一 高增态势的核心动能来自四重引擎：一是国内外科技巨头加速全球纵深布局，带动各类模态的多语种数 据爆发增长；二是视觉大模型持续突破，推动高精度、场景化视觉数据投入持续加码；三是交互体验迈 向拟人化，多情感、多角色、多音色的语音数据需求持续释放；四是垂直行业渗透加速，文旅、医疗、 教育等领域垂类大模型落地提速，专业化、场景化数据需求日益凸显。 格隆汇2月27日丨海天瑞声(688787.SH)公布2025年年度业绩快报，公司2025年度实现营业收入37,697.20 万元，较上年同期增长59.00%；归属于母公司所有者的净利润为1,407.45万元，较上年同期增加 24.16%；归属于母公司所有者的扣除非经常性损益的净利润为1,116.65万元，较上年同期增加 124.52%。 ...

海天瑞声发布业绩快报，2025年度实现营业总收入3.77亿元，同比增长59.00%；净利润为1407.45万 元，同比增长24.16%。报告期内，全球人工智能产业从"技术突破"迈向"规模化落地"，公司训练数据业 务迎来强劲增长，其中视觉业务收入增长超过200%，文本业务收入增长超过100%，语音业务收入亦保 持合理水平。 ...

（文章来源：证券日报） 证券日报网2月26日讯 ，海天瑞声在接受调研者提问时表示，过往多年，公司持续为境外多家头部科技 大厂的全球人工智能产品的本地化及出海提供关键的多语种、多模态数据支撑。随着全球化AI应用场 景的快速落地，市场对高质量、多语种、场景化训练数据的需求持续提升。具体而言，驱动需求的产品 线主要包括但不限于：多语种语音识别数据：服务于智能助手、客服机器人等产品的全球化部署与口音 适配。多语种手写体数据：支持金融票据识别、表单处理、手写笔记数字化等应用在不同语言文字区域 的准确理解。多语种文本数据：涵盖自然语言理解、内容审核、机器翻译等任务所需的多语言文本语料 库。公司凭借在多语言、多模态数据处理领域长期积累的全球供应链管理能力和技术know-how，正在 持续获取并交付此类项目，从而推动境外数据业务的快速发展。 ...

（文章来源：证券日报） 证券日报网2月26日讯 ，海天瑞声在接受调研者提问时表示，公司非常看好具身智能数据领域作为高增 长的新兴赛道，已组建具身智能数据专项团队开展相关业务，并开始在全国多个城市启动专项公司的选 址。一方面，国家已将具身智能列为未来产业予以重点培育。另一方面，控制机器人认知决策等"大 脑"仍处于发展早期，大模型在泛化能力上仍有不足，因此想实现规模化落地将依赖大量高质量、细颗 粒度的训练数据，因此这一领域市场潜力巨大。当前，国内、外多家头部科技企业已启动具身智能数据 的规模采购，需求涵盖机器人在真实场景下的交互数据、构建仿真世界模型所需数据等。同时，部分地 方政府也在积极规划建设"具身智能训练场"，旨在工程化生产高质量具身数据，赋能科研与产业应用。 目前，公司已与多家机器人本体厂商展开合作，并与若干家头部科技大厂及地方政府启动订单需求对 接，正在推进相关样例数据交付与训练场方案的设计与落地，为下一步具身智能数据业务的加速发展奠 定坚实基础。 ...

机器之心发布 2025 年，人工智能的发展重心正在发生一次根本性转移：从追求模型的规模，转向构建其理解与解决复杂现实问题的能力。在这一转型中，高质量数据正成为定 义 AI 能力的新基石。作为人工智能数据服务的前沿探索者，数据堂深度参与并支撑着这场变革的每一个关键环节。本文将深入解读 2025 年 AI 五大技术趋势及其 背后的数据需求变革。 「人情味」与「实时性」革命 趋势解码：追求更细腻的情感与更自然的实时互动 当前，语音合成技术已超越追求「清晰准确」的基础阶段，正同时向两个深度智能化维度演进：一是为合成语音注入情感、个性与文化适配性，让虚拟助手、数 字人、有声内容更具感染力和亲和力；二是从单向反应升级为支持实时打断、重叠对话与上下文连贯的全双工自然交互，这已成为高端智能座舱、实时翻译、拟 真客服等前沿场景的刚需。技术的核心挑战在于，让 AI 不仅能「读」出文字，更能「理解」语境与情绪，并像真人一样实时聆听、思考与回应，实现有情感、有 逻辑的连续对话。 数据需求跃迁：从「清晰样本」到「生动语料」与「交互流」 训练数据的重心正经历双重跃迁。一方面，需构建服务于音色、韵律、情感和风格精细控制的「表现力语料库」， ...

Core Insights - The company experienced a significant decline in net profit in Q3, with a nearly 90% year-on-year drop, attributed to a decrease in revenue from standardized data set products and a slowdown in income growth from large government projects [1][2] Group 1: Financial Performance - For the first three quarters of the year, the company achieved a revenue of 234 million yuan, representing a nearly 57% year-on-year increase, while net profit attributable to shareholders was approximately 4.18 million yuan, up about 10% [1] - The decline in Q3 profitability was primarily due to a lower proportion of high-margin standardized data set product revenue and a slowdown in income growth from significant government projects [1][2] Group 2: Business Development - The company has established a clear cooperation model with local governments, focusing on building high-quality data sets based on local characteristics and ensuring data security [2] - Collaborations have been initiated with cities like Chengdu, Changsha, and Baoding, and specific projects are underway with local governments in Guangxi and Hohhot [2] Group 3: International Expansion - The company benefits from the international strategies of overseas tech giants, leading to strong demand for high-quality, multilingual OCR recognition and text understanding data [3] - A localized data delivery system established in Southeast Asia has entered stable operation, contributing to continuous growth in overseas business revenue [3] Group 4: Emerging Opportunities - The company is optimistic about the growth potential in the embodied intelligence data sector and has formed a specialized team to develop related business [3] - The company is exploring the establishment of additional bases in Southeast Asia to further enhance overseas custom business and high-end data labeling market [3] Group 5: AI Applications - The company provides various data products and services to ByteDance, a significant client, which may be utilized in their AI applications [4] - The rise of AI agents in smart terminals is expected to create new data demands, particularly for diverse data types and multilingual voice data [4]

Core Insights - Beijing's innovative landscape is highlighted by the transformation of companies like Sanyuan Gene, Haitan Ruisheng, and Haibo Sunchuang into platform-based innovators, showcasing a shift from single-product firms to comprehensive solution providers in their respective fields [7][9][10]. Group 1: Company Transformations - Sanyuan Gene has evolved from a single-product pharmaceutical company to a platform-based innovative pharmaceutical enterprise, achieving significant growth in R&D investment and patent numbers [8][9]. - Haitan Ruisheng has transitioned from a data company to a foundational infrastructure provider for AI model implementation, with its asset scale increasing from 500 million to nearly 900 million yuan and maintaining a revenue growth rate of over 24% [7][10]. - Haibo Sunchuang has established itself as a leader in energy storage integration, leveraging its origins in Haidian District to become a key player in the tech innovation sector [4][10]. Group 2: Market and Economic Context - During the 14th Five-Year Plan period, Beijing ranked third globally in the number of unicorn companies, with R&D expenditure intensity around 6%, placing it among the top innovative cities worldwide [5][11]. - By the end of 2024, Beijing is expected to have 115 unicorn companies and 1,035 national-level specialized and innovative "little giant" enterprises, leading the nation in both metrics [11]. - The city has seen over 300 new tech companies established daily, surpassing Silicon Valley in density, and has achieved direct financing exceeding 5.6 trillion yuan during the 14th Five-Year Plan [11][19]. Group 3: Policy and Support - Beijing's government has implemented a range of supportive policies for tech companies, including funding for R&D, tax incentives, and a comprehensive service network to facilitate innovation and growth [12][18]. - The "15th Five-Year Plan" emphasizes the development of strategic emerging industries, including integrated circuits, robotics, and AI, aiming to enhance the competitive edge of these sectors [22][24]. - The introduction of the "merger and acquisition 19 articles" aims to accelerate the listing process for tech companies and improve the quality of existing listed companies, fostering a more integrated innovation ecosystem [40][42][43]. Group 4: Future Outlook - Sanyuan Gene aims to achieve annual revenue exceeding 1.5 billion yuan by the end of the 15th Five-Year Plan, positioning itself as a pioneer in gene engineering [22]. - Haitan Ruisheng plans to solidify its role as a key player in the digital transformation of multiple industries by the end of the 15th Five-Year Plan [20][21]. - The focus on building industry clusters and enhancing collaboration among companies is expected to lower innovation costs and accelerate technology application [28][30].

Core Viewpoint - The company confirmed that Google and Alibaba are important clients and that it provides AI training data for them [1] Group 1 - The company has been providing AI training data to both Google and Alibaba [1] - The company will adhere to relevant laws and regulations regarding information disclosure concerning Alibaba's revenue contribution [1]