Summary of Haitai Ruisheng's Conference Call Company Overview - **Company**: Haitai Ruisheng - **Industry**: AI Data Services Key Points Strategic Initiatives - Haitai Ruisheng is strategically aligning with AI data initiatives, collaborating with China Mobile and other partners to connect with national-level data annotation bases, with expected delivery of millions of data entries by Q4 2025 in Chengdu and Changsha [2][3] - The company is also working with the Guangxi government to build an authoritative corpus for ASEAN countries, with expected revenue of 20 million yuan by 2026 [2][4] Revenue and Growth - In Q3 2025, Haitai Ruisheng reported revenue of 77.64 million yuan, a year-on-year increase of 36%, driven by G-class business expansion and overseas content review services [3][7] - The revenue growth rate slowed to 30% in Q3 due to delays in project confirmations from state-owned enterprises, but the overall annual growth is expected to remain strong at around 57% [2][7] Client Base and Demand - The traditional client base consists of major domestic and international tech companies with high demand for multilingual and specialized data services [5] - Future demand is expected to focus on three areas: growing multilingual needs, high-end specialized data requirements, and enhancing interaction naturalness [5] Emerging Business Highlights - The company has expanded its overseas content review business, operating a base in the Philippines that contributed approximately 20 million yuan in Q3 [6] - Haitai Ruisheng is actively entering the embodied intelligence sector, collaborating with robotics companies and local governments to provide high-quality training datasets [6] Collaborations and Partnerships - Haitai Ruisheng collaborates with several state-owned enterprises in AI, including China Mobile, China Unicom, and China Telecom, participating in national AI community projects [8] - The partnership with Huawei has led to successful projects, including the "Boguang Model" for the Shaanxi Cultural Tourism Group, with plans to replicate this model in other regions [9] Market Outlook - The intelligent data market is projected to have significant growth potential, with domestic tech giants expressing substantial demand for embodied data, potentially generating tens of millions in revenue [10] - The development of multimodal data technology is increasing the demand for high-end data annotation, particularly in OCR and related fields [13] Future Plans - The company aims for a compound annual growth rate of 40% to 50% over the next two to three years, focusing on expanding overseas operations and enhancing delivery capabilities [16] - Plans to further develop the Philippine base and explore the Indonesian market, with a focus on high-end customized data annotation services [14][15]

Core Insights - Humanoid robots are transitioning from laboratory settings to practical applications in daily life and production, with data scarcity being a significant bottleneck for large-scale deployment [1][2] - The Shijingshan Humanoid Robot Data Training Center, the first of its kind in Beijing, is expected to produce over one million high-quality multimodal data entries annually, providing data services to various domestic and international large model and robotics companies [1][3] Group 1: Importance of Data - High-quality, diverse, and real-world multimodal data is essential for training AI models in robots, directly influencing their ability to operate effectively in various scenarios [2] - The lack of real-world data is identified as a fundamental bottleneck in the development of intelligent robotics, emphasizing the need for large-scale applications to generate authentic data [2] Group 2: Training Center Features - The training center spans 3,000 square meters and features diverse robotic forms, including dual-arm lifting robots, wheeled humanoid robots, and bionic quadrupeds, organized into training and application scenario areas [1][3] - The center aims to address three core issues in the robotics industry: insufficient cross-scenario data generalization, significant gaps between simulated and real environments, and the lack of standardized data formats and efficient iterative ecosystems [2] Group 3: Collaborative Ecosystem - The training center collaborates with over ten companies in the embodied intelligence industry, creating a comprehensive ecosystem that covers perception, decision-making, and execution [3] - The center's data output is already being utilized by leading domestic and international large model and robotics enterprises, indicating its role in fostering innovation and collaboration within the industry [3]