ByteTree, a physical AI data infrastructure provider, has secured 50 million yuan in a Pre-A funding round from investors including Yuanhe Houwang, Ince Capital, and Xinghaitu. The company plans to use the proceeds to build physical AI data pipelines, develop embodied intelligence algorithms, and construct high-quality datasets, alongside the R&D and application of robots in specialized scenarios. Founded in 2021, ByteTree has deep roots in autonomous driving data services. Having served nearly 100 automakers and intelligent driving suppliers, it has built a comprehensive edge in computer vision processing, scenario comprehension, and the industrial ecosystem. Building on that foundation, ByteTree leveraged its mature data-loop capabilities to pivot into the embodied intelligence sector in the second half of 2025. In just one year since, it has served nearly 20 embodied intelligence manufacturers, including several top-tier players valued at over 20 billion yuan. Image Source: ByteTree Focusing on physical AI applications, ByteTree has constructed a closed-loop business chain spanning "data collection — Shanhai data management platform — real-world scenario verification — high-quality datasets." For data collection, ByteTree utilizes methods like real-machine teleoperation and body-less data capture to tap into the real world. Backed by two training grounds and over 300 industrial, commercial, and agricultural process channels, its monthly capacity for embodied intelligence data now exceeds 50,000 hours. These two facilities are equipped with more than 200 humanoid robots and hundreds of body-less data collection devices, delivering an annual effective capacity of 350,000 hours. ByteTree's proprietary Shanhai platform handles critical stages—annotation, quality control, management, evaluation, and simulation—using AI automation tools to achieve large-scale data production. In scenario development, the platform supports secondary development of foundation models and validates data through real-world tasks, ultimately producing high-quality datasets that are verified, reusable, and iteratively optimized. ByteTree is refining embodied robot models for niche scenarios. Partners include leading embodied intelligence firms and listed companies, with use cases covering lighting production, office services, industrial manufacturing, and logistics. To date, ByteTree has delivered over 200,000 hours of effective embodied intelligence data. By 2027, it expects annual deliveries to surpass 1 million hours. ByteTree is also expanding globally. Having launched overseas operations in 2024, it established a Singapore entity covering over 40 countries and regions, including the EU, CIS, Southeast Asia, Latin America, the Middle East, and Australia. It operates more than 10 global modification and calibration bases and has logged over 1 million kilometers in compliant autonomous driving data collection. An overseas data delivery center is set to open in 2027, targeting international markets with a focus on the United States. About "Seeds Discovery":