Skip to main content

Xinghaitu

Category: Robotics / Embodied AI

Xinghaitu (Galaxea AI) is a pioneer in embodied intelligence, developing a 'unified brain' (foundation model) designed to control a diverse range of robotic forms, from mobile manipulators to humanoid platforms, through a 'One Brain, Many Forms' strategy. Xinghaitu was founded in 2023. The company is led by Jiyang Gao. Based in Beijing, China. Team size: 100-500. Total funding raised: $652.0M. Latest round: Series B+ (2 billion RMB, March 2025). Key investors include ["Ant Group","GL Ventures (Hillhouse)","IDG Capital","Meituan Longzhu","Capital Today","MiHoYo","Baidu Ventures","Cathay Capital"].

Founded
2023
Headquarters
Beijing, China
Team size
100-500
Total funding
$652.0M

Value proposition

Bridges the gap between high-level reasoning and low-level physical execution by providing a scalable AI brain that enables robots to perform complex, general-purpose tasks in unstructured real-world environments.

Products and solutions

["R1 Series: Full-sized wheeled humanoid robot (models: R1, R1 Pro, R1 Lite)","Galaxea G0: End-to-end VLA (Vision-Language-Action) foundation model utilizing a 'fast-and-slow' dual system","Galaxea Open-World Dataset: Large-scale, high-quality real-world robot action dataset","Ultra-light force-controlled robotic arms","Real2Sim2Real (RSR): Proprietary spatial intelligence engine for simulation and training"]

Unique value

Distinguished by its 'One Brain, Many Forms' philosophy, which decouples AI intelligence from specific hardware, allowing the same foundation model to adapt to various robotic bodies.

Target customer

Smart manufacturing facilities, logistics and warehouse operators, household service providers, and AI/Robotics research institutions.

Industries served

["Smart Manufacturing","Logistics & Warehousing","Consumer / Household Services","Education & Scientific Research","E-commerce"]

Technology advantage

Uses a 'fast-and-slow' dual-system VLA model where the 'fast' system handles real-time reflex actions and the 'slow' system manages complex reasoning, significantly improving response speed and energy efficiency. They leverage a massive real-world dataset to overcome the 'data scarcity' bottleneck in robotics.

How they differentiate

Xinghaitu differentiates through its 'One Brain, Many Forms' (一脑多形) philosophy, focusing on a unified foundation model (VLA - Vision-Language-Action) that decouples intelligence from specific hardware. They utilize a unique 'Fast-and-Slow' dual-system VLA model to balance real-time reflexes with complex reasoning.

Main competitors

["AGIBOT (Zhiyuan Robotics)","Galaxy General (银河通用)","Unitree Robotics","Figure AI"]

Key partnerships

["Tsinghua University (IIIS) - Deep research and talent collaboration","MiHoYo (HoYoverse) - Strategic investor and technology collaboration partner","Guyueju (古月居) - Strategic partnership for robotics education and ecosystem building","Lens Technology - Hardware and mass production partner","Investors including Ant Group, Hillhouse Capital (GL Ventures), and Baidu Ventures"]

Notable customers

["MiHoYo (HoYoverse)","Huawei Cloud","Volkswagen","Haier","Samsung","ByteDance","Physical Intelligence","Stanford University","MIT"]

Major milestones

["Founded in September 2023 by Dr. Jiyang Gao and Tsinghua professors.","Released the R1 Series robots (R1, R1 Pro, R1 Lite) for commercial shipment in Q3 2024.","Achieved a valuation of 20 billion RMB in March 2025.","Open-sourced large-scale real-world robot action datasets to the research community."]

Growth metrics

Reached unicorn status with a valuation exceeding 20 billion RMB ($2.8B USD) within 18 months of founding. Expanded from a research-focused team to over 100 employees.

Market positioning

Leading developer of embodied AI 'brains' and general-purpose robotic platforms, positioned as a top-tier Chinese unicorn in the AGI-Robotics convergence space.

Geographic focus

Primary focus on China's smart manufacturing and logistics sectors, with stated ambitions for expansion into North American and European markets for developer and enterprise tools.

Patents and IP

Holds multiple proprietary technologies in perception, planning, SLAM, and hardware-software co-design; specifically emphasizes its Real2Sim2Real (RSR) engine and DP3 (3D Diffusion Policy) algorithms.

About Jiyang Gao

Jiyang Gao is a prominent entrepreneur and researcher in the field of autonomous driving and robotics. He holds a Ph.D. in Computer Science from the University of Southern California (USC), where he studied under Professor Ram Nevatia, and a Bachelor's degree in Electronics from Tsinghua University (2015). Prior to founding Xinghaitu, he served as an Executive Director at Momenta, a leading Chinese autonomous driving startup, where he was responsible for highway autonomous driving systems and AI infrastructure. His previous research experience includes a stint at Waymo, where he contributed to significant motion prediction algorithms such as VectorNet and TNT.

Official website: