China Telecom's AI research institute, TeleAI, has created its first commercial company—an embodied AI startup named Gamma Robotics. This company is registered in the Xuhui district of Shanghai and is based in a major large model innovation center.
Gamma Robotics' main focus will be on developing self-evolving humanoid robots and creating whole-body models. The company plans to open a humanoid robot laboratory in collaboration with TeleAI.
The founder and CEO is Bai Chenjia, who previously headed TeleAI's Institute of General Embodied Intelligence and was involved in forming the embodied AI research team. Li Xuelun, CTO and Chief Scientist at China Telecom, as well as head of TeleAI, will serve as a scientific advisor.
Technical Objectives of Gamma Robotics
Gamma Robotics pursues two key technical goals. Self-evolution implies that robots continuously accumulate experience during real-world interaction and update their capabilities, allowing the models to improve during operation. The whole-body model aims for a highly articulated system formed by the human body and dexterous hands, with the goal of simultaneously solving balancing, complex contact, and fine manipulation tasks.
The company states its intention to model locomotion and dexterous manipulation within a unified structure so that movement, balance maintenance, grasping, and object handling are integrated into complex task execution skills.
According to the announcement, the team relies on TeleAI's work, which already provides connectivity between the embodied brain for planning, the cerebellum for motion control and synthetic data generation. China Telecom's cloud and computing resource network is expected to provide resources for training, inference, and deployment.
A distinctive feature is the operator's network infrastructure. Gamma Robotics will collaborate with TeleAI on coordinating AI Flow—TeleAI's intelligent communication network—at the device, edge, and cloud levels, connecting robots to edge nodes and cloud systems. The goal is for experience gained by one robot in the physical world to become a shared capability for other nodes, allowing robots to learn from each other and perform tasks collectively.
Initial demonstration projects will focus on servicing and operating China Telecom's data centers, as well as home and commercial sector services. Both parties plan to test the limits of the robots' capabilities, system stability, and effectiveness in real-world scenarios with ecosystem partners, after which they will refine products and delivery based on the feedback received.
For China Telecom, this spin-off connects its computing network and AI research with a dedicated business for creating humanoid robots.