Unitree and UBTech move from hardware to robotic brains and AI

In a major move for the robotics industry, two of China's leading humanoid robot companies, Unitree and UBTech, are changing strategy from focusing on hardware to focusing on developing core AI models and robotic operating systems, known as "robot brains."

From limbs to mind: a strategic turn

Previously, both companies were known for advanced and often affordable robotic devices. Now they are making it clear that the focus is shifting. Unitree, widely known for its agile H1 and Go2 quadropeds and other robots, is actively promoting the “Unitree Robot Morphology” model. In parallel, UBTech, the creator of the Walker series, is promoting “Walker S AI” and its own robotic operating system.

Unitree's efforts to create a "robot brain"

Unitree's recent announcements have focused on its large "Unitree Robot Morphology" model. The company says it is designed for real-time control and reasoning required by humanoid robots. This AI is said to enable complex tasks such as reasoning, logical analysis and autonomous navigation. As a key demonstration, a humanoid robot, Unitree G1, costing approximately $16 thousand, was shown, which performs tasks based on high-level commands: for example, recognizes, brings and transfers a specific drink to a person.

UBTech Strategy: Unified AI and Own OS

Similarly, UBTech is promoting a large "Walker S AI" model and its own robotic operating system. The company recently showed how Walker S performs a logistics task in a real production environment: the robot finds the right workstation, accurately assembles components, conducts quality checks and delivers the finished product - all autonomously, controlled by AI. UBTech's approach combines its AI model with an internally developed OS.

Industry Implications: "Intel Inside" for Robotics?

The parallel courses of the two major players indicate a maturing industry, where standardized, powerful hardware is becoming more affordable. The main field of competition is shifting to the software layer: in the AI ​​models that provide complex reasoning, and in the operating systems that control the robot's key functions.

Competition is intensifying

Now this focus puts them directly in line with other giants who are also actively investing in robotic AI. Google's RT-2, OpenAI (though it recently disbanded its robotics team), and Tesla's Optimus are all pushing similar ideas: powerful AI models that can generalize across different tasks.

The main question remains the same: which approach will be more effective? Are we talking about specialized models adapted to specific “shapes” of robots, like Unitree, or about more universal models that can adapt to different configurations? The race to create the ultimate “robot brain” has already begun, with Chinese companies lining up at the forefront.