China’s robot makers have mastered backflips: Now they want robots to think like humans

China's humanoid robots have already attracted attention with their ability to run, dance and perform backflips. But the country's next big robotics challenge is making these machines intelligent enough to perform real jobs without being programmed for every task.

The shift towards embodied AI is driving a major investment boom. Chinese companies developing robot intelligence raised 93.5 billion yuan ($13 billion) across 322 deals in the first half of 2026, five times the amount recorded a year earlier, according to research firm ITjuzi. The focus is moving beyond hardware. The real competition is now about building the AI brain that allows robots to understand their surroundings, learn new skills and adapt.

China’s Robot Race: Why AI Brains Matter

Embodied AI refers to technology that allows machines to understand and interact with the physical world. Unlike chatbots that mainly process text, these systems must learn how to grip objects, move around obstacles and handle unfamiliar situations. Chinese companies such as X Square, Galaxea and Spirit AI are developing systems designed to give robots more general-purpose skills. Some specialist startups have reached valuations of around $3 billion, according to company disclosures. Meanwhile, established manufacturers are also investing in AI research. UBTech advertised a chief scientist position with compensation of up to 124 million yuan, including equity and benefits.

How Are Chinese Companies Teaching Robots?

Training robots is different from training chatbots. Robots need examples of physical actions, not just information from internet text. Companies are collecting this data in several ways. Spirit AI has hired around 1,000 people to record everyday and industrial activities. Another company, Robbyant, says it trained its model using 160,000 hours of material, including 50,000 hours of human-guided robot movements. Some developers are also using computer simulations to teach machines how to perform industrial tasks. However, simulations cannot perfectly reproduce real-world conditions.

China vs US: Who Will Win The Robotics AI Race?

US companies such as Physical Intelligence and Skild AI are also developing AI systems that can help robots learn different tasks.

China has an advantage in manufacturing and access to industrial environments for testing. However, restrictions on advanced Nvidia chips remain a challenge for its developers. The bigger milestone will be robots that can complete unfamiliar tasks reliably. Unitree founder Wang Xingxing believes future AI models could give robotics its own "ChatGPT moment". But experts caution that successful demonstrations do not automatically mean commercial readiness. Making robots dependable across different workplaces could still take years.