In May 2026, a significant shift in China’s high-tech landscape became evident as three of the nation’s most prominent first-generation robotics firms moved toward public listings. Unitree, Leju, and DEEP Robotics, all founded around 2016, have reached critical IPO milestones nearly simultaneously. This convergence marks the end of a decade-long survival game for companies that focused on mechanical endurance and motion control long before 'Embodied AI' became a multi-billion dollar buzzword.
These firms were born in a different era, emerging just as the AlphaGo victory signaled a new horizon for machine learning. While today’s startups often lead with sophisticated Large Language Models, the 2016 cohort survived by navigating a fragmented domestic supply chain and high component costs. They prioritized 'the body'—the servos, actuators, and gait algorithms—over the 'brain,' carving out niches in education, power grid inspection, and consumer-grade quadruped robots to maintain cash flow during the industry’s leaner years.
The strategic divergence between these veterans and newer 'brain-first' startups, such as Agibot (Zhiyuan), is now coming to a head. While newcomers have secured astronomical valuations based on AI potential, the pioneers are presenting the public market with tangible track records. Unitree, for instance, has shipped over 30,000 units, demonstrating the kind of mass-production and supply-chain mastery that theoretical AI models have yet to replicate in the physical world.
As these companies approach the trading floor, the central question for investors is shifting from whether a robot can look like a human to whether it can function as a sustainable industry. Secondary markets provide a ruthless reality check that venture capital often avoids. The success or failure of these IPOs will set a 'valuation anchor' for the entire sector, determining whether the current 10-billion-yuan valuations of younger startups are grounded in reality or speculative inflation.
However, the battle for dominance is far from over as the two generations begin to converge. First-generation companies are aggressively investing their IPO proceeds into AI 'brains' to improve autonomy, while the newer AI-centric firms are building their own factories to solve hardware reliability issues. The upcoming year will likely see a winnowing of the field where only those who master the synthesis of physical reliability and cognitive intelligence will survive the transition from lab to factory floor.