Over the past year, the heat in the Embodied AI sector has not dissipated but has instead become more "realistic".
While earlier discussions focused on whether robots could be built, the industry is now asking: Who can actually deliver? Who can enter real-world scenarios? Who can turn technology into a sustainable business?
From the perspective of global valuation, top companies are forming different routes. Figure AI and Physical Intelligence are betting heavily on model and system capabilities, seeking to answer how robots should understand the world and complete tasks. Meanwhile, companies such as Apptronik, 1X, and Agility Robotics place greater emphasis on the robotic body, mass production, and landing in specific scenarios.
The characteristics of Chinese companies are even more distinct. Enterprises like UBTECH (优必选), Unitree (宇树), Agibot (智元), Galbot (银河通用), and Galaxea (星海图) are no longer staying in the product demonstration stage but have started entering real scenarios like factories, logistics, and retail. Compared to overseas companies that emphasize technical paradigms, the advantages of Chinese firms are more reflected in engineering capabilities, supply chain efficiency, cost control, and delivery speed.
This is also the most noteworthy aspect of Embodied AI: it is not a pure software track, nor a simple continuation of the traditional robot industry. Models determine the upper limit of capability, the body determines the carrier form, and the scenario determines whether this system can truly be established.
Therefore, valuation rankings themselves are not the most important factor. What matters more is which companies are closer to a commercial closed loop. Future competition will not only look at whose robot movements are flashier or whose model parameters are larger, but at who can truly integrate the model, body, data, and scenario. The first stage of Embodied AI was building the robot. The second stage is bringing robots into real scenarios. The next stage, the real watershed, will be: who can achieve large-scale delivery, who can earn sustainable profit, and who can replicate a scenario to more markets. From this perspective, the competition in Embodied AI has just begun. (Note: Data as of April 29, 2026. The list covers core humanoid robot and Embodied AI companies; valuations of unlisted companies are estimated based on the latest financing and public information, and rankings are for reference only.)
