China’s embodied AI push overseas is expanding from robot sales to open-source model ecosystems
Chinese embodied AI companies are taking a broader route overseas, moving beyond selling robots into exporting models, toolchains, and open-source technical infrastructure. The article cites data from EO Intelligence showing that 73 of 182 funded embodied AI companies have already expanded abroad, while another 21 are preparing to do so, bringing the combined share to about 51.6%. Unitree founder Wang Xingxing previously said overseas business accounted for about 50% of the company’s annual performance, while firms including AgiBot and UBTECH have put globalization at the center of this year’s strategy plans and launch events. The report argues that China’s edge rests on three layers: hardware bodies and motion control, real-world industrial and service scenarios that generate data, and open-source “brain” models that plug into global developer ecosystems. Ant Lingbo’s LingBot series is presented as a case study, with more than 33,000 GitHub stars and over 2.2 million downloads. The article also details how outside teams have reused or adapted LingBot models, including Stanford University’s Vision and Learning Lab, researchers from the National University of Singapore, Nvidia, AWS Labs, and independent developers building tools on top of the models. It concludes that two outbound paths are taking shape: a capital-heavy route focused on products and supply chains, and a platform route centered on open models and developer ecosystems.
