Gasgoo Munich-Kunlunxing Robotics has officially launched operations in the Beijing Economic-Technological Development Area.Kunlunxing is a general-purpose humanoid robotics company anchored by core technology in physical endogenous causality. Adhering to a "body + brain" dual-drive strategy, the firm focuses on general embodied intelligence. It is dedicated to building proprietary general-purpose humanoid robots and large-scale embodied intelligence models.The founding team consists primarily of core veterans from industry leaders such as Huawei, Alibaba, and Li Auto. A defining trait is that every member brings a track record of exceptional success in their respective fields.Company founder Ren Geng is a versatile corporate leader. He combines experience as the commercial head of a 100 billion yuan revenue business at Alibaba Cloud. He also served as a CEO for overseas markets during Huawei's global expansion and as president of a large group with over 50,000 employees.Co-founder Lang Xianpeng is a "rare AI technology expert" who has cultivated the domestic AI field for 15 years. He stands as the only "AI pioneer" in the embodied intelligence sector to successfully scale autonomous driving from scratch to maturity.The founders and team possess mature business management experience. This is combined with three core capabilities: "top-tier commercial execution," "world-class AI implementation," and a "robust hardware R&D system." These factors give the company the strength and confidence to navigate the embodied intelligence revolution.Benchmarking against Tesla's humanoid robot as an industry standard, Kunlunxing is building full-stack core technical capabilities.Regarding the "brain," current approaches lean toward "fitting that looks right" rather than "reasoning that holds up physically." This applies whether using VLA behavioral imitation or spatial simulation in world models. Neither approach allows the model to truly understand the laws of physical endogenous causality.Historically, autonomous driving relied on HD maps to compensate for shortcomings. Similarly, the current robotics industry's fixation on stacking data and using imitation learning for overfitting training is a transitional product. It is not the ultimate solution."If AI capabilities are insufficient, the problem should be solved fundamentally, not by giving it a crutch." In response, Kunlunxing has established a technical roadmap for the next stage of embodied intelligence. This roadmap is centered on physical endogenous causality. It aims to build the foundational base for general embodied intelligence through three core pillars: native models, Agent architecture, and data systems.
+75 others
