Former Li Auto AD Chief Launches Embodied AI Startup in Beijing Yizhuang

According to QbitAI, the embodied-intelligence startup project led by Lang Xianpeng, former head of autonomous driving at Li Auto, has officially surfaced. Kunlunxing, co-founded by Lang and former Alibaba Group vice president Ren Geng, has settled in the Beijing E-Town Economic-Technological Development Area. The company is targeting the general-purpose embodied-intelligence track and benchmarking against the Tesla humanoid robot, with the goal of building full-stack capabilities spanning both the robot body and the AI brain. The article says Kunlunxing moved quickly after registering in March 2026: its core team was in place within two weeks, and in less than two months it completed its corporate structure, talent aggregation, and R&D system setup. It also secured support from leading investment institutions at an early stage, with its valuation soon exceeding $1 billion. The founders’ backgrounds are a major focus of the report: Ren Geng previously worked at Huawei, Alibaba, and ENN, bringing experience in hardware engineering, cloud-computing commercialization, large-organization management, and industrial transformation; Lang Xianpeng has spent many years in AI and autonomous driving, having built Li Auto’s intelligent-driving R&D system, pushed Li Auto from supplier reliance toward full-stack in-house development, and delivered highway NOA, urban NOA, map-free NOA, end-to-end plus VLM solutions, and mass production delivery of VLA. Kunlunxing’s technical position is that AI capabilities should not be supplemented with external “crutches”; instead, the company aims to start from native models, Agent architectures, and data systems to build foundational embodied-intelligence capabilities centered on physically endogenous causality. Beijing E-Town is providing support through policy, funding, talent, scenarios, and other means, because the local government hopes to address Beijing’s shortcomings in embodied-intelligence mass production and commercialization teams. However, the article also preserves a risk assessment: embodied intelligence is still in its early stages, technical routes remain unsettled, and scenario deployment, commercial closed loops, data barriers, and competitive pressure have yet to be resolved. A star team and capital enthusiasm do not guarantee entrepreneurial success.