Sota Wujie
Category: Robotics / Embodied AI
Chinese embodied-AI startup building a "physical brain" (native 4D world-action models + multimodal VLA) for robots, targeting overseas commercial scenarios starting with European supermarkets. Sota Wujie was founded in 2026. The company is led by 胡德波 (Hu Debo). Based in Shanghai, China. Team size: 11-50. Latest round: Undisclosed.
- Founded
- 2026
- Headquarters
- Shanghai, China
- Team size
- 11-50
Value proposition
Builds the "native physical brain" for robots — a unified latent-space world-action model that lets robots understand and act on the physical world (contact, force, spatial/physical change), rather than just "seeing" it. Does not build full robot hardware; instead provides the brain + OS + data-collection stack and integrates with China's hardware supply chain for overseas deployment.
Products and solutions
Physica-Claw robot operating system (agentic AI architecture), native 4D physical world model / world-action model (unified latent space), multimodal VLA, self-developed head-mounted Ego vision capture + full-modal UMI handheld data-collection devices, three-finger end effector, chassis+dual-arm robot integration for early deployment.
Unique value
First company to deploy a native 4D physical world model in real commercial settings (European supermarkets), combining world-action models, multimodal VLA, and a Physica-Claw agentic robot OS with proprietary Ego + full-modal UMI data-collection devices. Founder's unique cross-border capability (ex-Huawei Europe, prior partnerships with 7 of the world's top-10 supermarket retail groups) is a hard-to-replicate moat for overseas compliance and deployment.
Target customer
Overseas (Europe/US) commercial enterprises — supermarkets, logistics and manufacturing groups with labor shortages and clear ROI for robot labor replacement.
Industries served
Retail (supermarket back-of-house sorting, store restocking, shelf organization, e-commerce picking), logistics, manufacturing, future home services.
Technology advantage
Native 4D world-action model unifying spatial representation, physical constraints, spatiotemporal reasoning, safety evaluation and action output — moving beyond 2D generation/simulation paradigms. Data flywheel: 100k hours cold-start, ~1M hours annual increment from real supermarket deployments. Chief scientist Liu Zhe (HUST PhD under Prof. Bai Xiang, HKU postdoc) brings 35+ top papers / 2,380+ citations in 3D/4D perception & autonomous-driving world models (EPNet, TANet, LION, GenieDrive, DrivePI, FASTER, Metis).
How they differentiate
Unlike peers chasing video generation/simulation ("virtual hothouses"), Sota Wujie deploys directly into real physical environments (European supermarket shelves) to train "scenario intelligence." Uses an agentic Physica-Claw OS now, with a path to collapse capabilities into a unified world-action model for scaling. Combines native 4D world modeling, proprietary full-modal data collection, and founder's unique cross-border compliance/deployment experience (ex-Huawei Europe, prior partnerships with 7 of the world's top-10 supermarket retail groups).
Main competitors
无界动力 (Wujie Power, embodied general-purpose brain / latent-space world model, ~$200M+ angel rounds, first $100M+ overseas order), 正行创新 (Zhengxing Innovation, physical-intelligence embodied AI, ~$100M angel round), 星动纪元 (Star Dynamics, industrial logistics, thousand-unit mass production), 智在无界, 自变量机器人 (AutoAgents) & 千寻智能 (Qianxun, end-to-end embodied route), 破壳时刻, NVIDIA GR00T as reference model.
Key partnerships
Europe's largest supermarket group (long-term framework agreement, ~1,000 robots forecast over 3 years, starting with POC), selected as a Shanghai municipal AI key project and signed at WAIC 2026 closing ceremony, open ecosystem collaboration with China's core robot-component supply chain partners, academic advisors incl. Stanford robotics professor, German humanoid-robotics research lead, SJTU robotics professor.
Notable customers
Europe's largest supermarket group (strategic partnership, first-phase POC, ~1,000-unit order forecast if POC passes).
Major milestones
Founded April 2026 by ex-Kepler CEO Hu Debo, signed long-term framework agreement with Europe's largest supermarket group (~1,000 robots over 3 years), selected as Shanghai municipal AI key project and signed at WAIC 2026 closing ceremony, plans to demo full brain capability (world model + multimodal VLA + Physica-Claw OS) in summer 2026 and run European POC by end of 2026.
Market positioning
Positions itself as the "machine brain" layer for embodied AI rather than a full-robot maker — a "physical-intelligence OpenClaw" / brain-driven full-ecosystem leader. Differentiates via vertical-convergence-then-generalization path: sink a native physical foundation model into one high-complexity scenario (supermarkets) first, then generalize to home and industrial logistics. Targets overseas markets where China's hardware supply chain + founder's global client relationships give an edge.
Geographic focus
China (Shanghai HQ) with primary go-to-market focus on Europe and US commercial scenarios.
About 胡德波 (Hu Debo)
Ex-CEO & co-founder of Kepler Robotics (开普勒机器人, humanoid robotics, 2023-2026); Ex-Huawei overseas country representative (Europe, from 2006); Ex-ZTE (joined 1999, led Europe/Russia/Asia-Pacific overseas business from 2001); Founded drone & underwater-robot startup in 2015 (partnered with HiSilicon on vision-AI drones). Royal Institute of Technology (Sweden) graduate. 70后 (born 1970s) veteran of global markets.