Shengshu partners with Anyverse Dynamics on robot world models
Shengshu Technology and embodied-robotics startup Anyverse Dynamics have signed a strategic partnership to build video-based generative world models for robots, though no products, funding or timelines were disclosed.
Shengshu Technology, the multimodal and world-model company behind the Vidu video model, and Anyverse Dynamics, a developer of general-purpose embodied robots, announced a "comprehensive strategic partnership" on March 12.
Under the agreement, Shengshu will contribute its multimodal foundation models and world models. Anyverse Dynamics brings its work on an embodied "universal brain" and "manipulation intelligence," along with hardware-software integration and deployment. The companies say they will jointly tackle technical bottlenecks in embodied intelligence and pursue scaled applications globally.
The technical idea behind the deal is that video captures physical space-time, causality and dynamics. That would make video a natural bridge between perception and action. The partners plan to pair the "physical understanding and imagination capabilities" of video generation models with robot execution. The goal is generative world models that let robots run a "mental rehearsal" before acting. The announcement presents this as a fix for the fragmented perception-reasoning-action pipelines that it says keep robots from operating reliably in complex real-world environments.
Anyverse Dynamics CTO Xu Wenda and Shengshu world model head Tan Hengkai signed the agreement. Shengshu founder Zhu Jun and CEO Luo Yihang attended, as did Anyverse founder and CEO Zhang Yufeng and co-founder Wang Yingyun.
The announcement does not give financial terms, specific models or robots, benchmarks, or a schedule for joint releases. For now, the partnership is a statement of intent rather than a product launch.