Alibaba Pivots Toward 'World Models' with $290 Million Investment in ShengShu

Alibaba Cloud is aggressively expanding its footprint in the next generation of artificial intelligence, leading a $290 million (2 billion yuan) Series B investment in ShengShu. The startup, known for its Vidu video generation tool, is focusing on 'world models' designed to replicate the physical world more accurately than traditional large language models (LLMs). This strategic shift acknowledges the inherent limitations of text-centric AI. By leveraging multimodal data—including vision, audio, and touch—ShengShu aims to bridge the gap between digital simulations and physical applications. The funding round also saw participation from Baidu Ventures and TAL Education, following a previous 600 million yuan raise from Qiming Venture Partners. The investment aligns with Alibaba's broader strategy to dominate embodied AI. The Chinese tech giant has recently deployed similar capital into other specialized firms, including a $50 million investment in Tripo AI and a $60 million investment in PixVerse, both of which are exploring AI grounded in physical space. The development of general world models is viewed as a critical prerequisite for the advancement of humanoid robots and autonomous systems. By connecting perception and action, these models allow AI to predict real-world behavior consistently, potentially unlocking new industrial, commercial, and home applications for robotics.