Alibaba Cloud has led a $290 million funding round in Chinese startup ShengShu, signalling a strategic shift toward a new frontier in artificial intelligence known as “world models”.

The investment, which also drew participation from Baidu Ventures and TAL Education, will support ShengShu’s development of its AI video platform Vidu and a broader ambition to build a general-purpose system capable of understanding how the physical world works.

At the centre of this push is the concept of a “world model”, an AI system trained on multimodal data such as video, audio and sensory inputs to simulate real-world environments. Unlike traditional large language models that focus on text, these systems aim to grasp physics, spatial relationships and cause-and-effect, making them more suitable for applications like robotics and autonomous driving.

ShengShu says it plans to use the new funding to create a model that bridges digital and physical environments, linking areas such as gaming and AI-generated video with real-world use cases like industrial automation and robotics. The goal is to enable machines not just to interpret information, but to act on it in a consistent and predictive way.

Read also

The move reflects a broader shift within the AI industry, where companies are beginning to recognise the limitations of text-based systems in handling real-world complexity. While language models have excelled in tasks like writing and coding, they struggle with physical reasoning — a gap that world models are designed to fill.

ShengShu’s latest model, Vidu Q3 Pro, is already gaining traction in the competitive AI video generation space, ranking among the top systems capable of producing video from text and images. Notably, the company launched its technology globally ahead of several high-profile rivals entering the same space.

For Alibaba, the deal is part of a wider investment spree aimed at strengthening its position in next-generation AI. The company has recently backed other startups working on 3D modelling and video-based AI, while also releasing its own open-source tools and robotics-focused models.

The latest investment underscores a growing belief across the tech industry: the future of artificial intelligence may depend less on better chatbots and more on machines that can understand and interact with the real world.