AI for Business

Alibaba Bets $290 Million on AI That Sees the World

Alibaba Cloud is placing a major wager on a different path for artificial intelligence. The company announced it has led a 2 billion yuan ($290 million) investment in ShengShu, a startup known for...

Share:

Alibaba Cloud is placing a major wager on a different path for artificial intelligence. The company announced it has led a 2 billion yuan ($290 million) investment in ShengShu, a startup known for its Vidu video generation tool. The move signals a strategic pivot toward developing "world models," a class of AI built to understand physical environments, not just process text.

This investment highlights a growing industry acknowledgment: large language models (LLMs) like those behind ChatGPT have inherent constraints. They are trained predominantly on written language, which lacks the sensory and spatial data needed to model real-world physics. In contrast, world models are engineered using multimodal data—video, audio, and even touch—to create a more coherent simulation of how objects and forces interact.

ShengShu stated the new capital will fund work on a "general world model" intended to bridge digital simulations (like games and AI video) with physical applications such as autonomous driving and robotics. "We aim to connect perception and action," said founder Zhu Jun, suggesting the goal is to create AI that can predict and respond to real-world events with greater consistency.

The company is not operating in a vacuum. Alibaba, alongside Baidu Ventures, recently invested in Tripo AI, which generates 3D models from photos and is also pursuing world model technology. In September, Alibaba backed PixVerse, which released a world model for interactive video generation earlier this year. These parallel investments underscore a concentrated push into AI that interacts with space and matter.

For business leaders, the implication is clear. The next frontier for practical AI deployment may depend less on conversational agents and more on systems that can navigate and manipulate complex environments. As noted by Wired co-founder Kevin Kelly, achieving human-like intelligence in machines requires reasoning, world understanding, and continuous learning. While LLMs contributed a form of knowledge, the understanding of the physical world remains a significant hurdle. Alibaba's latest financial commitment is a direct attempt to clear it.

Source: CNBC

Ready to Modernize Your Business?

Get your AI automation roadmap in minutes, not months.

Analyze Your Workflows →