Friday, April 10, 2026: For the past two years, the tech world has been obsessed with AI that can talk. But Alibaba Cloud is placing a massive bet that the next era of intelligence won’t just chat, it will navigate.
In a decisive move to move past the limitations of Large Language Models (LLMs), Alibaba recently spearheaded a 2 billion yuan ($290 million) Series B funding round for ShengShu, the powerhouse behind the Vidu video generation engine. This isn’t just another venture capital injection; it’s a strategic pivot toward “World Models”, AI designed to understand the physical laws of our universe rather than just the grammatical laws of our language.
0 Interest in Sora: How Alibaba is Funding the ‘OpenAI Killer’
While ChatGPT and its peers are masters of predicting the next word in a sentence, they lack a fundamental grasp of cause and effect in the physical world. They don’t know that a glass shatters when dropped or how a car should lean into a curve.
Alibaba’s aggressive investment strategy, which includes recent backing for Tripo AI (3D modeling) and PixVerse (interactive video), suggests a “reality-first” philosophy. By focusing on multimodal data (vision, sound, and touch), these companies are building a digital “sandbox” where AI can learn the consequences of physical actions.
Alibaba’s $290M Plan to Build a Real-World Brain
ShengShu’s vision for a “General World Model” aims to dissolve the border between two massive industries:
- The Digital Realm: Enhancing AI-generated video and gaming environments with consistent physics.
- The Physical Realm: Providing the “brain” for autonomous vehicles and humanoid robots.
By connecting perception (seeing the world) with action (interacting with it), Alibaba is positioning itself as the infrastructure provider for the “Embodied AI” revolution, where robots move through homes and factories with human-like intuition.
What Happens Next?
- The Rise of “Action-Models”: Within the next 12–18 months, expect a shift from “Chat” interfaces to “Action” interfaces. We will see the first generation of robots that don’t require complex coding for every task but instead “understand” their environment through these world models.
- The “Physics Gap” War: As OpenAI’s Sora remains largely behind closed doors, Chinese firms like ShengShu, Kuaishou, and ByteDance are flooding the market. The winner won’t just be who makes the prettiest video, but whose AI best predicts real-world outcomes.
- Hardware Integration: Look for Alibaba to integrate ShengShu’s tech directly into its logistics robots and cloud-based autonomous driving platforms. The goal is a seamless loop: the AI sees a warehouse, simulates the most efficient path in its “world model,” and then executes it in real-time.
“To replicate human intelligence, AI needs more than just a library of books; it needs a sense of gravity, space, and time,” says an industry analyst. “The battle for the ‘Knowledge AI’ is over. The battle for ‘Physical AI’ has just begun.”
