Inside Westlake University's New World Model Research Center
Westlake University, a private research university in the eastern Chinese city of Hangzhou, has established a new center devoted to world models — AI systems that build internal simulations of how the physical world behaves. The Westlake World Model Research Center was inaugurated on October 31 at the university’s Yungu campus, according to the technology outlet IT Home. It is described as a university-level interdisciplinary platform aimed at physical intelligence, the branch of AI concerned with machines that sense, move and act in real space rather than only in text.
The center’s leadership is a notable pairing of academia and industry. Its director is Yu Kaicheng, a doctoral supervisor and assistant professor in the university’s engineering school who is also co-founder and chief scientist of KMind, an AI company. The chief scientist role goes to Jin Yaochu, an AI chair professor and head of the university’s AI department. A group of principal investigators will take on research tasks across different directions.
World models have become one of the most crowded frontiers in AI research. The idea is to give a model an internal representation of physics, spatial relationships and cause and effect, so that a robot or an autonomous system can predict what will happen next instead of learning every situation from scratch. That capability is widely seen as a prerequisite for reliable robots, self-driving vehicles and other embodied systems that must cope with messy real environments.
For Chinese AI research, the center reflects a broader shift in how the country’s institutions organize work on hard problems. Companies rather than universities have driven much of the recent progress in large models, and several academic labs have responded by building structures that sit closer to industry — joint appointments, startup founders leading university centers, and platforms explicitly aimed at commercial bottlenecks. Westlake’s model, with a sitting startup executive running the center and a senior professor serving as chief scientist, follows that template.
Hangzhou gives the center a dense local ecosystem to draw on. The city is home to major AI and e-commerce firms and a growing cluster of robotics and embodied-AI startups, several of which have raised significant funding in the past year. A university platform focused on physical intelligence can plausibly serve as a shared source of talent and early-stage research for that cluster, though the announcement did not detail budgets, staffing numbers or specific projects.
The center’s stated goal is to break through core bottlenecks in the field. That is a broad ambition, and world models remain an open research problem globally: training them requires enormous amounts of video and sensor data, and validating whether a model truly understands physics is difficult. Chinese teams have been active in the area, and the competition to build systems that generalize beyond curated demonstrations is intense on both sides of the Pacific.
What the center actually produces will matter more than its launch. University research platforms in AI often take time to show results, and the involvement of a company co-founder raises the question of how openly findings will be shared. Still, the creation of a dedicated, cross-disciplinary institute signals that Chinese universities want a seat at the table in embodied AI, not just a supporting role supplying graduates to industry labs.
The next things to watch are concrete: which principal investigators join, what data and computing resources the center secures, and whether its first published work targets simulation, robot control or something narrower. For global readers tracking where physical AI is being built, Hangzhou is now a place to keep on the map.