Chinese startup Psibot has reached a $1.48bn valuation as it builds around so-called “world models,” a branch of AI aimed at helping machines better understand and navigate the physical world. The company’s rise points to growing investor interest in robotics-focused artificial intelligence, especially ideas that move beyond text and into real-world action.
Founder Viktor Wang frames the opportunity through a familiar AI milestone. He points to OpenAI’s 2019 GPT-2 release as an early sign that large models trained on internet text could generate human-like writing. In his view, robotics may now be approaching a similar turning point, with AI systems beginning to show the potential to model environments and support more capable machine behavior.
That comparison helps explain why world models are drawing attention. While chatbots and text generators have dominated the recent AI boom, the next major phase could center on systems that can predict, plan and respond in physical settings. For startups like Psibot, the bet is that progress in this area could unlock more practical and adaptable robots.
Psibot’s valuation also underscores how competitive the AI race has become in China, where companies are looking for the next breakthrough beyond language models alone. Even so, turning early technical promise into reliable products remains a major test. For now, the company’s funding momentum suggests investors believe world models could become an important part of robotics’ next chapter.
