August 28, 2026, (Inside AI) — Chinese embodied-AI startup PsiBot has closed a new funding round exceeding US$100 million, drawing industrial investors from automotive, electronics, and gaming sectors. The capital will accelerate research into embodied world models, human-operation data collection, and deployments in logistics and advanced manufacturing.
The round includes Ningbo Tuopu Group, a fund backed by Chery Holdings, Lens Technology, 37 Interactive Entertainment, Wuhu Investment Holding Group, and Fosun Fortune Capital. Existing shareholder Zhuhai Technology Industry Group also continued to invest.
PsiBot focuses on dexterous manipulation, a subfield of embodied AI where robots learn to handle objects with human-like precision. The startup is building a dual-system architecture that pairs its Psi-R2 operation-policy model with the Psi-W0 action-conditioned world model.
Industrial Backers Signal Shift Toward Physical AI
The investor lineup reveals a strategic pattern. Ningbo Tuopu Group supplies automotive components, while Chery Holdings is a major automaker. Lens Technology makes glass and touch modules for consumer electronics. 37 Interactive Entertainment brings gaming and simulation expertise. These backers are not passive financial players; they represent factories, supply chains, and production lines where dexterous robots could soon be deployed.
Read: PaXini Raises RMB1 Billion to Scale Tactile-Sensing Chips for Robots
This mirrors a broader trend in China’s embodied-AI sector. Industrial conglomerates are increasingly funding robotics startups not just for returns, but to secure early access to automation technology. PsiBot’s focus on logistics and advanced manufacturing aligns directly with the operational needs of its new investors.
The dual-system architecture is technically notable. An operation-policy model like Psi-R2 decides what action to take, while an action-conditioned world model like Psi-W0 predicts what happens next. This pairing allows robots to simulate outcomes before executing movements, reducing errors in unstructured environments.
Data Collection Remains the Bottleneck
PsiBot’s plan to invest in human-operation data collection addresses a critical industry challenge. Unlike language models that train on internet text, dexterous manipulation requires physical demonstration data. Collecting this data is expensive and slow, often involving human operators wearing motion-capture suits or teleoperating robot arms.
The startup’s emphasis on embodied world models suggests a bet on simulation and prediction to reduce reliance on real-world trials. If Psi-W0 can accurately model physical interactions, PsiBot could generate synthetic training data at scale, a potential competitive moat.
Competing Chinese embodied-AI firms have taken different paths. Some prioritize full humanoid robots, while others focus on specific tasks like welding or sorting. PsiBot’s dexterous-manipulation niche targets high-value, precision-dependent jobs that remain difficult to automate.
Read: Xiaomi Open-Sources Embodied AI Foundation Model Xiaomi-Robotics-1
Sources indicate the funding will also support deployments in logistics, where parcel sorting and palletizing demand flexible grasping. Advanced manufacturing, including electronics assembly, requires fine motor control that traditional industrial robots lack.
PsiBot did not disclose its valuation or the exact funding amount beyond the US$100 million threshold. The round’s industrial composition suggests the startup is positioning itself as a practical automation partner rather than a research lab.
Zhuhai Technology Industry Group’s continued investment signals confidence from existing backers. The state-linked entity likely sees PsiBot as aligned with regional industrial policy goals in Guangdong province, a manufacturing hub.
The embodied-AI sector in China has attracted billions in funding over the past two years. Government initiatives encourage robotics adoption in factories facing labor shortages and rising wages. PsiBot’s new capital gives it runway to compete in a crowded field.
