Business & FundingNov 3, 2026

China Physical AI Funding Roundup: Robot Deals Expand

Robot-training data and commercial deployment deals lead China's latest physical AI funding wave.

Funding Flows Into Training Data and Embodied Models

Deals Signal Faster Commercialization in Coffee and Logistics

What Physical AI Developers Should Watch Next

China physical AI funding is moving beyond single pilot projects and into the layers that make robots useful at scale. This week's roundup shows money flowing to training-data infrastructure, cross-embodiment robot models, and commercial deployment deals rather than just another humanoid demo. GenRobot AI's $100 million Series A+ is the headline, with backers including SAIC Motor, Baidu, and Galbot. Its capture rigs and Matrix platform turn human demonstrations into labeled training data, which is one of physical AI's hardest bottlenecks.

The second big theme is embodied AI models. RoboScience closed an A2 round of several hundred million yuan led by Amber Capital, Hongruida, and Shanghai Semiconductor Industry Investment. CEO Tian Ye, formerly on Apple's AI platform team, founded the startup with Stanford classmate Shao Lin. Its Visics model learns manipulation skills that transfer across robot bodies, and the company says its data pipeline cuts the cost of a training sample to 0.05 yuan while its video library grew from 1 million to 4 million clips this year. Elite Robots, Juqiao Industrial, Yicheng Innovation, and iSAGEBOT also secured financing to expand their hardware and sensor work.

Commercial bets are scaling too. Anyverse Dynamics and JD.com signed a strategic agreement to bring K15 dual-arm wheeled semi-humanoids to more than 1,000 7Fresh coffee shops, with product cooperation valued up to 1 billion yuan. Orbbec signed a 10,000-unit procurement agreement with Ace Robotics for depth cameras, RGB cameras, and handheld capture kits used to collect model training data. Pudu Robotics opened a European headquarters in Germany, while WeRide's Jingshuo subsidiary agreed to a 70 million yuan pre-Series A round.

For developers, the takeaway is that value is concentrating in the data and model layer. Whoever controls clean, cheap, large-scale demonstration data and can train a generalist robot model becomes essential infrastructure for every robot vendor. The trade-off is that many deal values and delivery timelines remain undisclosed, so execution risk still exists. Even so, the pattern is consistent: China's physical AI sector is moving quickly from announcements toward deployable systems.

If you build robots, watch how these startups turn data into a durable moat.

#china physical ai funding#robotics funding#physical ai#ai training data
View source ↗
China Physical AI Funding Roundup: Robot Deals Expand | Ali Fakih