DexForce raises 1 billion yuan Series B, betting virtual training can crack the humanoid data bottleneck

The Shenzhen startup says synthetic data beats collecting human demos; backers include Lenovo Capital and Lens Technology.

Shenzhen-based startup DexForce has raised 1 billion yuan (about $147 million) in a Series B round that values the company at more than 10 billion yuan, founder and chief executive Jia Kui said in an interview with Reuters published on October 15. The company is betting that synthetic data generated in virtual environments - not thousands of hours of human demonstrations - is the way to teach humanoid robots real-world tasks.

Jia, a machine-learning professor at the Chinese University of Hong Kong, Shenzhen, founded DexForce in 2021. The roughly 500-person company builds AI software that simulates physical environments and predicts how objects move and interact: it creates a digital version of a workplace, trains the robot in simulation, then transfers what it learned to physical machines. Reaching what Jia calls 'physical artificial general intelligence' - a humanoid that can take on new tasks in unfamiliar places on command - purely through real-world data collection would require 10 trillion samples, he said, making manual collection 'fundamentally inefficient.' Those are the company's statements from the interview.

The round's backers include Lenovo Capital, the investment arm of Lenovo, Lens Technology, and several state-owned funds. DexForce says its software is already used in industrial settings by Toyota, Samsung, Panasonic, Foxconn, Haier, and Hisense, while its humanoid robots make coffee and serve customers at malls and tourist sites in more than 15 Chinese cities. Jia declined to comment on reports that the company is considering a Hong Kong listing.