Chinese startup DexForce uses virtual data to train humanoid robots
3 sources · The Economic Times · Reuters · TradingView- Boom: DexForce is training humanoid robots with virtual data to cut development costs
- Neutral: The Chinese startup published its approach on October 15, 2026
- Boom: Virtual data replaces costly real-world data collection in robot training pipelines
The story in full
DexForce, a Chinese robotics startup, is training humanoid robots using virtual data as a method to reduce costs, according to reports published on October 15, 2026.
The approach centers on synthetic or simulated training data rather than expensive real-world collection, a technique aimed at making humanoid robot development more economically viable. DexForce's specific cost figures, robot models, and further technical details were not available in the source material.
Analysis
317 wordsOn October 15, 2026, Chinese robotics startup DexForce disclosed that it is training humanoid robots using virtual, simulated data rather than data collected from real-world physical interactions. The announcement drew coverage from Reuters and other outlets on the same day. Specific cost figures, the names of DexForce's robot models, and granular technical details about the simulation pipeline were not included in the available source material.
The significance of the approach lies in a well-known bottleneck in robotics development: gathering high-quality real-world training data is slow, expensive, and difficult to scale. If synthetic data can reliably substitute for physical data collection without sacrificing robot performance, it could meaningfully lower the barrier to entry for humanoid robot development. DexForce operates in a crowded and competitive Chinese robotics sector, where reducing costs is a strategic priority as companies race to bring capable humanoid robots to market. What remains genuinely in dispute is whether virtual training data produces robots that transfer their learned behaviors reliably to real environments, a problem researchers call the simulation-to-reality gap.
No reactions from the Pro-AI, Anti-AI, or Middle Ground camps had been published at the time of writing. The Pro-AI camp would typically highlight this as evidence that AI-driven simulation is accelerating robotics and democratizing access to advanced development tools. The Anti-AI camp would likely raise concerns about the safety and reliability of robots trained on synthetic rather than real-world data, and flag the geopolitical dimension of Chinese firms rapidly advancing humanoid robotics capabilities. The Middle Ground camp would probably acknowledge the cost-reduction logic while pressing for rigorous real-world validation before deployment in any consequential setting.
The key things to watch are any follow-up technical disclosures from DexForce about how their simulated data is generated, independent benchmarks comparing their robots' real-world performance to those trained on physical data, and whether other Chinese or international robotics firms adopt or respond to this approach in the months ahead.
Where do you stand?
Add your take
0 reader votesSign in with Google to pick a side and post. Your vote moves the story's Doom / Boom score.