Axis Robotics, a Physical AI data company founded in 2025, has raised $12 million in seed funding to scale its Compounding Data Engine. The round was led by Hack VC and included Nomad Capital, Pi Network Ventures, 10K Ventures, and several angel investors.

The investment will expand Axis's end-to-end system for generating the large volumes of diverse, structured robotic training data that Physical AI still lacks. While language models draw on vast internet text, robotic systems need billions of high-quality human-physical interaction trajectories. Current datasets remain scarce, poorly generalized across environments, and fragmented by different robot hardware.

Axis closes this gap with a closed-loop workflow that combines large-scale simulation, real-world egocentric capture, and continuous human-in-the-loop refinement. A Task Generation Engine produces highly varied robotic tasks through randomization of objects, layouts, and embodiments. Contributors then supply motion data either through a browser-based simulation platform or a mobile app that records real-world hand and body movements. Human corrections feed directly back into model training, creating a self-reinforcing cycle.

The company already runs a global network of more than 100,000 active contributors, these users generate over 1,200 hours of simulation data and more than 20,000 hours of real-world data each month. According to Axis, pretraining on its Sim Dataset V1 lifted success rates on the LIBERO-Plus benchmark by several percentage points and improved robustness to layout changes, sensor noise, and pose variation.

"Physical AI demands billions of human-physical interaction motion trajectories," said founder Chris Feng. "For years the industry lacked an efficient, infinitely scalable hybrid data production system. That's exactly what we built—a compounding data engine that links a global contributor network with constant model iteration."

Axis, headquartered in Singapore, is already supplying customized training packages to robot manufacturers and AI companies, including Booster Robotics, Manycore Tech, Dexmal, Lotus, and Geely Auto. The new capital will enlarge the contributor network, strengthen procedural generation, and support the release of Sim Dataset V2 in September and a DAgger Dataset in November.

Axis is positioning itself as core infrastructure for Physical AI.

This story was also published on my newsletter called Spyrigend.

Tip Jar

Enjoy the article? Consider leaving a tip to support my writing efforts. All donations are greatly appreciated!

Bitcoin: 3BcFDV3DYHT8rZurPWUF9kZUkqSFQ9kfAS

Etherium: 0x8AD199f90bf118a8ed708C73AA199EF7EF5444ba

Solana: A8cxXWcs6ARYNzgjK2nN4jkgh8Xx7a5EJFt8JeTRZ2rv

Disclaimer: This article is written purely for entertainment and educational purposes and should not be taken as financial advice in any way. Do your own research. If you are seeking financial advice, find a professional who is right for you.