Axis Robotics has closed a $12 million seed round led by Hack VC, with participation from Nomad Capital, Pi Network Ventures, 10K Ventures, and unnamed angel investors. The company plans to use the money to build what it calls a massively parallel, human-in-the-loop global data engine for Physical AI.
Why Physical AI needs a new data engine
Physical AI — robots that can perceive, reason, and act in the real world — faces three big hurdles: data scarcity, a generalization gap, and embodiment fragmentation across different robot hardware. Most robot data today is collected in controlled labs, which doesn't scale. Axis wants to fix that by getting humans into the loop at scale.
What the engine includes
Axis's Compounding Data Engine has four parts: a Task Gen Engine, a Browser-Based Sim Teleoperation Platform, an Ego Data Mobile Capture App, and a Data Processing Pipeline. The browser-based platform is the first web-based interface for generating high-quality robotic motion trajectories remotely. The company says it achieves 10 times higher throughput than lab-based collection.
More than 100,000 active contributors are already submitting data three to four times a day. That adds up to over 1,200 hours of simulation data and more than 20,000 hours of real-world ego-centric data every month.
Benchmark results and commercial partners
Axis recently launched Sim Dataset V1. On the LIBERO-100 benchmark, the dataset improved overall success by 4.9 points and outperformed a volume-matched RoboCasa baseline by 31.3 points. The company also has commercial partnerships with Booster Robotics, Manycore Technology, Feagine Robotics, Lotus Cars, Geely Auto, and SomaStacks.
The funding will go toward scaling the data engine further. Axis hasn't disclosed a specific timeline for its next milestones, but the seed round gives it room to grow its contributor base and expand its commercial reach.
