Axis Robotics raises $12M seed to fix Physical AI's data bottleneck
What's the deal? Axis Robotics has raised $12 million in seed funding led by Hack VC, with participation from Nomad CapitalDealroom has a profile for this one. Try Dealroom →, Pi Network Ventures, 10K VenturesDealroom has a profile for this one. Try Dealroom →, and several angel investors. The company announced the round on July 27.
What's the endgame? Axis is building a large-scale, human-robot collaborative global data engine to solve the data problems slowing Physical AI: scarce training data, weak model generalisation, and fragmentation across robotic hardware. Unlike language models trained on internet text, Physical AI needs vast real-world human interaction data.
How it works: Its "composite data engine" merges task generation, data collection, model training, and optimisation. A browser-based remote operation platform runs robot simulations, while a mobile app collects real human motion data that is cleaned and annotated into multimodal datasets.
The company reports a global network of over 100,000 active contributors, generating more than 1,200 hours of simulated data and over 20,000 hours of real-world first-person data each month.
Why it matters: Axis claims its datasets outperform benchmarks. In the LIBERO-Plus test, the π0.5 model trained on Axis data improved its success rate by 4.9 percentage points — a 31.3-point gain over the RoboCasa365 benchmark dataset.
By the numbers: The raise ranks in the 95th percentile of all-time seed rounds in robotics, across a sample of 3,871 deals. Founder Chris said the company is constructing a "sustainable data production system" to accelerate general robotic intelligence.
The traction: Axis has signed partnerships with Booster Robotics, Manycore TechDealroom has a profile for this one. Try Dealroom →, Feagine Robotics, Dexmal, Lotus Car, and Geely Automobile, supplying custom training data to robot makers, Physical AI firms, and industrial automation companies.
What's next? The funds will expand data generation capacity, grow the contributor network, and develop the core infrastructure behind the next generation of general robotic intelligence.
The signal: As robotics startups race toward general-purpose machines, the constraint is shifting from hardware to data. Axis is betting that whoever controls the data engine — not the robot — will shape how fast Physical AI matures.
Read more: Cointime
