Skip to main content
AI Market Watch
Loading...

Axis Robotics

Category: AI Infrastructure

Axis Robotics is building a distributed, crowdsourced data infrastructure for Physical AI, enabling anyone with a web browser to generate robot training data through browser-based simulation. Axis Robotics was founded in 2025. The company is led by Chris Feng. Based in Berkeley, California, United States. Team size: 11-50. Total funding raised: $12M. Latest round: Seed. Key investors include Hack VC, Nomad Capital, Pi Network Ventures, 10K Ventures.

Founded
2025
Headquarters
Berkeley, California, United States
Team size
11-50
Total funding
$12M

Value proposition

Axis Robotics solves the robot training data bottleneck by enabling anyone with a web browser to generate diverse, high-quality training data through browser-based simulation — replacing expensive teleoperation setups ($40K+/robot) with a distributed community model that produced 100K trajectories in 5 days from 18,000+ contributors.

Products and solutions

Robot Training Data Infrastructure (browser-based simulation platform), MetaSim / RoboVerse (cross-simulator meta-simulation layer), Task Generation Engine (multi-dimensional procedural generation; 207 tasks / 50K+ demos in V1), Sim Data Collection Platform (browser MuJoCo-WASM teleoperation), Mobile Egocentric App (real-world hand-pose capture), Data-Processing Pipeline (cleaning, domain randomization, model-ready formats), Compounding Data Engine (closed-loop generation, collection, training, DAgger post-training), Sim Dataset V1/V2, DAgger Dataset

Unique value

Only platform combining browser-based simulation, community-driven data collection, and proven sim-to-real transfer — community-contributed data has been used to train policies deployed on physical robots performing autonomous tasks. Crypto-powered incentive layer on Base blockchain for verifiable, ownership-ready datasets.

Target customer

Robot manufacturers, Physical AI companies, robotics research labs, industrial automation firms, and autonomous vehicle companies needing diverse robot training data at scale

Industries served

Robotics, Physical AI, Industrial Automation, Autonomous Systems, Manufacturing

Technology advantage

MetaSim cross-simulator abstraction layer (MuJoCo WASM-based); Task Generation Engine with multi-dimensional procedural variation (207 tasks / 50K+ demos in Sim Dataset V1); Browser-based data collection (zero hardware requirement); Proven sim-to-real transfer (Little Prince's Rose campaign); Crypto-governed contributor network with on-chain verification; Sim Dataset V1 lifted π0.5 LIBERO-Plus overall success from 83.9 to 88.8 (+4.9 pp) and beat volume-matched RoboCasa365 by 31.3 pp (arXiv:2607.21588)

How they differentiate

Unlike centralized lab-based teleoperation (Scale AI, XDOF) which requires expensive hardware and dedicated operators, Axis uses a distributed browser-based model where anyone can contribute. This enables 10x-100x more task diversity at a fraction of the cost. The crypto-native incentive layer on Base blockchain also differentiates it from traditional data infrastructure companies.

Main competitors

Scale AI (Physical AI data annotation), XDOF (robot training data pipelines), Physical Intelligence (π) (in-house robot foundation model training)

Key partnerships

Booster Robotics, Manycore Tech, Feagine Robotics, Dexmal, Lotus Cars, Geely Auto, Zeroth, AgiBot, Deep Robotics, EngineAI, Unitree, BitRobot

Notable customers

Booster Robotics, Manycore Tech, Feagine Robotics, Dexmal, Lotus Cars, Geely Auto, Zeroth, SomaStacks

Major milestones

July 27, 2026: Raised $12M seed round led by Hack VC, Q1 2026: Little Prince's Rose — 10,000+ valid trajectories in 3 days; product beta ~180K trajectories / 20,000+ users, March 2026: Main product launched on Base blockchain, Sim Dataset V1: π0.5 LIBERO-Plus 83.9→88.8 (+4.9 pp); +31.3 vs volume-matched RoboCasa365 (arXiv:2607.21588), 100,000+ global contributors, 1,200+ hours simulation data and 20,000+ hours real-world data collected monthly, July 2026: Axis V2 launched with human-gated DAgger post-training tasks

Growth metrics

100,000+ global contributors; 452.8 PFLOPS total hash power connected; 19,280.2 on-chain agents; 8,902 active nodes; 1,200+ hours simulation data/month; 20,000+ hours real-world data/month

Market positioning

Early-stage leader in the emerging Physical AI data infrastructure layer, positioned as the "crowdsourced data engine for Physical AI" — competing with centralized data collection approaches by leveraging distributed human participation and blockchain-based incentives.

Geographic focus

Global (headquarters in Berkeley, CA with global contributor network; partnerships with Chinese robotics companies like Booster Robotics, Geely Auto)

About Chris Feng

Ex-COO of Chainbase (blockchain data network); Senior Associate prior; Finance & Risk Management concentration at Fisher College of Business, The Ohio State University

Latest news about Axis Robotics

More AI Infrastructure companies

Official website: