Robotics Data
Companies in this report
AllSides builds high-fidelity 3D scanners that convert physical objects into relightable digital twins, powering one of the largest quality 3D datasets for physical AI and generative-3D model training. With customers including Amazon, Meta, and adidas, the Italian startup sits at the intersection of hardware scanning and AI-ready data infrastructure.
Build manufactures head-mounted cameras worn by factory workers to capture first-person video of real industrial tasks, then open-sources the resulting datasets for training humanoid and industrial robot foundation models. The company has released some of the largest egocentric-video datasets in existence — but its rapid scaling has also drawn investigative scrutiny over worker consent and compensation.
CortexAI supplies large-scale, real-world workplace data — egocentric video, robot trajectories, and human-in-the-loop rollouts — to robotics foundation model developers. As embodied AI labs race to train general-purpose robots, CortexAI is positioning itself as the data and evaluation layer the industry needs.
Mecka AI pays hundreds of thousands of contributors across 12 countries to capture human motion with iPhones and body sensors, then sells that data to embodied AI and humanoid robotics labs. Founded in 2025, it has raised $60M and claims nine-figure contracted revenue.
PrismaX is a San Francisco-based startup building infrastructure that connects robotics with mainstream adoption through AI, Web3, and blockchain technologies. Backed by a16z crypto and launched in 2024, the company sits at an unusual intersection of tele-operation, real-world data collection, and robotic foundational models.
XDOF builds the data-collection tools, pipelines, and annotation systems that frontier labs rely on to train robot foundation models. Founded by UC Berkeley researchers behind the GELLO teleoperation project, the company emerged from stealth in June 2026 with a $70M raise and the largest open-source bimanual robot manipulation dataset.