XDOF Enters Market: Data Infrastructure for Robot Training and $70M Raised

Friends, sharing an update from AI and robotics: startup XDOF builds infrastructure to collect and prepare data for robots.

- Problem: unlike LLMs, robots lack large high-quality datasets of physical interaction.
- What XDOF does: collects teleoperation and egocentric data, develops cleaning and annotation tools; raised $70M and serves ~20 clients.
- ABC dataset includes 130,000 trajectories, 300 hours of simulation and 100 hours of evaluations — scale previously unavailable to academia.
- The company also supplies operators and sensors — services that larger labs often outsource.

Why it matters: data quality and scale may become the main bottleneck for physical AI.

Would you invest in specialized robot data providers?

#AI #robotics #data #infrastructure