Introduction

XDOF has emerged from stealth with a clear mission to build the data infrastructure that powers general-purpose robots. Less than three months after coming out of stealth, the startup is already in advanced discussions for a major funding round, signaling strong investor appetite for companies solving the critical data-collection bottleneck in physical AI. The company focuses on collecting real-world teleoperation data through remote operation and sensor-equipped human operators, creating the training sets needed to teach robots everyday tasks.

What Happened

XDOF is in late-stage talks to raise a Series B round valuing the company at approximately $1.2 billion, with 8VC reportedly leading the investment. The round comes just months after the startup emerged from stealth and closed a $70 million Series A in June, which included participation from Andreessen Horowitz, Lux Capital, and Spark Capital. Despite not having planned to return to the market so soon, XDOF's annualized revenue of around $50 million has made the new round attractive to venture firms. Deal terms remain fluid and could still change, though the company and 8VC declined to comment on the discussions.

  • Founded by UC Berkeley researchers Philipp Wu and Fred Shentu in 2024
  • Emerged from stealth less than three months ago
  • Closed a $70 million Series A in June with Andreessen Horowitz, Lux Capital, and Spark Capital
  • Annualized revenue now approaching $50 million

Why This Matters

Physical robots lack the equivalent of internet-scale text data, making real-world data collection the primary bottleneck for general-purpose AI. XDOF positions itself as the Scale AI or Mercor for robotics, offering an outsourced data-supply chain that captures teleoperation signals, sensor recordings, and annotated task performance. The startup has partnered with UC Berkeleys AI Research lab to release what it calls the ABC dataset, billed as the largest collection of high-quality robot training data ever assembled. By combining remote robot teleoperation with human collectors wearing sensors to record everyday actions like folding clothes or flattening boxes, XDOF aims to create a scalable pipeline for physical AI development.

  • XDOF describes itself as the Scale AI or Mercor for physical robotics
  • Data scarcity remains the primary barrier to deploying general-purpose robots at scale
  • Partnering with UC Berkeley to release the ABC dataset, the largest high-quality robot training set to date
  • Plans to expand data collection globally through remote teleoperators and sensor-wearing operators

Key Takeaways

  • XDOF is in advanced talks for a $1.2 billion valuation just months after emerging from stealth, led by 8VC
  • The startups rapid revenue growth to $50 million annualized made the round inevitable despite the short interval since its Series A
  • XDOF aims to solve the data scarcity problem facing general-purpose robotics through a global teleoperation and sensor network
  • Backed by partnerships with UC Berkeley and already working with 20 AI and robotics labs, XDOF is positioning itself as critical infrastructure for physical AI

Conclusion

As XDOF moves toward closing its Series B, the startup underscores a fundamental truth about the next wave of AI hardware: breakthroughs in general-purpose robotics will be limited by data supply, not algorithmic theory. By building an outsourced, global data-collection network and preparing to release one of the largest robot training datasets ever assembled, XDOF is staking its position as foundational infrastructure for the physical AI era. Investors and industry observers will be watching closely to see how the $1.2 billion valuation discussions unfold and what they signal for the broader robotics funding landscape.