XDOF, a startup specializing in collecting real-world teleoperation data for training general-purpose robots, is in advanced discussions to raise a Series B round at a valuation of around $1.2 billion. Sources familiar with the deal indicate that 8VC is leading this funding round.
This potential funding comes less than three months after XDOF emerged from stealth mode. The company previously secured a $70 million Series A round in June, with participation from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. While XDOF did not initially plan to raise capital so quickly, its rapid expansion and annualized revenue nearing $50 million drew interest from venture capital firms, prompting the new round.
What XDOF Does in Robotics Data
Co-founded in 2024 by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO), XDOF aims to provide critical data infrastructure for the robotics industry. The company builds data pipelines, collection tools, and annotation systems that frontier AI labs and robotics companies often struggle to develop internally, essentially acting as an outsourced data-supply chain.
The company’s foundation stems from Wus PhD research into how robots learn from large datasets, where he identified a significant barrier: the lack of extensive, real-world data. This led to the creation of GELLO, a low-cost teleoperation system that allows humans to remotely control robotic arms to generate training data. This research became core to XDOF, which investors now compare to data-labeling giants like Scale AI or Mercor, but for physical robotics.
Unlike large language models (LLMs) that initially trained on vast internet data, physical robots lack a comparable real-world dataset. This makes high-quality data collection a key bottleneck in developing general-purpose robotic machines. XDOF addresses this by combining remote robot teleoperation with human collectors who wear sensors to record everyday tasks, such as folding clothes or flattening boxes.
XDOF is also partnering with UC Berkeleys AI Research lab to release ABC, which it believes is the largest collection of high-quality robot training data ever assembled. The startup plans to hire and train data collection teams globally, including teleoperators and egocentric operators. XDOF previously reported having 20 customers, including several frontier AI labs. Other companies attempting to collect real-world data for robot training include Mecka AI, alongside human-data platforms like Scale AI and Micro1.
Leave a Reply