Robot training data startup XDOF in talks for $1.2B Series B valuation

XDOF, a startup focused on collecting real-world data to train general-purpose robots, is in late-stage discussions for a new funding round valuing the company at roughly $1.2 billion, according to people familiar with the matter. The round would reportedly be led by venture firm 8VC.

The company was founded in 2024 by UC Berkeley researchers Philipp Wu, who serves as CEO, and Fred Shentu, the CTO. Just months ago, in June, XDOF raised a $70 million Series A with backing from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital.

Sources say XDOF hadn't planned to raise again so soon, but rapid growth — with annualized revenue nearing $50 million — drew renewed investor interest. The total amount being raised and whether it includes new capital in the valuation figure remain unclear, and terms of the deal are still not finalized.

An outsourced data pipeline for robotics

XDOF positions itself as a data-supply partner for frontier AI labs and robotics companies, building the pipelines, collection tools, and annotation systems needed to train physical robots — a task investors compare to what Scale AI and Mercor did for data labeling in the broader AI industry.

The approach traces back to GELLO, a low-cost teleoperation system Wu and Shentu built while researching the shortage of large-scale robotics datasets during their PhD studies. Unlike large language models, which could train on vast troves of internet text, physical robots lack an equivalent real-world dataset, making data collection a major bottleneck.

To address this, XDOF combines remote robot teleoperation with human collectors wearing sensors to capture everyday activities, such as folding clothes or flattening boxes. It is partnering with UC Berkeley's AI Research lab on a large robot-training dataset called ABC, and plans to hire and train global teams of teleoperators and sensor-wearing data collectors.

XDOF has previously said it works with 20 customers, including several frontier AI labs. Competitors in the space include Mecka AI, along with human-data platforms like Scale AI and Micro1 that are expanding into robotics.