Robotics Startup XDOF Nears Unicorn Status on $1.2B Valuation

Founded in 2024 by UC Berkeley researchers Philipp Wu and Fred Shentu, the startup serves as an outsourced data supply chain for the robotics industry. By providing the pipelines, collection tools, and annotation systems required to train general-purpose robots, XDOF addresses a critical bottleneck in the field: the lack of large-scale, high-quality physical datasets. Investors have likened the startup’s market position to Scale AI, noting that while LLMs drew on the internet for training, physical robots currently lack a comparable data source.

The company’s foundation lies in the GELLO project, a low-cost teleoperation system developed by the founders to enable remote robotic control. Today, XDOF utilizes this technology alongside teams of global human collectors who record everyday tasks, such as folding laundry, to feed their proprietary datasets. Having already secured 20 customers, including prominent frontier AI labs, the firm is currently collaborating with UC Berkeley’s AI Research lab to release ABC, a massive collection of robot training data. While terms of the pending investment remain fluid and neither XDOF nor 8VC have commented, the rapid interest underscores the intense demand for high-fidelity data in the race to build autonomous physical machines.