Less than three months after emerging from stealth mode, XDOF is in advanced talks led by 8VC to raise a Series B round at an approximately $1.2B valuation. Founded in 2024 by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO), the company announced a $70 million Series A round in June with participation from Thrive Capital, Andreessen Horowitz, Lux and Spark Capital. While XDOF’s annualized revenue is approaching $50 million, the total size of the new round and whether the valuation includes the new investment are unknown; the terms of the deal have not been finalized.
The company collects real-world teleoperation data to train general-purpose robots and aims to provide the data-collection, tooling and labeling infrastructure that robotics companies and advanced AI labs would struggle to build on their own. XDOF is built on the low-cost GELLO system previously developed by Wu and Shentu, which enables people to remotely control a robotic arm to generate training data.
XDOF sees the lack of a comprehensive training dataset like the internet for physical robots as a major bottleneck. In collaboration with UC Berkeley’s AI Research lab, the company plans to release ABC, which it believes is the largest collection of high-quality robot training data to date. In the data-collection process, it uses not only teleoperators who remotely control robots but also people who record everyday tasks such as folding clothes and flattening boxes with sensors. Planning to establish data-collection teams around the world, XDOF previously said it had worked with 20 customers, including several advanced AI labs. Mecka AI, Scale AI and Micro1 also operate in a similar field.
Why it matters
The importance of this development lies in the fact that training data, just like physical hardware, is emerging as a central bottleneck in the advancement of robotic systems. XDOF’s combination of the data it collects through teleoperation and sensors with tools and labeling infrastructure focuses on the data-generation problem that robotics companies and advanced AI labs struggle to solve on their own. The ABC collection planned in collaboration with the Berkeley AI Research lab is one of the areas to watch in assessing how this approach translates into research and product development processes; however, the collection has not yet been released. The presence of companies such as Mecka AI, Scale AI and Micro1 shows that competition is also taking shape around this data infrastructure. The terms of the new round and whether they include the disclosed valuation remain unclear.
Background
Series is not a new name in the FikirPilot archive: we have published 3 news reports mentioning the name in the last 90 days; the latest was dated September 4, 2026.