XDOF Eyes $1.2 Billion Valuation in 8VC Led Series B Talks
XDOF is nearing a Series B led by 8VC at a $1.2 billion valuation as its robot training data business approaches $50 million in 2026 revenue.
Summary
As of September 4, 2026, XDOF is in late-stage talks for an 8VC led Series B at about a $1.2 billion valuation, less than three months after leaving stealth. The raise size and whether the valuation includes new capital remain unknown, terms could change, and neither XDOF nor 8VC commented. Founded in 2024 by UC Berkeley researchers Philipp Wu, CEO, and Fred Shentu, CTO, XDOF had not planned to raise again after its $70 million June Series A from Thrive Capital, Andreessen Horowitz, Lux and Spark Capital. Investors approached as annualized revenue neared $50 million. XDOF has 20 customers, including several frontier AI labs.
XDOF supplies data pipelines, collection tools and annotation systems for general-purpose robotics, addressing the lack of an internet-scale real-world training dataset. Its foundation is GELLO, Wu and Shentu’s low-cost system for remotely controlling robotic arms to generate data, which produced an influential robotics paper. XDOF combines robot teleoperation with sensor-wearing collectors recording tasks such as folding clothes and flattening boxes, and plans worldwide teams of teleoperators and egocentric operators. It is partnering with UC Berkeley’s AI Research lab on ABC, which it expects to be the largest collection of high-quality robot training data. Investors liken XDOF to Scale AI or Mercor for physical robotics; competitors include Mecka AI, Scale AI and Micro1.
Positives
- Annualized revenue approaching $50 million prompted investors to pursue XDOF shortly after its previous financing.
- Twenty customers, including several frontier AI labs, already use XDOF’s robotics data services.
- ABC is intended to become the largest assembled collection of high-quality robot training data.
- The $70 million June Series A drew Thrive Capital, Andreessen Horowitz, Lux and Spark Capital.
- GELLO offers a low-cost way for human operators to generate robotic-arm training data remotely.
Risks & concerns
- The Series B amount and whether the $1.2 billion valuation includes new capital remain unknown.
- Late-stage negotiations are unfinished, leaving the valuation and other terms subject to change.
- Physical robotics lacks an internet-scale real-world dataset, creating a critical obstacle for general-purpose machines.
- Mecka AI, Scale AI and Micro1 are pursuing the same expanding market for real-world robotics data.
- XDOF and 8VC have not commented on the proposed financing.