8VC is in advanced talks to lead a Series B for XDOF at a valuation near $1.2B, several people with knowledge of the matter told TechCrunch. Negotiations are ongoing and the shape of the round could still shift, the people cautioned. A deal would cap a swift ascent for the startup, which went public with its robot data operation less than three months ago.
Philipp Wu and Fred Shentu, both UC Berkeley researchers, started the company in 2024. Wu serves as chief executive and Shentu as chief technology officer. A $70M Series A closed in June with Thrive Capital, Andreessen Horowitz, Lux and Spark Capital on the cap table. The company had not planned another raise so quickly, but annualized revenue near $50M brought investors to its door instead.
The thesis behind XDOF is that physical robots face a data famine. Web-scale text powered large language models, but machines that fold laundry or pack boxes have no equivalent corpus, so every hour of human-guided operation becomes valuable raw material. The firm builds collection tools and annotation pipelines for that material, blending remote teleoperation rigs with sensor-wearing humans who record chores like folding clothes and flattening boxes.
UC Berkeley’s AI Research lab is a partner on the ABC dataset, which XDOF is compiling to widen the pool of usable demonstrations. Collector hiring is expanding worldwide as the project grows. XDOF says about 20 customers already draw on its pipelines, several of them frontier AI labs.
