The XDOF Series B valuation has reached roughly $1.2 billion in late-stage talks, with venture firm 8VC set to lead the round, according to several people with knowledge of the deal. The terms are not yet final.
The speed here is the story. XDOF only emerged from stealth three months ago, closed its Tracxn-tracked Series A on 18 June 2026, and wasn’t planning another round. Then the revenue started coming in, and the VCs started calling.
Annualised revenue is approaching $50 million, the people said. That number, not any particular product launch, is what pulled investors back to the table so fast.
From a PhD Project to a $1.2bn XDOF Series B Valuation
XDOF was founded in 2024 by two UC Berkeley researchers: Philipp Wu, who serves as CEO, and Yide ‘Fred’ Shentu, the CTO. According to Shentu’s academic site, he is a machine learning and robotics PhD student advised by Professor Pieter Abbeel, and was previously a founding engineer at Covariant AI.
The company grew out of a research paper. Wu, Shentu, and three UC Berkeley colleagues published the GELLO paper in September 2023, describing a low-cost teleoperation system that lets a human operator control a robotic arm remotely to generate training data. The GELLO paper on arXiv has since accumulated 386 citations, according to Semantic Scholar. Not bad for what started as a workaround to a data shortage.
That shortage is exactly the problem XDOF is now selling solutions to. Physical robots don’t have an equivalent of the internet to train on. Unlike large language models, which gorged on text already sitting online, general-purpose robots need real-world movement data collected by actual humans doing actual tasks: folding clothes, flattening boxes, reaching across tables. XDOF builds the pipelines, collection tools, and annotation systems to capture and organise all of it, essentially acting as an outsourced data-supply chain for the robotics industry. Investors describe it as the Scale AI or Mercor for physical robotics.
The $70 million Series A brought in Thrive Capital, Andreessen Horowitz, Lux Capital, and Spark Capital. SiliconANGLE reports that WndrCo also participated, a detail not confirmed by the original funding announcement. Alongside the round, XDOF released ABC-130K, which SiliconANGLE describes as the world’s largest open-source bimanual robot manipulation dataset, developed in partnership with UC Berkeley’s AI Research lab. Cooley LLP advised XDOF on that round, with partners Stephane Levy and Michael Bruno leading the team.
8VC’s Firepower and What the Deal Does Not Yet Tell Us
The reported Series B lead, 8VC, has been deploying at pace. The firm closed Fund VI in March 2025 with $998 million in new LP capital. According to an SEC EDGAR Form D filing dated 29 June 2026, as tracked by Value Add VC, 8VC also announced Fund VII with a target of $1.5 billion. The firm, then, has plenty of capital to put to work, which makes a billion-dollar-plus robotics data bet plausible on its face.
What we don’t know is the total capital being raised in this Series B round, or whether the $1.2 billion valuation is pre- or post-money. The people familiar with the deal could not confirm either point, and XDOF and 8VC declined to comment.
XDOF currently counts 20 customers, including several frontier AI labs. Its collection model combines remote teleoperation, where operators steer robots from a distance, with egocentric data capture, where human collectors wear body sensors to record physical movement. The startup plans to expand those teams globally.
For the robotics data sector, the valuation signals that investors are willing to pay a premium for early leadership in what could become a very large infrastructure layer. The more interesting number to watch is how that $50 million annualised revenue figure moves once the Series B closes and the hiring begins in earnest.
