The robotics data startup is reportedly back in the market just months after its Series A, as investors chase the bottleneck behind physical AI.
XDOF is already back in fundraising mode. Less than three months after coming out of stealth, the robotics data startup is in late-stage talks to raise a Series B at a valuation of about $1.2 billion, according to people familiar with the deal. 8VC is said to be leading the round.
A fast return to market
The timing is unusual even by venture standards. XDOF raised a $70 million Series A in June, with participation from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. The company was not planning to raise again so soon, the people said, but its growth changed the conversation. Annualized revenue is approaching $50 million, and investors began approaching the company about another round.
The terms are not final and could still change. TechCrunch was unable to learn the total amount being raised or whether the $1.2 billion figure includes the new capital. XDOF and 8VC did not respond to requests for comment.
Why investors are leaning in
XDOF was founded in 2024 by UC Berkeley researchers Philipp Wu, now CEO, and Fred Shentu, CTO. The company is built around a problem that has become central to robotics: there is no internet-scale dataset for physical machines to learn from. XDOF is trying to fill that gap by building the data pipelines, collection tools, and annotation systems that frontier AI labs and robotics companies would otherwise have to create themselves.
That pitch has made XDOF look, to some investors, like a Scale AI or Mercor for robotics. The comparison is not about the product category alone. It is about the role. In large language models, the internet supplied the raw material. In robotics, the raw material has to be gathered in the real world, task by task, movement by movement.
From GELLO to ABC
Wu’s work on the company traces back to his research as a PhD student, when he was studying how robots learn from large datasets and ran into the same obstacle XDOF is now trying to solve: a lack of large-scale data. He and Shentu built GELLO, a low-cost teleoperation system that lets a human operator control a robotic arm remotely to generate training data. That work led to an influential robotics paper and became the foundation for XDOF.
The startup is now partnering with UC Berkeley’s AI Research lab on what it says will be the largest collection of high-quality robot training data ever assembled, called ABC. To capture it, XDOF combines remote robot teleoperation with human collectors wearing sensors to record everyday tasks such as folding clothes and flattening boxes.
The company also plans to hire and train data-collection teams around the world, including teleoperators who steer robots remotely and egocentric operators who wear body sensors to capture movement data.
The race for real-world robot data
XDOF said previously that it was already working with 20 customers, including several frontier AI labs. It is not alone in chasing this market. Mecka AI is also trying to collect real-world data for robot training, while companies such as Scale AI and Micro1 are pushing human-data platforms beyond LLMs.
For now, the bigger story is how quickly XDOF has moved from stealth to a possible nine-figure valuation jump. In a market where physical AI still lacks the training data that powered the last wave of model building, that shortage is becoming its own asset.
