XDOF Nears $1.2 Billion Valuation in Series B Funding Talks
Less than three months after emerging from stealth mode, XDOF, a startup specializing in collecting real-world teleoperation data to train general-purpose robots, is reportedly in late-stage discussions to raise a Series B round. Sources familiar with the deal indicate the funding round is expected to value the company at approximately $1.2 billion, led by prominent venture capital firm 8VC.
Founded in 2024 by UC Berkeley researchers Philipp Wu (CEO) and Fred Shentu (CTO), XDOF has quickly attracted substantial investor interest. Earlier this year, TechCrunch reported on the startup’s $70 million Series A round, which featured participation from Thrive Capital, Andreessen Horowitz, Lux Capital, and Spark Capital. Despite not initially planning to raise capital so soon after that round, XDOF’s rapid growth—with annualized revenue approaching $50 million—has drawn renewed attention from venture capitalists eager to support its expansion.
While the exact amount being raised in the Series B round remains undisclosed, and the terms are still subject to change, the valuation signals strong confidence in XDOF’s business model and market potential. Both XDOF and 8VC have declined to comment on the ongoing negotiations.
Building the Backbone for Robotics Data Supply Chains
XDOF aims to solve a critical bottleneck in robotics development: the scarcity of large-scale, high-quality real-world data. Unlike large language models that benefit from vast datasets harvested from the internet, physical robots require meticulously collected and annotated data reflecting real-world tasks to improve their learning capabilities. XDOF positions itself as an outsourced data-supply chain for the robotics industry, providing the infrastructure, tools, and annotation systems that leading AI labs and robotics companies find difficult to develop internally.
Philipp Wu’s academic background underscores this mission. During his PhD studies, Wu encountered significant obstacles due to the lack of large datasets for robot learning. This challenge led him and Shentu to develop GELLO, a cost-effective teleoperation system enabling human operators to remotely control robotic arms, generating valuable training data. Their research culminated in a well-regarded robotics paper, laying the groundwork for what would become XDOF.
Investors now compare XDOF to data-labeling powerhouses like Scale AI and Mercor, which fueled the recent AI boom by supplying crucial data infrastructure. However, XDOF’s focus is on physical robotics—a sector where comparable datasets have yet to be amassed, making their work foundational for the future of general-purpose robots.
Collaborations and Data Collection Initiatives
To further its mission, XDOF is partnering with UC Berkeley’s AI Research lab to release ABC, which it claims to be the largest collection of high-quality robot training data ever assembled. The startup employs innovative data capture methods, combining remote teleoperation of robots with human collectors who wear sensors as they perform everyday tasks such as folding clothes and flattening boxes. This dual approach allows the company to gather rich, nuanced datasets critical for advancing robotic learning.
Looking ahead, XDOF plans to expand its global workforce by training teams of data collectors, including teleoperators who steer robots remotely and operators equipped with egocentric sensors to capture detailed movement data. The startup is already collaborating with around 20 customers, including several leading AI research labs, demonstrating early market traction.
The landscape of real-world data collection for robot training is becoming more competitive, with other startups like Mecka AI and human-data platforms such as Scale AI and Micro1 pushing the boundaries beyond large language models.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
For further details, read more Here.
