XDOF is in late-stage talks to raise a Series B funding round at a valuation of approximately $1.2 billion, according to people with knowledge of the deal. The talks come less than three months after the startup emerged from stealth and secured a $70 million Series A round, driven by rapid revenue growth that is approaching $50 million annually.
Rapid Growth Drives New Funding Talks
According to people familiar with the matter, venture capital firm 8VC is leading the late-stage discussions for the new financing. XDOF wasn’t planning to raise capital so soon following its June Series A round, which included participation from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital, alongside WndrCo and other backers mentioned in supplemental reports. However, surging commercial traction pushed annualized revenue near $50 million, prompting investors to approach the company.
Both XDOF and 8VC didn’t respond to our request for comment. Sources note that the terms of the transaction are not final and remain subject to change, while the total amount of capital being raised and whether the $1.2 billion valuation incorporates the new funds have not been publicly disclosed.
Solving the Robotics Data Bottleneck
Founded in October 2024 by UC Berkeley researchers Philipp Wu (CEO), Yide (Fred) Shentu, and Nemo Jin, XDOF builds data pipelines, teleoperation tools, and annotation systems to supply training data for physical robots. As Wu observed during his PhD research, large-scale real-world datasets for physical robotics have historically lagged far behind the vast text libraries available for large language models. To tackle this, Wu and Shentu developed GELLO, an affordable teleoperation system allowing human operators to control robotic arms remotely to generate structured training data.
Did you know? Unlike software algorithms that can pull from the entire internet, physical robots require specialized real-world movement data, creating a massive supply-chain bottleneck that startups like XDOF aim to resolve.
Industry investors now liken XDOF to foundational data-labeling giants like Scale AI or Mercor. The company currently employs roughly 60 people and maintains about 20 customers, including several frontier AI labs and robotics enterprises. To scale its operations further, XDOF plans to hire and train global networks of data collectors, including remote teleoperators and egocentric operators wearing body sensors to record everyday physical actions like folding clothes.
Open Datasets and Industry Partnerships
Alongside its commercial software tooling and data infrastructure, XDOF is actively collaborating with academic institutions to release massive training corpuses. The startup has partnered with UC Berkeley’s AI Research lab and MIT to build and release what it describes as the largest bimanual robot manipulation dataset ever assembled, known as ABC or ABC-130K. According to supplemental coverage, this dataset spans 130,000 trajectories across 195 bimanual tasks, accompanied by simulation resources designed to accelerate embodied AI research.
Frequently Asked Questions
Who founded XDOF?
XDOF was co-founded in October 2024 by UC Berkeley researchers Philipp Wu (CEO), Yide (Fred) Shentu, and Nemo Jin.
What is XDOF’s valuation in the reported Series B talks?
According to sources with knowledge of the discussions, XDOF is negotiating a valuation of approximately $1.2 billion.
What does XDOF actually do?
The startup builds data pipelines, teleoperation tools, and annotation systems to supply high-quality physical training data for robotics firms and frontier AI laboratories.
Want to stay updated on the latest breakthroughs in robotics data infrastructure and AI funding? Subscribe to our newsletter or join the conversation in the comments below.
Worth a look