Technology· Startups

XDOF, just three months out of stealth, is in talks for a Series B at a $1.2B valuation

XDOF, a robotics data startup founded by UC Berkeley researchers, is pursuing a Series B funding round at approximately $1.2 billion valuation led by 8VC, less than three months after exiting stealth mode. The company's rapid trajectory toward $50 million in annualized revenue has attracted investor interest for a follow-on round after raising $70 million in Series A funding in June.

By AI NewsroomPublished about 19 hours agoUpdated about 19 hours ago2 views

XDOF, just three months out of stealth, is in talks for a Series B at a $1.2B valuation

Why It Matters

XDOF's accelerated fundraising timeline reflects intense investor competition in the robotics AI sector, where high-quality training data represents a critical constraint for developing general-purpose robotic systems. The startup's valuation and growth trajectory suggest venture capitalists view data infrastructure for physical robotics as a foundational layer comparable to earlier data-labeling platforms that supported the broader AI boom.

Key Facts

Time to Series B: Less than three months after emerging from stealth
Series B valuation: Approximately $1.2 billion
Series A funding: $70 million raised in June 2024
Annualized revenue: Approaching $50 million
Current customers: Already working with 20 customers including several frontier AI labs

XDOF has rapidly become one of the most sought-after startups in the robotics data space. Co-founded in 2024 by UC Berkeley PhD researchers Philipp Wu and Fred Shentu, the company emerged from stealth with a clear mission: building the data infrastructure that AI labs and robotics companies cannot efficiently construct themselves. Their approach involves combining teleoperation technology—where human operators remotely control robotic arms—with egocentric sensing systems that capture everyday human movements for model training.

The company's technical foundation traces back to Wu and Shentu's academic research on GELLO, a low-cost teleoperation system that generated an influential robotics paper. This work demonstrated the acute scarcity of large-scale, high-quality training data for physical robots, a bottleneck that XDOF directly addresses. Unlike large language models that could initially train on internet-scale text, general-purpose robots lack an equivalent real-world dataset, making organized data collection a critical prerequisite for progress.

XDOF's growth metrics have evidently caught the attention of venture investors. The company raised $70 million in its Series A round in June from investors including Andreessen Horowitz, Thrive Capital, Lux, and Spark Capital. Rather than following the typical multi-year funding cycle, the startup has attracted follow-on interest from 8VC within months, reportedly driven by annualized revenue approaching $50 million. The company's Series B terms remain under negotiation and could still shift before finalization.

The startup's go-to-market strategy involves recruiting and training global teams of data collectors and remote teleoperators while partnering with UC Berkeley's AI Research lab to release a large-scale dataset called ABC. XDOF currently serves roughly 20 customers spanning multiple frontier AI laboratories. Investors have drawn comparisons to Scale AI and Mercor, the data-infrastructure companies that became essential to the previous wave of AI development, suggesting similar conviction that XDOF will prove foundational to robotics advancement.

Competition in robot data collection is intensifying, with startups including Mecka AI and data platforms like Scale AI and Micro1 also pursuing similar opportunities. However, XDOF's early customer traction and the combination of teleoperation and human sensor technology have positioned it at the forefront of the emerging sector.