XDOF, a startup co-founded by UC Berkeley researchers Philipp Wu and Fred Shentu in 2024, has emerged from stealth mode and is now in advanced discussions to secure a Series B funding round led by 8VC, potentially valuing the company at approximately $1.2 billion, according to multiple sources familiar with the negotiations. This development comes less than three months after the company’s initial public appearance and follows a $70 million Series A round in June, which attracted investments from Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital. The accelerated timeline for the new funding round is attributed to XDOF’s rapid growth, with annualized revenue nearing $50 million, prompting venture capitalists to initiate discussions despite the company’s recent capital raise.

XDOF’s business model centers on providing data pipelines, collection tools, and annotation systems to frontier AI laboratories and robotics firms, effectively serving as an outsourced data supply chain for the robotics industry. The company’s foundation is rooted in Wu and Shentu’s prior research, including the development of GELLO, a low-cost teleoperation system designed to generate training data for robots. Leveraging this technology, XDOF collaborates with UC Berkeley’s AI Research lab to compile what it claims is the largest collection of high-quality robot training data, known as ABC. This dataset is amassed through a combination of remote robot teleoperation and human data collectors equipped with sensors to record everyday tasks, such as folding clothes and flattening boxes. XDOF intends to expand its global workforce of data collectors, including teleoperators and egocentric operators, to further enhance its data offerings. The company currently boasts a client base of 20 customers, including several frontier AI laboratories, positioning itself alongside competitors like Mecka AI and established data platforms such as Scale AI and Micro1 in the burgeoning market for real-world robot training data. Source

This is a summary. Read the original at techcrunch.com.