Less than three months after completing its Series A financing, XDOF has been reported to be entering into new rounds of financing negotiations. According to insiders, this startup, which collects real-world teleoperation data for general-purpose robots, is in talks to raise about $1.2 billion in Series B financing, with 8VC leading the investment.

This company was co-founded in 2024 by researchers from the University of California, Berkeley, Philipp Wu and Fred Shentu. TechCrunch previously disclosed that XDOF completed a $70 million Series A financing in June, with investors including Thrive Capital, Andreessen Horowitz, Lux, and Spark Capital.

Revenue growth drives a new round of negotiations

Insiders say that XDOF was not originally planning to refinance so soon, but the company has been growing rapidly, with annual revenue approaching 50 million US dollars. It is this performance that has led several venture capital firms to approach the company proactively, initiating a new round of negotiations.

However, the transaction terms have not yet been finalized, and it is currently impossible for the outside world to confirm whether the total financing amount and valuation include additional funds. Neither XDOF nor the 8VCs have responded to requests for comments.

Robot Data Supply Chain

The business focus of XDOF is to build data pipelines, collection tools, and annotation systems for cutting-edge AI laboratories and robotics companies. It aims to transform what was originally a difficult process of building these data systems in-house into an outsourced “data supply chain.”

Co-founder Wu previously discovered in research that one of the biggest obstacles to robot training is the lack of real data on a sufficient scale. From this, XDOF developed the low-cost remote operation system GELLO, which allows humans to remotely control robotic arms to generate training data.

Expand the collection network

XDOF is still collaborating with the University of California, Berkeley AI Research and lab, preparing to release a robot training dataset they call ABC. The company states that this could be the largest high-quality robot training dataset to date.

To collect this data, XDOF has combined remote teleoperation with wearable sensors for data collection. The company plans to recruit and train data collection teams globally, covering both remote operators and personnel responsible for collecting wearable motion data. Previously, XDOF stated that it had collaborated with 20 clients, including several leading AI laboratories.