XDOF Startup Nears Billion-Dollar Valuation Within Months

In a scene that reflects the growing enthusiasm toward the AI-powered robotics sector, a startup called "XDOF" has managed to capture investors' attention within a short period of no more than three months since going public. According to informed sources, the company is engaged in advanced negotiations to close a Series B funding round that would value it at nearly $1.2 billion, led by "8VC." This rapid growth raises important questions about the true value of data in the world of robotics.

From a Berkeley Lab to a Billion-Dollar Company

XDOF was founded in 2024 by two researchers from the University of California, Berkeley: Philipp Wu, who serves as CEO, and Fred Shinto, who holds the position of CTO. The company had announced last June a Series A funding round worth $70 million, in which several prominent names participated, such as "Thrive Capital," "Andreessen Horowitz," "Lux," and "Spark Capital."

According to sources, the company did not intend to raise new funding this quickly, but its remarkable growth — which pushed its estimated annual revenue to around $50 million — prompted venture capitalists to approach it with an offer for a new investment round. The total amount of capital planned to be raised has not yet become clear, and the terms of the deal have not been finalized and may be subject to change.

Why Is Data the Biggest Obstacle for Robots?

The core of XDOF's work lies in addressing one of the most complex challenges facing the robotics industry: the scarcity of high-quality, real-world data needed to train machines to perform tasks. Unlike large language models that were initially trained on the vast content available across the internet, physical robots do not have a comparable real-world database that can be leveraged.

The roots of the idea go back to Wu's PhD studies, when he faced great difficulty finding massive data suitable for his research on how robots learn from large datasets. From this was born his collaboration with Shinto on a project called "GELLO," a low-cost remote control system that allows a human operator to guide a robotic arm to generate training data, which resulted in influential research in the field of robotics that later became the foundation upon which the company was built.

A Data Supply Chain for the Industry

Investors describe XDOF as the "Scale AI" or "Mercor" of the world of physical robotics, in reference to the data-processing giants who helped drive the AI boom. The company aims to build the data pipelines, collection tools, and labeling systems that advanced AI labs and robotics companies find difficult to create themselves, effectively operating as an external data supply chain for the sector.

To collect this data, XDOF adopts a dual approach that combines:

• Remote control of robots (teleoperation) by specialized operators.
• Human data collectors wearing sensors to record daily tasks such as folding clothes and flattening cardboard boxes.

The company plans to hire and train teams of data collectors around the world, including remote operators and others wearing body sensors to capture motion data from a first-person perspective.

Broad Partnerships and Ambitions

XDOF is collaborating with the AI research lab at Berkeley to release what it considers the largest high-quality robot training dataset ever assembled, under the name "ABC." The company had also previously revealed that it is already working with 20 customers, including a number of advanced AI labs.

The company enters a race in which it competes with other players seeking to collect real-world data to train robots, such as "Mecha AI," in addition to human data platforms that have begun expanding beyond the field of language models, such as "Scale AI" and "Micro1." It appears that the bet on physical data will remain one of the most prominent areas of competition in the coming years.

✦ بقلم فريق دروب أيديا