XDOF: the robot-data startup worth $1.2 billion three months out of stealth

By Carlos Montiel | Enterprise AI Specialist
Leer en español →
Published: 2026-09-07 | By: Carlos Montiel | Reading time: ~5 min

XDOF, a startup that collects real-world teleoperation data to train general-purpose robots, is in advanced talks for a Series B led by 8VC at a valuation of roughly $1.2 billion. The company left stealth mode just three months ago.

What XDOF actually sells: not a robot, but the raw material to train them

XDOF doesn't build robots or compete with physical-AI labs. It builds the data pipelines, collection tools, and annotation systems that frontier AI labs and robotics companies can't easily assemble on their own. In practice, it acts as an outsourced data supply chain for the entire robotics industry.

To capture that data, the company combines remote robot teleoperation with human collectors wearing body sensors to record everyday tasks — folding laundry, flattening boxes — and plans to hire and train collection teams worldwide: both teleoperators piloting robots remotely and "egocentric" operators capturing human movement with body-worn sensors.

The round: 20 customers, several of them frontier labs

XDOF already works with 20 customers, including several frontier AI labs, which explains why 8VC is willing to lead a billion-dollar Series B for a company that only left stealth in June 2026. The speed of the round — zero to $1.2 billion in a single quarter — reflects how much unmet demand there is for real physical-manipulation data, the input that now limits robotics progress more than compute or model availability does.

What it means for companies evaluating "physical AI"

XDOF confirms a pattern already seen with text and images: when a type of data is scarce and valuable for training models, specialized intermediaries emerge to collect and sell it as a service, rather than every lab collecting it in-house. For manufacturing or logistics companies in Latin America considering robotics automation pilots, this suggests the "we don't have data for our specific process" gap can be solved by hiring data collection as a service, instead of assuming that capability has to be built from scratch.

It's also a market signal: if frontier AI labs are willing to pay for third-party teleoperation data rather than generating it solely with their own robots, it's because the race for "physical AI" — general robots learning real-world tasks — is accelerating faster than language-model-centric media coverage suggests.

Carlos Montiel
Enterprise AI Solutions Architect
LLMs, Agents & Orchestration Specialist
guatemalia.com/#contacto · info@guatemalia.com

Need to implement AI in your company?

Carlos Montiel is an enterprise AI solutions architect. He implements LLMs, Agents, RAG and orchestrators for companies across Guatemala and Latin America. Reach out for a consultation.

Contact Carlos Montiel

info@guatemalia.com