XDOF Raises $70M to Build the Data Layer Behind Physical AI

California-based robotics infrastructure company XDOF has raised $70M from Thrive Capital, Spark Capital, Andreessen Horowitz, Lux Capital, and WndrCo to expand the data infrastructure powering the next generation of Physical AI systems. XDOF builds the collection pipelines, teleoperation systems, annotation workflows, and operational tooling used to train robot foundation models. As robotics companies race to improve machine intelligence in the physical world, the company is focused on a challenge that receives far less attention than the robots themselves: generating high-quality training data at scale.

The founding team includes Philipp Wu (CEO), Yide "Fred" Shentu (CTO), and Nemo Jin (COO), whose backgrounds span the UC Berkeley Berkeley Artificial Intelligence Research (BAIR) Lab, Covariant, and Tesla. XDOF reports approximately 20 customers and roughly 60 employees as it emerges from stealth. The funding highlights a broader shift across artificial intelligence. As Physical AI and robotics move from research projects into commercial deployment, infrastructure providers are becoming increasingly important pieces of the technology stack.

What Happened

XDOF announced a $70M funding round backed by Thrive Capital, Spark Capital, Andreessen Horowitz, Lux Capital, and WndrCo. The investor syndicate is notable because it spans firms that have historically backed foundational technology platforms and infrastructure businesses. That matters because XDOF is not building robots for consumers. It is building the operational layer that helps robotics companies train them.

While much of the market focuses on increasingly capable machines and model breakthroughs, Co-Founder & CEO Philipp Wu, Co-Founder & CTO Yide Fred Shentu, and Co-Founder & COO Nemo Jin have been building the infrastructure required to generate, manage, and operationalize robotics training data. Technology history has a habit of rewarding the companies that make entire ecosystems possible. Infrastructure rarely receives the spotlight, yet it frequently becomes one of the most valuable parts of the stack.

Why This Matters

Physical AI refers to artificial intelligence systems capable of interacting with the physical world through robots, machines, and autonomous systems. Unlike large language models that learn from vast amounts of internet text, robots learn through observation, repetition, and interaction. They require enormous amounts of real-world data to understand movement, manipulation, force, environments, and task execution.

That means collecting data, cleaning data, annotating data, evaluating outcomes, and repeating the process thousands of times. XDOF is attempting to industrialize that workflow. The company has built a full-stack data infrastructure platform that helps robotics companies and frontier AI labs collect, organize, annotate, and operationalize the data required to train robot foundation models. Sophisticated robotics systems are only as capable as the datasets supporting them, and for many organizations, building internal data operations from scratch can become as difficult as building the models themselves.

Market Context

Artificial intelligence is entering a new phase. The first wave focused on software-based intelligence. The next phase increasingly centers on machines interacting with the physical world. That transition is creating demand for entirely new categories of infrastructure.

Physical AI, robotics, autonomous systems, and embodied intelligence all require high-quality real-world training data. The challenge is not simply generating intelligence. The challenge is creating repeatable systems capable of producing the data that intelligence depends on. The funding environment reflects this shift, with venture capital firms increasingly evaluating not only robotics manufacturers and AI model developers but also the infrastructure providers supporting them. The XDOF funding round arrives as investment activity across robotics, embodied AI, industrial automation, and AI infrastructure continues to accelerate.

Competitive Landscape

XDOF operates within a growing segment of the robotics ecosystem: robotics data infrastructure. The company recently released the ABC Dataset, which contains more than 130,000 robot trajectories and over 3,500 hours of real-world interaction data. The dataset serves as a training resource for Physical AI systems learning real-world manipulation tasks.

The release demonstrates both scale and intent. Many companies discuss the importance of robotics data. Far fewer are building operational systems capable of generating it consistently. XDOF is positioning itself as the infrastructure partner for organizations that need large-scale robotics datasets without building the underlying machinery themselves. That creates a different competitive profile than traditional robotics startups because the company sits underneath the ecosystem rather than competing directly within it.

What This Signals

The most important signal from the XDOF funding announcement is not the size of the round. It is what investors appear to believe comes next. The market is beginning to recognize that Physical AI requires a different support structure than software-based AI. Models matter. Hardware matters. But neither improves without reliable streams of high-quality data.

That reality creates opportunities for companies operating below the surface of public attention. Data pipelines, annotation systems, teleoperation networks, and operational workflows rarely generate headlines, yet those capabilities often determine which technologies successfully scale beyond controlled environments. The infrastructure layer is becoming a category of its own.

The Bigger Industry Shift

Physical AI is moving from experimentation toward deployment. Organizations can no longer rely solely on academic datasets, research demonstrations, or small-scale testing environments. They need systems capable of generating training data continuously and at industrial scale.

This is where XDOF fits into the broader AI ecosystem. The company is positioning itself as a foundational layer between robotics builders and the data required to improve robot performance. As Physical AI adoption expands, infrastructure providers that can create, manage, and operationalize training data may become increasingly strategic assets. Whether the future belongs to humanoid robotics, industrial automation, logistics systems, or entirely new categories of autonomous machines, one reality remains unchanged: data is becoming a critical resource. XDOF is betting that the next phase of robotics growth will be driven not just by better models, but by better infrastructure. Investors appear to agree.