XDOF
"Defining motion for autonomous systems."
US robotics data infrastructure startup founded Oct 2024; raised $70M to deliver production-scale teleoperation datasets for robot AI training.
XDOF is a robotics infrastructure company founded in October 2024 by Philipp Wu (CEO), Fred Shentu, and Nemo Jin (COO). The company emerged from stealth in June 2026 with $70 million in funding from Thrive Capital, Andreessen Horowitz (a16z), Spark Capital, Lux Capital, and WndrCo. XDOF addresses what it identifies as the primary bottleneck in robotics development: the availability of high-quality, production-scale training data for robot foundation models.
The company's core offerings span three tiers. At the highest-value tier, XDOF collects teleoperation data directly on the target robot being deployed, using human operators to demonstrate manipulation tasks. The second tier involves teleoperated robots gathering more generalizable data, including the GELLO low-cost teleoperation system co-authored by XDOF's founders. The third tier covers egocentric data collected from humans performing everyday tasks using custom wearable sensors the company plans to develop. XDOF delivers production-scale datasets covering complex bimanual manipulation tasks including pick-and-place, folding, insertion, tool use, and assembly, alongside policy training support, robot evaluation services, and full-stack tooling encompassing hardware design, data operations, and model development workflows.
In June 2026 XDOF released ABC-130K, described as the largest open-source teleoperation dataset at the time of release, comprising more than 130,000 demonstrations across 195 bimanual manipulation tasks. The dataset was developed in collaboration with researchers from UC Berkeley, Carnegie Mellon University, MIT, and Amazon. Demonstrated tasks include folding T-shirts, unfolding boxes, and loading AirPods cases. As of June 2026 XDOF had approximately 60 employees and was working with 20 customers, including several frontier AI laboratories. The company's mission is to unlock a world with abundant, useful physical intelligence by building world-class infrastructure for the most ambitious robotics builders.
Primary type & automation activities this supplier delivers:
ABC-130K Teleoperation Dataset
- Model: ABC-130K
- Launch Year: 2026
Largest open-source teleoperation dataset with 130,000+ bimanual manipulation demonstrations across 195 tasks, developed with UC Berkeley, CMU, MIT, and Amazon.
Company Facts
Founded
2024
Primary Role
Software/Algorithm
Company Size
-
Primary Region
North America
Annual Sales
-
Funding Stage
Early-Stage VC
Funding Total
-
Keywords
- XDOF
- robotics data
- teleoperation
- AI infrastructure
- open-source dataset
- robot training data
- teleoperation dataset
- bimanual manipulation
- ABC-130K
- GELLO
- embodied AI infrastructure
- robot foundation model
- Software/Algorithm
- Component Supplier
- Control & Software
- Cloud & Data
- Manufacturing
- Logistics & Supply Chain
- Education & Research
ROBOTTODAY Weekly, June 15 – 19, 2026
This week in robotics: Japan commits $65B to physical AI, UBTECH’s U1 humanoid companion robot nears 5,000 pre-orders, XDOF raises $70M for robot foundation model infrastructure, and Waymo recalls ~3,900 robotaxis after construction-zone failures. June 15–19, 2026.
Robotics Infrastructure Startup XDOF Emerges from Stealth with $70M in Funding
XDOF has officially launched after securing $70 million in funding to create infrastructure for robot foundation models. The company aims to develop essential datasets, robotic systems, and software tools that will enable robotics firms and research institutions to enhance the capabilities of physical AI systems. This significant investment comes from prominent venture capital firms, including Thrive Capital, Andreessen Horowitz, Spark Capital, Lux, and WnderCo. The funding will support XDOF's mission to advance the field of robotics and artificial intelligence, addressing the growing demand for more sophisticated and efficient robotic solutions.
AI AI Funding & Investment Robotics Amazon Andreessen Horowitz Carnegie Mellon
Collecting robot training data is dirty, unglamorous work. Some AI labs are already paying XDOF to do it.
Recent discussions in the field of artificial intelligence highlight a significant challenge facing the development of physical AI systems. Experts emphasize that in order for physical AI to achieve milestones comparable to those of large language models (LLMs), a critical data issue must be addressed. As of October 2023, the existing datasets are insufficient to support the complex learning and operational needs of physical AI. This gap in data could hinder progress and innovation in creating AI that can effectively interact with and navigate the physical world. Addressing this problem is essential for advancing the capabilities of physical AI, ensuring that it can perform tasks with the same proficiency as its software counterparts.
AI Startups a16z robots Thrive Capital
