Robots & Robotics

Humyn Labs Secures $47M to Build Robot-Training Data Infrastructure

Humyn Labs has secured $26 million in equity and $21 million in venture debt to expand real-world data capture and preparation for robotics and physical AI.

By Kseniia Klichova | Edited by Daniel Krauss Published:
Humyn Labs Secures $47M to Build Robot-Training Data Infrastructure
Humyn Labs is expanding the real-world data infrastructure used to train robots after securing $47 million in equity and venture debt. Photo: Humyn Labs

Key Notes

  • Humyn Labs announced $47 million in financing, comprising a $26 million equity round led by Accel and $21 million in venture debt led by Lighthouse Canton.
  • The company records real-world work and prepares multimodal datasets for AI labs and robotics developers, with capture and enrichment operations in more than 20 countries.
  • Funding will expand deployment-readiness infrastructure in Brazil, South Korea and India, alongside capture hardware, data processing and international teams.

Humyn Labs has secured $47 million to expand the collection and preparation of real-world training data for robots and physical AI. The financing backs a business built around recording people doing work, then converting those observations into signals that model developers can use.

The November 19 funding announcement separates the total into a $26 million equity round led by Accel and $21 million in venture debt led by Lighthouse Canton. Prosus Ventures participated in the equity financing, while Trifecta Capital joined the debt component.

Human Work Becomes Structured Training Data

Founded in 2026 by Manish Agarwal and Ishank Gupta, Humyn records activity in factories, warehouses, farms and homes. The company says its proprietary capture devices collect jobs to laboratory requirements, with an enrichment pipeline converting recordings into labelled datasets across vision, sound, motion and touch.

Its platform description shows the intermediate work behind that proposition: tracking hands, estimating depth, identifying objects and separating a task into actions such as reaching, grasping and placing. Those labels give a training system more structure than an unannotated video alone.

The engineering challenge is to preserve the relationship between what a person sees, how they move and when contact occurs. A useful recording needs to make those signals comparable over time. It also needs clear provenance so a developer can trace the resulting training sample back to its source.

Expansion Targets Capture and Deployment Infrastructure

Humyn says the capital will support deployment-readiness infrastructure in Brazil, South Korea and India, alongside capture hardware and enrichment engines. It also plans to strengthen its presence in San Francisco, London and Seoul. The company reports capture and enrichment operations in more than 20 countries.

The announcement says annualized revenue has exceeded $50 million within six months of operations. That is a company-reported run-rate figure, rather than a statement that $50 million has already been booked over a full year. The release does not provide a valuation.

Humyn’s data workflow includes validation, processing, labelling and human review. It describes synchronizing input streams and filtering unsuitable captures before exporting training formats. Such checks address the usability of a dataset; the model trained on it still needs evaluation on the intended robot and task.

Robot Learning Creates Demand Beyond the Robot Itself

The financing adds to investment in the infrastructure surrounding robotics. Our Mosaico report examined tools for storing and retrieving robot sensor recordings. Humyn focuses further upstream, creating and enriching records of human activity.

Our coverage of DexUMI demonstrations shows another route for transferring human manipulation into robot-learning data. The approaches use different collection hardware, but both depend on a useful mapping between the demonstration and the machine. Humyn’s expansion will be judged by how consistently its data supports that transfer across customers’ real tasks.

Disclaimer: RobotsBeat is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, Robotics, technology, software, and digital innovation sectors. These relationships do not influence RobotsBeat's editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.

Artificial Intelligence (AI), Business & Markets, News, Robots & Robotics, Startups & Venture