“We Don’t Build Robots — We Build the Data That Makes Them Smart”
Noitom Robotics, a Beijing-based data infrastructure company for the embodied AI industry, has closed a Pre-A+ financing round led by Qiming Venture Partners, with participation from 5Y Capital, Legend Capital, and follow-on investments from Matrix Partners China and InnoAngel Fund. The round was oversubscribed.
Combined with its earlier Pre-A round (led by Alpha Startups), Noitom Robotics has now raised hundreds of millions of RMB (~$30–50M USD), cementing its position as a critical enabler in the global humanoid robotics supply chain.
Unlike traditional robotics firms, Noitom Robotics does not manufacture hardware.
It delivers high-quality, scalable, multi-modal training data — the fuel for Vision-Language-Action (VLA) models that power next-generation robots.
“We’re grateful to investors who back a ‘robotics company that doesn’t build robots.”
Strategic Positioning: Data as the New Bottleneck
As VLA models mature and the scaling law for embodied AI takes hold, the industry’s primary constraint is shifting from algorithms to real-world human behavior data.
Noitom Robotics addresses this by offering:
- Factory-grade data: Ultra-high-precision, multi-sensor captures of human tasks in controlled industrial environments
- In-the-wild data: Visually diverse, real-world recordings for environmental generalization
- Human-centric design: Focus on transferable behaviors, not robot-specific demos
The company leverages over a decade of motion capture expertise from its parent, Noitom Ltd., which holds 70% of the global professional motion capture market and has shipped tens of thousands of systems worldwide.
This heritage enables Noitom Robotics to:
- Capture complex human actions with sub-centimeter precision
- Structure and standardize raw behavior into model-ready datasets
- Map human motion to diverse robot morphologies (bipeds, quadrupeds, arms) via cross-body infrastructure
Building the “Data Factory” for Physical AI
The new capital will accelerate:
- Engineering of scalable data production pipelines
- Global “data factory” deployment for consistent, high-throughput output
- Core team expansion in AI, robotics, and data systems engineering
The goal is to create an end-to-end data delivery platform that allows robot developers to:
- Order task-specific datasets on demand
- Receive structured, labeled, and simulation-ready data within weeks
- Train models that generalize across robots and environments
This service is already live: Noitom Robotics supports dozens of humanoid and embodied AI companies globally, including leading Chinese and international OEMs and model developers.
Investment Takeaway:
While the world watches robot demos, the real race is for behavioral data at scale.
Noitom Robotics represents a rare profile:
- Proven technology (spin-out from a category-dominant motion capture leader)
- Clear product-market fit (paying customers in robotics and AI)
- Defensible infrastructure (data factories, cross-body mapping, standardized pipelines)
For investors seeking exposure to the humanoid boom without hardware risk, Noitom Robotics offers a pure-play on the data layer — the foundation upon which all general-purpose robots will be trained.
As Dr. Dai puts it:
“Robots don’t learn from code. They learn from watching us.”
And Noitom Robotics is building the world’s most advanced library of human motion — one task at a time.
