The turning point for mass production of humanoid robots has been reached, but the “brain” and data gap are still bottlenecks (Nomura)

Street Signal | The turning point for mass production of humanoid robots has been reached, but the “brain” and data gap are still bottlenecks (Nomura)

A Nomura Securities report pointed out that China’s robotics industry will present three major changes at the 2026 World Artificial Intelligence Conference (WAIC): body scale, brain lag, and data bottlenecks.

Embodied intelligence has passed the inflection point of mass production, with nearly 60 humanoid robots performing real tasks, and the narrative focus has shifted to shipment and deployment.

However, the current demand base is mainly based on non-production scenarios (entertainment, consumption, government and education total about 95%), industrial applications are only 5%, and the economics of factory testing have not yet been verified.

The key bottleneck lies in the high-quality data gap (up to 20 times), which restricts the maturity of the "brain". One sentence conclusion: The "body" of humanoid robot industrialization takes the lead, but the "brain" and data are the core bottlenecks restricting its large-scale commercialization.

Investors need to be wary of the risk of capital formation ahead of real demand. Good/bad: Good for China’s humanoid robot industry chain, but hardware and software need to be distinguished. The current market may be over-hyping the concept of mass production and ignoring the long-term challenges of software and data. Catalysts:

1) Repeat orders from unrelated industrial customers;

2) Verification of the economic payback period of a single humanoid robot;

3) Acquisition of high-quality training data and progress in model iteration.