Huafu Securities believes that advances in general-purpose models and toolchains could help embodied AI reach a critical milestone, with data collection and robot components two areas worth watching.
Tianfeng Securities’ machinery team maps opportunities for deploying physical AI to specific manual tasks: apparel focuses on handling flexible fabric pieces; packaging and logistics cover loading, unloading, sorting and handling; welding highlights control systems and workstations. To understand the size of the opportunity, the corresponding worker pools must be considered alongside product coverage and development stage.
CICC Machinery divides physical AI data-collection initiatives into perception, motion capture, simulation, and scenario-based closed loops. The key to understanding these companies is to distinguish between collection products already launched, existing orders, and business directions extending into embodied intelligence, while clarifying the relationship between overall earnings and specific business lines.
The material identifies high-quality training data as an industrial bottleneck for physical AI, with opportunities concentrated in collection hardware, real-world data and simulation services. Estimates point to steep growth in market size, but companies are at markedly different stages of realization, from having made sales or accumulated data to having disclosed no revenue yet.
The growth drivers of four industries are diverging: physical AI depends on collecting real-world data, optical connectors already have earnings and orders to support them, refrigerants rely on quota constraints to stabilize prices, and overseas auto expansion is shifting from an export rush to local production.
