Abstract: Restricted by the construction space and navigational requirements, developing floating wind turbines is an urgent and practical need for developing offshore wind energy resources. Currently, the primary industrial reason for limiting the development of floating wind turbines is the high cost of construction. The basic premise for solving this problem is to accurately simulate the motions of wind turbines in complex environments by coupling aero-hydrodynamic responses, thereby supporting the integrated optimal design of wind turbines. The hybrid testing method combines physical test simulation (experimental substructure) and numerical simulation (numerical substructure), allowing for the full utilization of their respective advantages and simultaneously simulating a complete structural response. Thus, this approach has been extensively developed in civil engineering, as well as in the automotive, aerospace, and other fields. This study summarizes the advantages of hybrid tests in multi-load coupling analysis of wind turbine structures: Hybrid testing can not only divide the structure into experimental and numerical substructures at the structural level but also separate the aerodynamic simulation and hydrodynamic simulation at the load level. Meanwhile, the aerodynamic and hydrodynamic effects can be coupled through the substructural boundaries to achieve a complete structural and full-coupling analysis of floating wind turbines. This paper summarizes a few exploratory studies available on hybrid tests of floating wind turbines, identifying the key scientific issues in floating wind turbine hybrid tests in conjunction with the development of hybrid testing methods. Specifically, they include: The “real-time” problem: the contradiction between the high-efficiency and high-precision calculations of numerical substructure, and the "real-time" loading of the experimental substructure must meet the requirements of the time-domain continuous loading of wave/wind; The substructure coupling problem: the design of truncation of the physical substructure and the accurate extrapolation to its full-scale full-depth structural responses. In summary, the offline hybrid testing method, combined with machine learning, can better address the above problems, which warrants further attention and in-depth research.