Accurate prediction of stress responses in offshore wind turbines is crucial for fatigue analysis and estimating remaining useful life, both vital for ensuring the resilience of these structures under demanding environmental conditions. The high costs of maintaining offshore wind turbines make predicting their remaining operational life critical for enhancing operational efficiency. However, a significant challenge lies in the fact that not all turbines in a wind farm are fully instrumented, limiting the collection of comprehensive data across the entire farm. Typically, only a few turbines are equipped with both Supervisory Control and Data Acquisition (SCADA) systems and dynamic monitoring equipment, while most rely solely on the former. This study addresses this issue by employing Gaussian Process (GP) regression models to predict bending moment responses at various heights along the turbine tower and foundation, using SCADA data from a fully instrumented turbine. The research utilizes a physics-informed kernel trained on SCADA measurements, such as wind speed, power output, pitch, yaw, and acceleration, collected from the tower's top. This kernel integrates key SCADA parameters to model the complex relationships between these measurements and the bending moment responses. Through transfer learning, the GP model is adapted to predict bending moments in turbines that are only partially instrumented, relying solely on available SCADA data. The study is conducted using two identical 6 MW offshore wind monopiles that are fully instrumented, allowing for the validation of the proposed strategy. Initial results show the GP model’s effectiveness in accurately predicting structural responses from SCADA data, providing a cost-effective solution for monitoring and maintaining wind turbines. This approach offers promising potential for improving the reliability and performance of wind energy systems by delivering detailed insights into turbine behavior without extensive instrumentation requirements.