This paper introduces a novel nonlinear adaptive blade pitch control scheme for floating wind turbines (FWTs) with a semi-submersible platform. Initially, a control-oriented physical model of the FWT is developed using the platform’s quasi-coordinates and rotor speed. Presented in matrix form, this model captures the key physical phenomena governing turbine-platform dynamics, achieving a balance between model accuracy and complexity. Subsequently, an adaptive pitch controller is implemented to maintain rotor speed at its rated level while mitigating platform pitch motion. To address uncertainties in the FWT dynamics, an online learning approximator is employed for real-time estimation. Theoretical analysis guarantees rotor speed tracking and ensures bounded signals throughout the closed-loop system. Finally, the established FWT model and pitch control scheme are verified through simulations carried out using the FAST software, confirming their effectiveness.