Communication problems in the sensor‐to‐controller and controller‐to‐actuator channels can cause both the controller and actuator to run asynchronously with the original system in different operating modes. This paper investigates an optimal fault‐tolerant control approach for Markov jump power systems (MJPSs) with asynchronous controller and actuator. Firstly, an asynchronous controller is proposed to deal with incomplete information (i.e., system modes) transmission in the sensor‐to‐controller channel. Secondly, a new asynchronous actuator faults model is constructed to simultaneously represent the two partial losses, of modes information in the controller‐to‐actuator channel, and of control effectiveness (LoCE) caused by actuator faults. Under this framework, two related hidden Markov models (HMMs) are formed, which reveal that both the controller and actuator are asynchronous with the controlled system in different modes. By using Lyapunov and optimal approaches, sufficient conditions are derived to ensure that MJPSs are mean square stable with an optimal guaranteed cost. Finally, Monte Carlo simulations are used to verify the effectiveness of the proposed control method.