This article develops a method to address the observer-based fuzzy control for a class of discrete-time nonlinear networked control systems under multichannel attacks, in which taking semi-Markov process to characterize the switching among various attack threats on multiple communication channels. The nonlinear networked control systems are modeled by a Takagi–Sugeno fuzzy model, where the membership function of the controller depends on the estimated state after observation. Moreover, for solving the problem of inaccessible actual attack modes, an observed-mode sequence driven by emission probabilities is adopted. In the framework of the hidden semi-Markov process with a double-layer stochastic structure, the analysis for $\varepsilon$-error mean square stability and conditions for observer/controller synthesis of the resultant system are obtained via a Lyapunov function relied on both real and estimated attack modes. Finally, a simulation of the nonlinear mass-spring-damper mechanical system is conducted, demonstrating the proposed method's feasibility.