This work highlights the gains of a fast nonlinear model-based predictive control (NMPC) scheme applied to a 10 kW proton exchange membrane fuel cell (PEMFC). The freshness of the approach is based on a particular parameterization of the control action to decrease the optimization problem dimension. Due to its short computational time, its reliability and its low sensitivity to noise, an artificial neural network (ANN) model is designed and used as a predictive model. The performance of the proposed control strategy is confirmed thanks to simulations through varying control scenarios. Results show good performance in setpoint tracking accuracy and robustness against plant/model mismatch. Moreover, for similar setpoint tracking accuracy, the proposed NMPC scheme appears to be thirty times faster than a classical NMPC strategy. Therefore, the fast NMPC scheme proposed in this work appears to be a promising candidate to achieve real-time implementation.