AbstractFault diagnosis is a novel technology crucial for monitoring the proper functioning and ensuring the stability of mechanical devices and components. Nevertheless, most existing data‐driven methods for rolling bearing fault diagnosis exhibit limited diagnostic capabilities in scenarios characterized by noise interference and inadequate training data. To address this issue, this paper proposes a novel intelligent fault diagnosis method for rolling bearings based on Capsule Network with Fast Routing algorithm (FCN). Firstly, the vibration signal is transformed into a time‐frequency map through continuous wavelet transform (CWT), and the transformed time‐frequency map is input into the network model to enable the network to learn features more fully. Subsequently, this paper introduces FCNs into capsule networks, effectively mitigating the extended training times typically associated with capsule networks and reducing the demands on training equipment. Extensive experiments are conducted utilizing two distinct bearing datasets to assess the method's stability and generalization. The results of these experiments demonstrate the proposed approach's ability to maintain robust fault diagnosis capabilities, even in the presence of noise interference and limited training data. This innovative method lays the foundation for intelligent rolling bearing diagnosis and is readily adaptable to other rotating components.