The advent of quantum computing has stimulated research into quantum-enhanced classifiers, promising significant improvements over classical counterparts. Quantum Support Vector Machines (QSVMs), can provide quantum speedups potentially useful for cybersecurity applications such as anomaly detection in Cyber-Physical Systems (CPSs). However, practical assessments on operational data reveal significant challenges. This paper compares QSVM with classical SVM classifiers using real-world Operational Technology (OT) traffic data to explore whether quantum-enhanced models can approximate or potentially match the performance of classical machine learning algorithms that are already well-established and widely adopted for this use case. Our empirical evaluation demonstrates that Classical SVMs achieved 98% accuracy when trained on the full OT network traffic dataset. Due to current quantum kernel limitations, the QSVM was restricted to a reduced dataset of samples with 5000 for training and 2500 for testing, attaining an accuracy of 89%. Notably, a classical SVM trained on the same subset reached only 78%. This comparative analysis reveals important insights into the readiness of quantum-based methods for real-world cybersecurity applications and highlights directions for future research. While current hardware constraints prevent fully exploiting quantum speedup, the QSVM’s superior performance on reduced data is a promising indicator of its potential. Nevertheless, classical approaches currently remain the most practical and effective solution for anomaly detection in large-scale OT environments.