Machine learning is used to empower a machine tool, which gives rise to a new generation machine tool, i.e. cyber-physical machine tool. The use of four sensors to measure the cutting force, vibration, acoustic emission, and spindle motor current of an end milling machine is proposed. Sixty-five cutting tests using an end milling machine were conducted, during which sensor data was recorded. The flank wear exhibited on the tool following each cut was then measured using a microscope. This provided a labelled data set on which to train four machine learning algorithms: Support Vector Regression, Random Forests, Feed-Forward Back-Propagation Artificial Neural Networks, and Polynomial Regression. These were then compared and it was found that an artificial neural network provides the most accurate predictions of tool flank wear, with a mean absolute percentage accuracy of 90.11%. Using this trained neural network model, a real-time tool wear prediction system was implemented in LabVIEW. This tool condition monitoring system can be used to increase efficiency of manufacturing processes. © 2020 IEEE.