Ultra-precision machining (UPM) demands extreme accuracy, often at the nanometer scale, and stringent stability. To enhance its intelligence and digitalization, developing a digital twin (DT)—a real-time virtual replica of the machining process—can revolutionize process control, predictive maintenance, and efficiency. A UPM DT integrates real-time sensor data, high-fidelity simulations, and AI-driven analytics to monitor and optimize machining operations continuously. It contains three layers: the physical layer, the data layer, and the service layer. Machining status is accurately mapped by leveraging multi-sensor fusion technology and MTConnect-based data transmission. The fundamental services offered by the service layer are intelligent monitoring, dynamic control and process planning. Since UPM processes are sensitive to slight disturbances, in-process predictive monitoring, powered by real-time data analytics, AI, and IoT sensors, enables the timely diagnosis of machining performance. This predictive insight facilitates dynamic control and process planning to guarantee product quality and improve overall system performance. The application of the proposed digital twin system is presented, where a data model is constructed based on the MTConnect protocol. An intelligent chatter detection model is developed and integrated into the intelligent monitoring module to prevent surface defects before they occur. The result shows that the DT system realizes effective real-time prediction of machining conditions, which is updated in the information model and facilitates control operations. Additionally, versatile HMI techniques, such as VR and AR, are developed for remote monitoring and offer immersive visual experiences.