The research develops an upgradeable predictive maintenance solution that uses IIoT technology together with AI analytics to support lathe machines in small and medium-sized enterprises. A proposed system collects vital machine parameters such as vibration and temperature and RPM and power consumption by implementing IoT sensors as data acquisition points. Mathematical models in machine learning process the collected data to detect anomalies and make failure predictions. The NodeMCU microcontroller serves for data retrieval needs and MQTT functions as a system to establish immediate messaging capabilities. The solution offers scalable design and cost-efficiency which makes it suitable for SMEs. Through PowerBI dashboard alongside Android application users can evaluate data and receive warnings and create visual analytics interfaces. The system operates in actual industrial settings for comprehensive testing which provides entire system capabilities beginning from hardware deployment through data processing to live user access. These test cases showed that the system predicted failures with 96.7% accuracy using an F1-score of 0.58 which helped SMEs reduce their unplanned downtime by 40% without exceeding $70 per sensor module implementation costs.