This paper explores the integration of Artificial Intelligence (AI) with the Internet of Things (IoT) to revolutionize predictive maintenance practices within smart factories. As industries increasingly adopt IoT-enabled devices, the ability to forecast machine failures and reduce operational disruptions has become essential for maintaining productivity and reducing costs. Traditional maintenance approaches, such as reactive and preventive maintenance, often lead to unforeseen downtime or over-servicing, and neither optimally support smart manufacturing environments. Predictive maintenance, powered by AI algorithms, enables real-time data analysis from IoT devices to predict potential failures with high accuracy, allowing preemptive measures to be taken only when necessary. This paper outlines the architecture of an AI-driven predictive maintenance system, reviews key AI techniques applied in this context, and analyzes case studies showcasing successful implementations. Challenges such as data privacy, high implementation costs, and the need for specialized skills are also discussed. The results demonstrate the substantial impact of AI on reducing maintenance costs and enhancing machine longevity, underscoring the relevance of AI in IoT-enabled industrial environments.