Efficient asset management in the Oil and Gas industry is hampered by traditional reactive and preventive maintenance strategies, which often result in costly unplanned downtime, excessive resource utilization, and potential safety hazards. AI-driven Predictive Maintenance (AI-PdM) presents a transformative solution, leveraging machine learning models and sensor data to forecast equipment failures, thereby optimizing maintenance interventions. This study undertakes a systematic literature review, utilizing the Scopus database (2019–2024) and bibliometric analysis to examine the application and impact of AI-PdM in the O and G sector. Specifically, it investigates how AI-PdM enhances operational efficiency and, through this mechanism, improves financial performance. Our analysis reveals a significant surge in research interest, particularly post-2021. Key findings confirm substantial benefits, including reduced unplanned downtime, lowered maintenance costs, improved safety, extended asset lifespan, and enhanced environmental protection, supported by case studies from major industry players. Applications span drilling, pipelines, pumps, and compressors. Despite benefits, adoption faces challenges related to data quality, integration with legacy systems, model interpretability (necessitating Explainable AI - XAI), and cybersecurity. Empirical evidence suggests AI-PdM can decrease downtime by 30%, yielding tangible financial returns mediated by operational efficiency gains. Future directions include deeper integration with IoT, edge computing, 5G, and a growing focus on sustainability. This review provides actionable insights for industry stakeholders and identifies research avenues.