The digital transformation of energy systems increasingly relies on Artificial Intelligence (AI) and the Internet of Things (IoT), yet the literature lacks an integrated perspective on their joint contribution to sustainability outcomes and the UN Sustainable Development Goals (SDGs). This paper fills that gap through a systematic review and qualitative meta-synthesis of 42 peer-reviewed articles (2023–2026), framed within the Multi-Level Perspective (MLP) on socio-technical transitions. PRISMA 2020 guides the search and screening process, while the synthesis follows the seven-step meta-synthesis approach of Sandelowski and Barroso (2007). Our synthesis identifies five thematic clusters through which AI and IoT co-shape the energy transition: renewable forecasting and optimization, smart grid stability, demand-side efficiency, predictive asset maintenance, and decentralized market architectures. Evidence indicates that AI-IoT convergence is associated with improvements in forecasting precision, grid reliability, and operational productivity, with reported contributions to SDG 7, SDG 9, SDG 11, SDG 12, and SDG 13. However, the strength of this evidence varies: some relationships are empirically demonstrated, others are indirect or context-dependent, and several remain theoretically inferred. These distinctions are made explicit throughout the review. However, the review also surfaces critical tensions that complicate purely optimistic narratives notably rebound effects, the substantial energy cost of AI computation, and rising cybersecurity threats. To reconcile these dual dynamics, we propose a conceptual framework that systematically aligns AI-IoT capabilities with SDG-anchored transition pathways, offering researchers and policymakers a balanced lens for evaluating both technological opportunities and their inherent sustainability trade-offs.