Despite the continuing focus on less energy it uses, the development of cloud and edge computing has increased the demand of enforcing strong SLAs. This paper comparatively gestures towards energy efficiency and SLA compliance of the different ML classifiers under the cloud and edge computing architectures. In this work, we evaluate the potential of conventional and AI-based SVM, NN, DT, and RF schemes to minimize the energy consumption and satisfy SLA commitments. Although we illustrate the efficacy of our approach with high-level visualizations, as these stacked bar graphs, radar chart, bubble plot and box graph, we provide a complete evaluation in terms of time execution, SLA fulfillment rate and energy saving reduction rate. We hope that such study would provide new understandings over energy and SLA tradeoffs among classifiers when the AI models- mainly neural networks- have shown enhancements to reach the ultimate energy saving and yet satisfy the SLA as best as possible. This study also has the impacts by has a uses and towards optimization of resources in cloud edge computing systems which also aims for sustainable computing and adaptive resource management for future distributed systems.