This paper presents management and monitoring of energy consumption within High-Performance Computing (HPC) Artificial Intelligence (AI) workloads and insight of the first launched EuroHPC JU system, EuroHPC Vega. The Vega entered its fourth year of operation, as it provides essential infrastructure for the Slovenian scientific community, and projects within scientific sphere. It is necessary to curb the latter and introduce effective and essential mechanisms for energy sustainability, leading to high energy consumption and carbon emissions in Data Centers (DC). Novel approaches within AI domain and workloads grow in size and the complexity requires a considerable amount of computational resources. Due to the different architectures between DCs, introducing different mechanisms with the tendency towards reducing power consumption, and more efficient approaches can be used. Differences in systems and their architectures lead to different end solutions, without a unified framework to govern such systems. The current setup, energy-efficient concepts, technologies, and mechanisms are presented, including configuration adjustments made to improve energy consumption management and monitoring on a case of HPC Vega.