TITLE:
Artificial Intelligence and Energy Overconsumption: Data Center Electricity Demand, Cooling Burdens, and Regional Sustainability Constraints
AUTHORS:
David Crovato, Jose Rosania
KEYWORDS:
Artificial Intelligence, Energy Consumption, Data Centers, Electricity Demand, Cooling Systems, Energy Infrastructure, Grid Capacity, Renewable Energy Integration, Sustainability, Carbon Emissions, Energy Policy, Resource Constraints, AI Scalability, Power
JOURNAL NAME:
Energy and Power Engineering,
Vol.18 No.4,
April
23,
2026
ABSTRACT: Artificial intelligence is increasingly embedded in everyday life, but its expansion depends on physical infrastructure with significant electricity, cooling, and water requirements. In this paper, “overconsumption” refers not simply to high electricity use, but to levels of AI-related infrastructure demand that intensify grid strain, raise carbon emissions where fossil generation remains dominant, increase cooling and water burdens, or displace other uses of limited energy resources. Rather than treating AI as a purely digital technology, this study examines it as a material system shaped by energy, environmental, and regional constraints. Using secondary sources from peer-reviewed research, international energy institutions, government reports, and industry analyses, the paper distinguishes between total data center electricity demand and the share specifically linked to AI workloads. It compares measured observations with scenario-based forecasts to examine how AI-driven expansion may alter electricity demand across regions. Particular attention is given to cooling intensity, water use, climate conditions, and grid limitations. The analysis finds that AI-related data center growth is becoming an increasingly important driver of electricity demand, especially in regions with high cooling requirements, limited grid capacity, or water stress. It also finds that sustainability outcomes differ sharply across regions, meaning that the environmental burden of AI expansion is unevenly distributed. While improvements in hardware efficiency, cooling systems, and renewable integration may reduce some pressures, continued AI growth without coordinated infrastructure planning risks worsening resource strain and reinforcing regional inequalities.