Data Centers Must Tackle AI and Legacy IT Energy Use

Generative AI is pushing data centers toward power-hungry GPU systems, raising electricity, cooling and carbon-management costs. Yet the sustainability challenge extends beyond new AI clusters: conventional servers still run cloud services, databases and core business applications at enormous scale. That makes energy efficiency a portfolio-wide issue. Companies focusing only on AI infrastructure risk overlooking the larger installed base of mainstream IT, where hardware refreshes, software rationalization and higher utilization can deliver immediate savings.

Hewlett Packard Enterprise Chief Sustainability Officer Monica Batchelder said on April 22, 2026, that AI accounted for only 10% to 15% of worldwide data-center electricity use in 2024, versus 85% to 90% for general-purpose IT. Uptime Institute modeling cited by HPE puts AI’s share at just 20% to 30% by the end of 2030. HPE urged operators to modernize aging hardware, retire inefficient applications, rationalize storage, improve workload placement and adopt more efficient airflow and liquid-cooling systems.