1. Local Solutions

At the facility and community level, the goal is to shrink the immediate footprint on power and water systems.

According to the U.S. Department of Energy, improving energy efficiency is one of the fastest levers. This includes better server utilization, more efficient chips, and reducing idle computing. Cooling is another major opportunity, with ASHRAE emphasizing alternatives like liquid cooling and advanced airflow design, which can significantly cut both electricity and water use compared to traditional air cooling.

Water risk can also be reduced by shifting to closed-loop or air-based cooling systems and by siting data centers in regions with more abundant water resources. Some operators are beginning to reuse waste heat for nearby buildings, which turns a local burden into a local benefit. Grid strain can be managed by colocating data centers with renewable energy sources or by using on-site generation and energy storage to smooth demand spikes.

2.Broader Solutions

At the larger scale, the issue is not just efficiency but the source and scale of energy. According to the International Energy Agency, one of the most important steps is accelerating the transition to low-carbon electricity, since data center emissions are largely indirect. This includes long-term renewable energy contracts, investment in grid decarbonization, and in some cases pairing data centers directly with wind, solar, or nuclear power.

Researchers at Lawrence Berkeley National Laboratory emphasize that efficiency alone will not offset rising demand, especially with AI. That means systemic solutions also include rethinking demand itself, such as improving software efficiency, optimizing AI models, and avoiding unnecessary computation.

Policy also plays a role, through energy standards, transparency requirements, and incentives for low-impact design and location.

Download PDF

  • Our downloads have everything you need to supplement this course.