Guide to building better AI data centers in WA
A new guide for AI data center developers, developed by the Pacific Northwest National Laboratory and industry groups, aims to promote smarter and more responsible construction practices in Washington state. The guide focuses on energy efficiency, grid reliability, and mitigating the environmental impact of data centers, which are projected to quadruple energy consumption by 2035.
The Pacific Northwest National Laboratory (PNNL) in Richland, Washington, in collaboration with industry groups including the National Electrical Manufacturers Association (NEMA) and ASHRAE, has developed a new AI data center energy performance framework. This guide aims to help developers construct AI data centers in a smarter and more responsible manner, particularly regarding energy efficiency and grid reliability.
Patrick Hughes, Senior VP at NEMA, highlighted that NEMA's contribution focused on electrical systems, recommending that data centers incorporate energy storage, microgrids, and on-site generation to reduce their reliance on local power grids and mitigate impact on communities. This is crucial as AI data centers are projected to consume four times their current energy by 2035. Such measures could also enable data centers to export power to the local utility during peak demand, acting as grid assets rather than liabilities.
The framework also addresses the challenge of keeping pace with rapid data center build-out and innovation, which have outstripped the development of standards and building codes. ASHRAE contributed expertise on modern cooling methods that consume less water and energy. The guide emphasizes the importance of making data centers more self-reliant to avoid burdening local power grids and increasing residential utility bills. Hughes also stressed the increasing need for skilled electricians and electrical engineers to design these facilities responsibly in the future.