AI Energy Surge

By Lena Park

Artificial intelligence requires a large and growing amount of electricity to function. As AI systems become more widely used, the energy needed to build and run them has increased substantially. Governments, companies, and researchers are paying close attention to this shift, and for good reason. The scale of growth is significant enough that it is starting to influence how countries plan their energy infrastructure.

Large AI models are trained and operated on servers housed in data centers. These facilities run continuously, consuming electricity both for processing and for keeping the hardware cool. According to the International Energy Agency, training a single large AI model can use as much energy as hundreds of homes consume in a month, and that is before the model is even made available to the public. Once deployed and handling millions of requests every day, the energy use continues at a steady and substantial rate. Unlike most software, AI systems run under constant, high-volume demand, which means the electricity costs do not pause. Morgan Stanley also states that the biggest tech companies are expanding their data center infrastructure at a rapid pace. Google, Microsoft, Meta, and Amazon have all committed to major investments in new facilities to support AI workloads. Some projections estimate that data centers will account for a significantly larger share of global electricity consumption by 2030 than they do today. In certain regions, this spike in demand is already putting pressure on local electrical grids. Some utility companies have noted that a single large data center can draw as much power as a small city, and that building the infrastructure to support it takes years of planning that has not always kept up with the pace of expansion.

MIT Technology Review also states that water is another resource under pressure. Data centers rely on large amounts of water to cool their systems, and in regions already affected by drought or limited water supply, this adds another layer of environmental concern beyond electricity alone. The physical footprint of AI infrastructure extends across land, water, and the raw materials needed to manufacture the hardware itself.

The energy demands of AI are real and growing. How the industry manages this, and how quickly clean energy can be scaled to meet the demand, will have a direct impact on global emissions in the years ahead. There is no straightforward solution, but it is a challenge that requires serious planning rather than being treated as secondary to technological progress.

Works Cited

"AI’s Energy Usage and Climate Footprint: What Big Tech Isn’t Telling You." MIT Technology Review, 20 May 2025, www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/.

"Energy and AI: Energy Demand from AI." International Energy Agency, 2024, www.iea.org/reports/energy-and-ai/energy-demand-from-ai.

"Powering AI: Energy Market Outlook 2026." Morgan Stanley, 2024, www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026.

Page updated

Google Sites

Report abuse