Gartner, Inc. predicts that by 2027, power shortages will limit operations in 40% of AI data centers due to the surging demand for electricity driven by AI and generative AI (GenAI). Electricity consumption for AI data centers is forecasted to grow by 160% over the next two years, with incremental power usage reaching 500 terawatt-hours (TWh) annually in 2027—2.6 times higher than 2023 levels.
Bob Johnson, VP Analyst at Gartner, highlights that hyperscale data centers for GenAI applications are creating a massive energy demand, outpacing utility providers’ ability to expand capacity. This delay in infrastructure development could lead to energy shortages, restricting new data center growth and increasing operational costs.
Electricity Costs and Sustainability Challenges
As power demand surges, electricity costs are expected to rise significantly, impacting AI operations. Gartner advises organizations to anticipate higher operational expenses and secure long-term power contracts. The growing reliance on fossil fuels to meet immediate power demands could hinder zero-carbon sustainability goals, forcing some facilities to delay retiring older power plants.
While renewable energy sources like wind and solar cannot provide the 24/7 reliability needed for data centers without alternative backups, technologies such as enhanced battery storage and small nuclear reactors could offer long-term solutions. Gartner recommends optimizing GenAI applications to reduce computing power requirements and considering alternatives like edge computing and smaller language models to mitigate sustainability challenges.
