Every answer from artificial intelligence is processed in a data centre, and data centres are using more and more electricity. This article answers three questions: how much electricity artificial intelligence uses today, how much it will use in the coming years, and what this means for electricity generation and prices, including in Georgia.
Consumption Is Still Small but Growing Fast
According to the International Energy Agency (IEA), the world’s data centres used 485 terawatt-hours in 2025, about 1.5 percent of the electricity consumed worldwide. By 2030 this volume will roughly double. The energy needed to process one question is small and falling quickly, but the number of questions and the share of complex tasks are growing even faster, so total consumption keeps rising.
The Effect on Prices Appears Where Centres Are Concentrated
The global share is small, but consumption is concentrated in a few places. In Ireland, data centres account for 23 percent of national electricity consumption. In the largest US electricity market, PJM, about 40 percent of the cost of recent capacity auctions is linked to data centres. New demand needs new generation, and the IEA estimates that gas and coal will cover about 40 percent of it. This is why several countries and states have changed their rules, so that the cost of new centres does not fall on other consumers.
What This Means for Georgia
Georgia has no surplus electricity. In 2025 the country’s net imports were about 1.05 billion kilowatt-hours, and the deficit is largest in winter. A single 100-megawatt data centre would use about 0.88 billion kilowatt-hours a year, about 6 percent of national consumption. Tariffs have already risen: from April 2026 the household tariff went up by 27.2 percent.
Georgia’s experience with cryptocurrency mining showed that cheap electricity attracts large loads, and a tariff increase sends them to another country. A large data centre therefore makes sense for Georgia only if new capacity is built and separate terms are set for large consumers.
This article was prepared by the academic team of the Business and Technology University (BTU) together with BTUAI and draws on publicly available research and official data. The full report, with all charts and sources, is available attached.
Citation: Business and Technology University (BTU) & BTUAI. (2026). Artificial Intelligence and Electricity. Tbilisi, Georgia.