February 16, 2026 - AI Has a Power Problem

An analysis on the Life in the Markets podcast describes a power constraint for artificial intelligence. The technology requires substantial electricity. Chips for large language models and advanced computation use enormous amounts of power while data centers need continuous reliable supplies. By 2028 new data centers are projected to require 44 gigawatts of additional power capacity. Only about 25 gigawatts are scheduled to come online. This shortfall could lead to fewer data centers, reduced profit margins from higher energy costs or the need for rapid expansion of energy production. Inflation data shows electricity and utility costs rising significantly. Oil prices declined year-over-year. Energy generation is becoming the pressure point if AI demand continues to increase. The assessment concludes that AI has a power problem. The issue may restrain growth. It may shift capital allocation and alter market expectations for the sector.