AI adoption could strain global power grid as costs drop

McKinsey and Boston Consulting Group have warned that the growing affordability of AI could lead to increased electricity demand. As companies expand AI use beyond chatbots, cheaper models and lower token prices may drive more frequent and higher-volume AI adoption, potentially straining the electric grid. Data centers, already major power consumers, are expected to see global electricity demand rise by 24% annually through 2030, slowing to 5% between 2030 and 2040. McKinsey noted that data-center electricity demand is the fastest-growing load segment in OECD power markets. Meanwhile, BCG found that leading AI adopters are actively managing token spending, with three-quarters of its 'future-built' companies setting explicit token budgets. Half of these companies also encourage employees to use paid AI tools to boost adoption, compared to 25% of less advanced firms. McKinsey emphasized that while AI efficiency improvements may reduce power intensity, the overall demand is likely to rise as AI becomes more widely used.