AI Data Centers Could Require 50 Gigawatts of Power by 2030, IEA Projects
The International Energy Agency projects that AI data centers could require 50 gigawatts of power by 2030, raising concerns about strain on global electrical grids. A bill has been proposed in the U.S. to restrict data center development on federal land as scrutiny of AI infrastructure grows.
AI data centers could require 50 gigawatts of power by 2030, according to projections from the International Energy Agency released in November 2026. The figure has prompted growing scrutiny of the environmental and grid impacts of AI infrastructure.
A bill has been proposed in the U.S. Congress to restrict data center development on federal land, reflecting concerns about the scale of energy consumption tied to AI computing.
The power demands of AI are driven by the massive computing clusters needed to train and run large language models and other AI systems. Data centers require not only electricity for computing but also significant cooling infrastructure.
Google announced plans to launch AI server satellites by 2029 to test solar-powered orbital computing, a potential long-term solution to the energy challenge. The project involves a partnership with Blue Origin.
OpenAI unveiled "Jalapeño," its first custom inference chip developed with Broadcom, aimed at reducing the company's dependence on Nvidia's supply chain and potentially improving energy efficiency.
The energy debate is one of several infrastructure challenges facing the AI industry. A series of major cloud outages in November 2026 affected Microsoft 365, AWS, and Cloudflare, causing disruptions in aviation and financial services and raising questions about the resilience of global digital infrastructure.


