IEA Projects Global AI Data Center Power Demand to Reach 1,200 TWh by 2035

Global artificial intelligence workloads are set to trigger a massive surge in grid demand over the coming decade, according to new projections from the International Energy Agency (IEA) detailed by Yahoo Finance (https://finance.yahoo.com/energy/articles/worlds-top-energy-watchdog-made-183500306.html). The energy watchdog estimates that global electricity usage by AI data centers will expand from 415 terawatt-hours (TWh) in 2024 to 945 TWh by 2030, representing more than a twofold increase. Looking further out, the IEA projects power requirements to reach 1,200 TWh by 2035, nearly triple the baseline figure recorded in 2024.

To put the anticipated 2030 consumption figure into perspective, the agency noted that AI-driven power demand will exceed the total current electricity consumption of Japan. While data center expansion is occurring worldwide rather than in a single territory, historical trends underscore the trajectory: power usage across data centers has expanded at an annual rate of 12% since 2017, outstripping total energy consumption growth by more than fourfold.

As regulated electrical grids struggle to keep pace with rapid infrastructure buildouts, off-grid energy providers are absorbing significant demand. Bloom Energy (NYSE: BE), which manufactures hydrogen fuel cell systems designed for independent power generation, entered 2026 with a product backlog of $6 billion—reflecting a 140% jump compared to its backlog at the start of 2025. In addition to hardware sales, the company recorded a service contract backlog of $14 billion at the beginning of 2026, establishing recurring multi-year revenue streams.

The massive power requirements of hyper-scale technology platforms have also catalyzed interest in clean, continuous nuclear power generation. Constellation Energy (NASDAQ: CEG), one of the largest nuclear plant operators in the United States, is already working with major technology enterprises, including Meta Platforms (NASDAQ: META) and Microsoft (NASDAQ: MSFT). Unrestricted by utility rate regulations, Constellation retains the flexibility to price its power generation at prevailing market rates.

Upstream fuel suppliers are similarly positioned to capture shifts in demand resulting from nuclear energy adoption. Canadian producer Cameco (NYSE: CCJ), among the world's largest suppliers of uranium, anticipates that global demand for nuclear fuel will outpace supply by the mid-2030s. The projected market deficit is expected to give raw material producers enhanced pricing leverage as nuclear capacity scales up to support technology infrastructure.

Traditional regulated utilities are also taking strategic steps to serve the expanding artificial intelligence ecosystem. NextEra Energy (NYSE: NEE), which maintains a 3% dividend yield and a history of annual dividend increases extending over three decades, is pursuing an acquisition of Dominion Energy (NYSE: D). The transaction would grant NextEra direct exposure to key data center hubs where Dominion currently holds utility territories.

The IEA data highlights that the power requirements of artificial intelligence are reshaping the broader energy sector. As tech firms race to secure gigawatts of capacity for training and inference operations, energy companies ranging from fuel cell makers and uranium miners to nuclear operators and traditional utilities are aligning their capital allocation strategies with the technology industry's infrastructure pipeline.