AI-Driven Data Centers to Triple Energy Demand by 2035, Posing Grid Challenges

# Data Center Energy Demand Forecasted to Soar Nearly 300% Through 2035

The global energy landscape is undergoing a seismic shift. As artificial intelligence continues to reshape industries and transform how we work, data centers have emerged as the new power-hungry giants of the digital age. Recent forecasts paint a striking picture: energy demand from data centers is projected to skyrocket in the coming decade, with some projections suggesting growth approaching 300% by 2035. This explosive expansion raises critical questions about grid capacity, energy infrastructure, and our ability to power the AI revolution sustainably.

## The Scale of Growth Ahead

The numbers are staggering. Gas-power generation for data centers alone is expected to more than double from 120 terawatt-hours (TWh) in 2024 to 293 TWh in 2035, with much of this growth concentrated in the United States[1]. Meanwhile, BloombergNEF’s latest forecast projects data-center power demand will hit 106 gigawatts (GW) by 2035—a 36% jump from their previous outlook published just seven months ago[2]. This rapid revision underscores how quickly projections are shifting as AI deployment accelerates beyond earlier expectations.

The uncertainty surrounding these figures is itself noteworthy. Modeled energy use projections through 2030 range dramatically from 200 TWh per year to over 1,050 TWh per year[3]. At the high end, this would represent approximately a quarter of all U.S. electricity generation in 2023. Even more conservative estimates from the Lawrence Berkeley National Laboratory suggest data centers will consume between 325 TWh and 580 TWh annually by 2030, representing 6.7% to 12% of total U.S. electricity consumption[3].

## AI’s Insatiable Appetite

What’s driving this unprecedented surge? Artificial intelligence is the primary culprit. Currently, AI accounts for approximately 5-15% of data-center power use, but this share is expected to increase dramatically to 35-50% by 2030[1]. As machine learning models grow more sophisticated and computational demands intensify, data centers must expand their infrastructure exponentially to meet these needs.

This growth, while substantial, should be contextualized within broader energy trends. The International Energy Agency (IEA) projects that a 530 TWh rise in electricity demand in data centers by 2030 would represent only 8% of the overall increase in global electricity demand during that period[1]. Nevertheless, the concentration of this demand in specific regions—particularly the United States—creates acute challenges for local grids and energy planners.

## The Energy Mix Challenge

A critical aspect of this expansion involves the fuel sources powering these facilities. Currently, the data-center electricity mix is roughly 60% fossil fuels and 40% clean power. By 2035, the IEA’s central scenario envisions this ratio flipping to 60% clean power and 40% fossil fuels[1]. This transition would be driven primarily by the wider global expansion of renewable energy, with some projects directly funded by data-center companies themselves.

However, this optimistic scenario masks a troubling reality: significantly more gas and coal power will likely still be required to meet data-center demand[1]. This means ramping up existing plants and constructing new fossil fuel infrastructure even as the industry pursues renewable energy. The tension between these competing imperatives represents one of the central challenges facing energy planners and policymakers.

## Grid Strain and Infrastructure Concerns

The rapid growth in data-center energy demand is reshaping U.S. electricity demand quickly and dramatically[2]. Utilities and grid operators face unprecedented pressure to expand transmission and distribution infrastructure while simultaneously managing the integration of renewable energy sources. The concentration of data centers in specific regions—driven by existing fiber-optic networks, cooling water availability, and proximity to tech hubs—creates localized bottlenecks that could constrain growth.

Regional electricity grids are already experiencing strain. While some studies found no evidence of national electricity demand growth, specific regional and utility demands are expected to increase substantially[3]. This uneven distribution means certain areas will bear disproportionate burdens of accommodating data-center expansion.

## Looking Forward

The trajectory through 2035 presents both opportunities and challenges. While the IEA expects electricity demand growth to roughly halve after 2030—from around 90 TWh per year out to 2030 down to less than 50 TWh annually thereafter—the cumulative impact remains enormous[1]. By 2035, total data-center electricity demand is projected to reach 1,193 TWh, reflecting the new reality of the digital economy.

The data-center energy boom is not merely an energy story; it’s an economic, environmental, and geopolitical one. How we power this expansion—whether through renewable energy, fossil fuels, or nuclear power—will shape our progress toward climate goals and energy security. The coming decade will test whether our energy infrastructure can keep pace with technological innovation, and whether we can harness AI’s potential while managing its substantial environmental footprint.

The forecast is clear: data centers will demand unprecedented amounts of energy. The question now is whether we’re prepared to meet that demand responsibly.


Original source: TechCrunch – Data center energy demand forecasted to soar nearly 300% through 2035