Data Centers on the Brink: The Alarming Surge in Electricity Consumption by 2035

TL;DR

• Global data center electricity use is rising fast, with the International Energy Agency estimating it at about 415 TWh in 2024 and projecting roughly 945 TWh by 2030 in its base case.
• By 2035, the IEA sees demand reaching about 1,200 TWh in its base case, while more aggressive scenarios push it above 1,700 TWh; other analysts project similarly steep growth.
• The surge is being driven primarily by AI, raising fresh concerns about grid capacity, emissions, cooling, and the sustainability of rapid hyperscale buildouts.

The electricity appetite of data centers is becoming one of the defining infrastructure stories of the AI era. Recent forecasts show demand accelerating sharply through the end of the decade and into 2035, with the scale of growth now large enough to matter for national grids, power planning, and climate targets.

A power load that is no longer marginal

The International Energy Agency estimates that data centers consumed around 415 TWh of electricity in 2024, equal to roughly 1.5% of global electricity demand. In its base case, that figure more than doubles to around 945 TWh by 2030 and reaches about 1,200 TWh by 2035.

The same IEA analysis says a more aggressive scenario could push 2035 demand above 1,700 TWh, while a more efficiency-driven pathway could hold consumption closer to 970 TWh. That spread underscores how much the outcome depends on AI adoption, hardware efficiency, and policy responses.

AI is the main accelerant

The IEA identifies AI as the single most important driver of the coming surge, alongside broader demand for digital services. It also says that from 2024 to 2030, data center electricity consumption grows by around 15% per year in its base case, more than four times faster than total electricity growth in all other sectors.

Industry forecasts echo that concern. Deloitte estimates that AI data center power demand could rise from 4 GW in 2024 to 123 GW by 2035, a more than thirtyfold increase. Deloitte also says AI could grow from 12% of total data center power demand in 2024 to 70% by 2035.

The U.S. buildout is especially intense

In the United States, several analyses suggest the boom could be even more concentrated and harder to forecast. The World Resources Institute notes that one review by Rystad Energy found more than 100 GW of data center demand coming online between 2024 and 2035, while other estimates vary widely depending on assumptions.

BloombergNEF, as reported by Utility Dive and TechCrunch, projects U.S. data center power demand could reach 106 GW by 2035, up from about 40 GW today. TechCrunch also reports that much of this growth is expected to move into more rural areas as large sites become harder to place near cities.

Why the comparison to a whole country matters

The most striking comparisons are meant to convey scale. The IEA says global data center consumption could reach roughly 945 TWh by 2030 in its base case, and its higher 2035 case exceeds 1,700 TWh. That kind of consumption is comparable to the electricity use of major industrial economies, which is why analysts increasingly frame data centers as a system-level energy issue rather than a niche corporate one.

Separate forecasts vary substantially, but they point in the same direction. Precedence Research projects 1,300 TWh of global data center consumption by 2035, up from 460 TWh in 2024. Even where estimates differ, the common theme is sustained, high growth.

Emissions, grids, and cooling pressure

The IEA says emissions from electricity use by data centers rise from 180 million tonnes today to 300 million tonnes in its base case by 2035, and as much as 500 million tonnes in a higher-growth case. While that remains a smaller share of total energy-sector emissions, the report says data centers are among the fastest-growing sources of emissions.

The grid challenge is not only about total energy use but also location and timing. The WRI notes that some studies find no large national increase, even as regional and utility-level demand spikes sharply, which can strain local infrastructure long before national statistics fully reflect the pressure. That helps explain why utilities, regulators, and developers are increasingly focused on substation upgrades, transmission access, and long-lead power procurement.

The efficiency question

Not all projections point to runaway growth forever. The IEA notes that efficiency improvements could materially reduce demand, including a scenario in which global data center electricity use reaches around 970 TWh by 2035 instead of 1,200 TWh or more. In that lower-growth pathway, the global share of electricity consumed by data centers stays closer to 2.6%, rather than rising toward 4.4% in the higher-demand case.

That gap is why chip efficiency, workload optimization, better cooling, and more strategic siting are becoming central to the conversation. The coming years will likely determine whether AI infrastructure becomes a manageable part of the energy system or a persistent stress test for power planners.

What to watch next

The most important near-term signals are utility interconnection queues, new power purchase agreements, grid upgrades, and the pace of AI deployment. If the current wave of construction continues, the industry may need to prove that its expansion can be matched by low-carbon generation, faster transmission buildout, and more efficient computing.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.