Data Centers: how they work, what they use, where they are

How a data center works

A data center is a building (or a campus of buildings) that houses the servers, storage and networking equipment that run online services, plus the power and cooling equipment that keeps them running. The IEA describes servers, storage and network equipment installed in racks, supported by cooling, uninterruptible power supplies, backup generators and other infrastructure.¹

The IEA’s breakdown of where a data center’s electricity goes is below. The shares vary by facility; the IEA gives ranges, not a single split.

Source for the table: IEA, Energy demand from AI.¹

Types of data center

LBNL sorts U.S. data-center space into internal (in-house), enterprise, colocation (space rented to many customers), hyperscale (very large sites run by the biggest cloud companies), communications service provider and edge types.² It reports that colocation and hyperscale facilities held 74% of U.S. server capacity in 2023 and projects 85% by 2028, and that the share of server energy used in internal data centers fell from over 60% in 2014 to nearly 10% in 2023.²

Measuring efficiency

Power usage effectiveness (PUE) is the ratio of a facility’s total energy to the energy used by its computing equipment; lower is better (glossary). LBNL reports an average U.S. PUE of about 1.6 in 2014 and 1.4 in 2023, and a projected range of 1.15 to 1.35 by 2028.²

How AI adds to power demand

The IEA reports that electricity use by accelerated servers (those with GPUs and similar chips, mostly for AI) grows about 30% a year in its Base Case, against about 9% for conventional servers.¹ LBNL attributes the more-than-doubling of U.S. data-center electricity use between 2017 and 2023 largely to GPU-accelerated servers.² The IEA says a typical AI-focused data center uses as much electricity as 100,000 households, and that the largest ones now under construction will use about 20 times that.³

LBNL estimates that inference (answering requests) accounted for nearly 60% of AI-server energy in 2023 and projects that training would account for 50–53% by 2028.² The 2028 figure is a projection.

Power

Worldwide

• Data centers used about 415 TWh of electricity in 2024, about 1.5% of global electricity use.³
• The United States accounted for 45% of that, China 25% and Europe 15%.³
• Data-center electricity use has grown about 12% a year since 2017, more than four times faster than total electricity use.³
• Projection: the IEA’s Base Case reaches about 945 TWh in 2030 (slightly more than Japan’s total electricity use today) and about 1,200 TWh in 2035, with a range of 700 to 1,700 TWh across its cases.³
• The IEA puts data centers at roughly one-tenth of global electricity-demand growth to 2030.³
• The IEA’s Base Case has data-center emissions rising from about 180 million tonnes of CO₂ to about 300 million tonnes by 2035, under 1.5% of energy-sector emissions.³

United States

• LBNL: U.S. data centers used 176 TWh in 2023, 4.4% of U.S. electricity. That compares with 76 TWh (1.9%) in 2018.²
• Projection: LBNL’s 2028 range is about 325 to 580 TWh, or 6.7% to 12.0% of U.S. electricity.²
• Projection: the EIA’s AEO2026 estimates data-center servers at about 7% of commercial-sector electricity use in 2025 and 22% to 33% by 2050, depending on the case.⁴
• The EIA found that growth in U.S. commercial electricity demand between 2019 and 2023 was concentrated in a few states, led by Virginia (+14 billion kWh) and Texas (+13 billion kWh).⁵

Concentration and the grid

• The IEA reports that nearly half of U.S. data-center capacity sits in five regional clusters, and that about half of U.S. capacity now under development is in existing large clusters.³
• The IEA estimates that about 20% of planned data-center projects worldwide could be delayed by grid constraints.³
• Ireland: data centers used 6,969 GWh in 2024, 22% of the country’s metered electricity, up from 5% in 2015.⁶
• Stanford’s AI Index 2026 estimates that AI data-center power capacity reached about 29.6 GW by the fourth quarter of 2025.⁷ That is an estimate built from chip data, not a metered figure.

Water

Water use has two parts. Direct use is water a facility itself consumes, mostly for cooling. Indirect use is water consumed by the power plants that supply its electricity. The USGS explains how “consumption” (water not returned to its source) differs from “withdrawal.”⁸

• LBNL: direct water use by U.S. data centers rose from about 21.2 billion liters in 2014 to about 66 billion liters in 2023. Indirect water use associated with their electricity was about 800 billion liters in 2023.²
• Projection: LBNL estimates direct water use by hyperscale data centers alone at 60 to 124 billion liters in 2028.²
• Reported Google estimates that a median Gemini text prompt uses about 0.26 mL of water. Google says this has not been independently verified.⁹ See the bottle-of-water myth.
• Reported Amazon says that from October to March its Project Rainier data centers in St. Joseph County, Indiana will not use water for cooling, and from April to September only for a few hours a day on an average day.¹⁰

Cooling design differs by site and climate, so one site’s figure does not predict another’s. The deeper explanation is on Is data-center water use just cooling?

Major sites and programs by country

Most announced figures are plans or company statements, not measurements, so they carry the Reported label. Capacity is quoted in the unit each source uses (gigawatts of power, not square feet) and should not be added across rows. This is a sample of widely documented projects, not a complete list.

How many data centers, by country

Stanford’s AI Index 2026 reports data-center counts for 2025 compiled from Cloudscene.⁷ These are counts of facilities, not capacity: one hyperscale campus and one small server room each count as one, so the table says little about how much computing a country has. Only the countries the source lists are shown.

Myths vs facts

Keep reading

• Are AI data centers eating the world’s electricity?
• Is data-center water use just cooling?
• Does each AI question use a bottle of water?
• All Myths vs Facts
• AI glossary
• Data-center news