Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteAI data centres use electricity to run servers and specialised chips that train and operate AI models, as well as to power storage, networking, cooling and other facility systems. Demand rises with the amount of computing installed and used, the kinds of AI tasks performed, and the efficiency and power availability of each site. AI is a major driver of growth, but it does not account for all data-centre electricity use.
Where a data centre’s electricity goes
A data centre is more than a room full of processors. It contains servers that process workloads, storage systems that hold data, networking equipment that moves it, and auxiliary systems that keep the site operating. The International Energy Agency (IEA) estimates that servers use around 60% of electricity in modern data centres on average. Storage uses about 5%, while networking can account for up to 5%.
As an Amazon Associate I earn from qualifying purchases.
Cooling and environmental control are another substantial load, but their share varies widely: about 7% in efficient hyperscale facilities and more than 30% in less-efficient enterprise facilities, according to the IEA’s 2025 analysis. UPS batteries, backup generators, lighting and other site infrastructure also draw power. The percentages are averages or ranges for different facility types, not a universal breakdown for every data centre. IEA: Energy demand from AI – Energy and AI (2025)
Why AI changes the demand picture
Accelerated servers add computing capacity
AI workloads often use servers equipped with specialised accelerators such as graphics processing units (GPUs), alongside conventional processors. In the IEA’s 2025 Base Case, electricity use by accelerated servers grows faster than use by conventional servers. The agency estimates accelerated servers account for almost half of net data-centre electricity-demand growth from 2024 to 2030. That is a substantial contribution, not evidence that every increase in data-centre demand is caused by AI.
Different AI tasks have different energy needs
Running a simple text-generation request is not equivalent to generating video, performing complex reasoning or carrying out agentic tasks. The IEA’s 2026 update identifies video generation, reasoning and agentic tasks as workloads that can use far more energy per query than simple text generation. These are comparisons between categories of tasks, not a fixed energy figure for every query or model.
Growth is broader than AI alone
Data centres also serve cloud computing, storage, streaming and other digital services. In the IEA’s 2025 Base Case, AI-driven accelerated servers account for almost half of net growth over the 2024–2030 period; other IT equipment and facility infrastructure contribute as well.
How much electricity do data centres use?
The IEA’s 2026 update estimates global data-centre electricity consumption at 485 terawatt-hours (TWh) in 2025 and projects about 950 TWh in 2030—around 3% of global electricity demand. The agency reports that total data-centre consumption grew 17% in 2025, while consumption at AI-focused data centres grew 50%. These are estimates and projections for broad categories, not measurements of a typical facility or an individual AI request. IEA: Executive summary – Key Questions on Energy and AI (2026)
Free tools Windows power users keep installed
One-click scans. No signup required.
The IEA’s 2025 report estimated 415 TWh of data-centre use in 2024, about 1.5% of global electricity consumption. That figure comes from an earlier report and should not be mixed with the newer outlook as if both belonged to one unchanged forecast series. IEA: Executive summary – Energy and AI (2025)
Rank #3
What affects electricity demand?
- How much computing is installed: More servers and accelerators can raise a facility’s maximum and actual power needs.
- How intensively equipment is used: Demand depends not just on installed capacity but on workloads and utilisation.
- Workload mix and model capabilities: AI uptake, the complexity of tasks and new uses can change how much computing is required.
- Hardware and software efficiency: More efficient equipment and computing methods can reduce electricity needed for a given workload, though gains may be offset if usage expands.
- Facility overhead: Cooling and other supporting systems add to the power drawn by IT equipment; their share depends in part on facility type and efficiency.
- Power and equipment constraints: Grid connections, electricity infrastructure, cooling capacity and equipment supply affect how quickly new sites can be built or expanded.
The IEA describes the energy demand of AI as the result of three fast-moving uncertainties: efficiency improvements, surging uptake and changing model capabilities, which can enable more energy-intensive uses. Its 2026 outlook says bottlenecks constrain more aggressive near-term scenarios, while easing constraints and wider adoption of energy-intensive tasks could raise demand beyond 2030.
Why a global percentage can hide local grid pressure
A global share is not a measure of the strain on a particular electricity network. Data centres are concentrated in specific places, so a facility or cluster can create substantial local demand even when the sector remains a modest share of worldwide consumption. The IEA’s 2025 executive summary estimates data centres account for around one-tenth of global electricity-demand growth to 2030 and warns that concentrated siting can make grid integration challenging. Grid constraints and project delays are relevant planning concerns. IEA: Executive summary – Energy and AI (2025)
Electricity consumption also has environmental implications beyond the power used at the facility: emissions depend on how that electricity is generated, and cooling can require water. The European Commission identifies both electricity-related emissions and cooling-water needs as concerns. It says it proposed a common EU data-centre rating scheme on 21 September 2026 to improve transparency on energy and water use; a proposal should not be mistaken for a fully implemented scheme. European Commission: Energy performance of data centres
How to compare claims about data-centre demand
Before comparing a forecast, facility or headline figure with another, check that they describe the same thing:
- Scope and date: Is the number global or regional, for all data centres or AI-focused sites, and observed consumption or a projection?
- Workload: Does it distinguish conventional servers from accelerated servers and account for the mix of AI tasks?
- Efficiency and utilisation: Does it account for equipment efficiency, how much the servers are used and whole-facility overhead?
- Infrastructure: Are grid connections, power equipment, cooling and the site’s location part of the estimate?
- Scenario assumptions: What does the forecast assume about AI adoption, efficiency gains and the pace at which bottlenecks are resolved?
Forecasts are scenarios, not guarantees. The IEA’s 2025 report sets out alternative cases reflecting uncertainty in AI uptake, efficiency and bottlenecks; its projections should not be described as observed outcomes. The 2026 update keeps the central outlook near the prior trajectory while noting potential for higher demand beyond 2030 if constraints ease and energy-intensive uses spread.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
