IEA scenario sees data centre electricity demand doubling

IEA scenario sees data centre electricity demand doubling

Global data centre electricity consumption could double by 2030. The IEA’s World Energy Outlook 2026 puts demand at approximately 1,000TWh under stated policies, with infrastructure constraints affecting the pace and…

IEA scenario sees data centre electricity demand doubling
Summary
  • The IEA’s World Energy Outlook 2026 places global data centre electricity consumption at approximately 1,000TWh in 2030 under STEPS.
  • The scenario reflects increased AI infrastructure investment, but it is not a prediction that every planned facility will be completed.
  • Grid capacity, electrical equipment supply, cooling infrastructure and project commissioning remain critical constraints on translating proposed capacity into operating demand.

Global data centre electricity consumption doubles to approximately 1,000 terawatt-hours (TWh) by 2030 under the Stated Policies Scenario in the International Energy Agency‘s World Energy Outlook 2026.

The report, released on 19 November, examines how electricity demand could develop under different policy and market conditions. Its Stated Policies Scenario, commonly abbreviated to STEPS, considers existing and announced policy settings alongside the agency’s wider assessment of economic and energy developments.

The 1,000TWh figure is a global scenario outcome rather than a forecast of electricity consumption guaranteed to occur. The IEA explicitly describes its scenarios as alternative ways of examining possible energy futures, not predictions of a single expected result.

Artificial intelligence infrastructure investment is a major contributor to the increase. More powerful computing equipment, expanding customer demand and larger data centre projects are adding to the electricity required by digital infrastructure.

The global figure does not describe a specific increase in European consumption. Growth in individual markets will depend on the facilities delivered, the available electricity supply and the extent to which computing equipment is installed and used.

Data centres are also part of a broader increase in electricity demand. The World Energy Outlook examines expanding electrification across industry, transport, heating, cooling and other activities, meaning computing infrastructure will compete for electrical capacity with several sectors undergoing their own transitions.

Grid connections determine when capacity can operate

A data centre development requires an electricity connection capable of supporting its planned load. The capacity available through that connection influences the amount of computing equipment that can operate, alongside the power required by cooling and other supporting systems.

Large projects may need new substations, transformers, transmission equipment or upgrades to distribution networks before they can enter service at their intended scale.

Those requirements introduce dependencies beyond the construction of the data centre building. A developer can complete substantial building work while still awaiting equipment or a suitable connection to the electricity network.

Connection conditions differ between locations. Some networks may accommodate additional large consumers with relatively limited reinforcement, while others require substantial investment or lengthy infrastructure works.

Electricity prices also influence site selection and operating economics. Data centres draw power over extended periods, so differences in energy costs can affect competitiveness even when suitable grid capacity is available.

Developers therefore need to assess the capacity they can obtain, the timing of delivery and the commercial arrangements governing electricity supply.

A planned facility’s electrical capacity is not the same as electricity consumption. A proposed 100MW installation may be built in phases, leaving part of its eventual electrical allocation unused until further computing equipment has been commissioned.

Annual consumption depends on the power actually drawn over time. A constant average load of 100MW over an entire year would consume approximately 876GWh, while a facility operating below that average would use less electricity.

The distinction becomes important when comparing development pipelines with the IEA’s demand scenarios. Proposed megawatts cannot simply be added together and treated as operating electricity consumption.

Planning approvals, customer agreements and grid connection applications also represent different levels of project maturity. None establishes that computing infrastructure has completed installation and begun serving production workloads.

Equipment supply affects the commissioning sequence

Electrical infrastructure depends on a range of specialist equipment, including transformers, switchgear, power conversion systems and distribution assemblies. Construction schedules must account for the availability and installation requirements of those components.

Supply chain constraints can affect the timing of a project even where the developer has secured financing and land. Equipment may also need factory testing, transport and further verification after being connected at the site.

High-density computing introduces additional requirements. Systems using large numbers of accelerators can concentrate substantial electricity demand within individual racks, increasing the importance of local power distribution and thermal management.

Many high-density configurations use direct liquid cooling to transfer heat from processors into circulating coolant. The heat must then pass through distribution equipment and into the facility’s wider cooling and heat rejection system.

The ability to cool computing hardware depends on available heat exchanger capacity, coolant temperatures, flow conditions and the supporting pumps and controls. These requirements must be coordinated with the electrical systems supplying both the IT equipment and the cooling plant.

An installation with sufficient electricity capacity may still be unable to operate its full computing load if the cooling system has not been completed or commissioned.

Network infrastructure creates another dependency. Large computing clusters require suitable connections between processing equipment, storage systems and external services. Their availability depends on hardware supply, installation, configuration and testing.

New equipment can also change the relationship between computing work and electricity consumption. More efficient processors may reduce energy requirements for a particular workload, while higher utilisation or expanding demand can increase total consumption.

Operating demand depends on actual utilisation

The total energy consumed by data centres reflects the number of active installations, their equipment configurations and the amount of time that systems operate at different loads.

Installed capacity can remain underused during early commissioning or customer occupation. Conversely, an established facility may increase annual consumption without expanding its electrical connection if the installed equipment operates at higher utilisation.

Supporting infrastructure contributes additional electricity demand. Cooling systems, power distribution losses and building services consume energy alongside the processors, storage and networking equipment performing computing work.

Efficiency measures can influence the amount of supporting energy required. Their effects depend on facility design, climate, operating temperatures and equipment utilisation, and cannot be inferred solely from a technology supplier’s maximum performance rating.

Operational flexibility may also affect how data centres interact with electricity networks. Some computing activities can potentially be scheduled or adjusted, but the amount of flexibility available depends on workload requirements, customer agreements and the technical characteristics of the installation.

Services requiring continuous availability may offer limited scope to reduce demand without affecting users. Backup power systems are primarily designed to protect critical loads, and their existence does not independently establish that a facility can offer sustained grid flexibility.

The IEA’s scenario therefore needs to be interpreted alongside the practical constraints affecting infrastructure delivery and operation. Investment in computing equipment, planned electrical capacity and measured consumption are related, but they represent different stages and characteristics of the industry.

Under STEPS, the agency’s modelling places global data centre electricity consumption at approximately 1,000TWh in 2030. The amount ultimately consumed will depend on the pace of construction, available power infrastructure, equipment delivery, utilisation and the demand for the computing services those facilities provide.


Stay updated with the latest insights and trends in the data centre industry by subscribing to our newsletter.

← Back

Thank you for your response. ✨