Data centres and the UK National Accounts

As the UK economy rapidly adapts to growth in artificial intelligence (AI) and cloud computing technologies, data centres play a fundamental role in meeting the additional demands for data storage and compute capacity. Data centres are now designated as critical national infrastructure. This designation places data centres alongside functions, such as energy and water supply. It reflects their role in supporting essential services and economic activity, reinforcing their importance in national infrastructure planning.

This supporting methodology article explains where data centres appear within business investment statistics and other areas of the national accounts and environmental accounts. This article also explains why we are currently unable to separately identify either the investment in or output of data centres.

It is one of a series of articles we will publish on the digital economy including digital infrastructure and the development of an AI thematic account. The account forms part of our Surveys and economic statistics improvement plan. It provides one potential framework for tracking AI-related investment, production and use across the economy, while helping to assess the role of supporting infrastructure such as data centres.

As the use of AI, cloud computing and digital infrastructure continues to grow, we will continue to assess how these activities are represented within economic statistics.

Data centres are the physical facilities, which are primarily dedicated buildings, that house computing infrastructure that is typically used to deliver memory and processing power for artificial intelligence (AI), cloud computing, data storage, and server hosting.

Data centres come in several types, each with varying energy needs and capacity. Data centres do not have an unique industrial classification. However, they can be broken down into different categories, based on several organisational sources, including the National Grid Electricity System Operator's (NESO's) Data Centres report (PDF, 247KB) and the House of Commons Library's Data centres: planning policy, sustainability and resilience research briefing. These categories are:

• enterprise – owned and used by a single organisation

• co-location – typically owned by specialist operators but used by multiple tenants who rent space to install their own servers, storage and networking equipment; the data centre operator maintains the facility and provides power, cooling, connectivity and security

• hyperscale – large, purpose-built campuses, often owned and operated by a single global cloud provider and often located outside city centres because of land and power constraints

• AI – optimised for the high-performance computing needs of AI but not exclusive providers

The UK data centre industry has expanded substantially over the past two decades. There are no regularly updated official data sources available on data centres. However, the Department for Energy, Security and Net Zero (DESNZ) has produced a narrow but consistent time series of operational data centres, as outlined in their Energy Trends: June 2026 special feature article on data centre electricity consumption.

DESNZ's analysis is based on matching meter-level electricity consumption data to data centres serving external organisations. These include co-location, managed hosting and hyperscale facilities, consistent with definitions used by other organisations like the NESO. While these estimates exclude enterprise data centres, they provide a useful indication of data centre expansion in the UK. They indicate that data centres increased from 221 to 239 – an increase of 18 data centres from 2020 to 2024.

Looking further back, research from Oxford Economics estimates that the number of data centres increased from 52 in 2000 to 266 in 2024, as outlined in their blog on the UK's data centre boom. Much of the recent expansion in UK data centres has been caused by growth in hyperscale facilities, according to private sector research from CBRE UK. These facilities are often developed either directly by major cloud providers or by specialist co-location operators building capacity for cloud computing and AI workloads.

The growth of data centres in the UK is also reflected in their electricity consumption. Data centres consume an estimated 4.5 terawatt-hours (TWh), which is 2% of the 249.2 TWh1 of electricity consumed from the grid in Great Britain in 2024, according to DESNZ's statistics on data centre electricity consumption (PDF, 210 KB). This reflects a 1.3 TWh (41%) increase between 2020 and 2024. This increase could also reflect expanding capacity of existing data centres.

As AI and cloud use grow further, data centre and related energy demands are expected to continue in coming years, though factors, such as planning constraints, available energy supply and housing demands, could affect that growth. More information on the expected future of data centres can be found in Barbour ABI's The future of data centres in the UK: continued growth, sustainability and regional shifts blog.

Main components of a data centre and their treatment in economic statistics

The UK National Accounts provide an integrated framework for describing economic activity and are compiled in line with international standards. They bring together a range of data sources to give a consistent assessment of production, income, expenditure, and wealth across all sectors of the economy.

At the centre of the framework is gross domestic product (GDP), which measures the size of the economy. GDP can be estimated from three perspectives:

• production (value added by industries)

• income (wages and profits generated)

• expenditure (spending on goods and services, including investment and trade)

These approaches are balanced on an annual basis through the Supply and Use Tables, to produce a single, consistent set of estimates on a detailed product and industry basis.

The National Accounts also provide information on both economic flows and stocks. They capture flows, such as consumption, investment and trade over time, and measure assets and wealth through the national balance sheet. These statistics are published through outputs including the Blue Book, the Pink Book, and our regular quarterly and monthly GDP releases.

Data centre components within UK National Accounts outputs

This section outlines the main components of a data centre pre-build, during construction and in operation, and where they are seen within the National Accounts outputs.

Any data stored by the data centre on databases are not currently included in GFCF. Under the System of National Accounts (SNA) 2025, "data" will be recognised as an investment asset, alongside the assets discussed in this article. However, as estimates of investment in data in the UK are not currently available, data are excluded from this analysis.

Legal fees

Examples of this component are:

• fees incurred for acquiring land or establishing ownership

This component is included in "costs of ownership transfer" in gross fixed capital formation (GFCF).

Building the data centre

Examples of this component are:

• data centre construction

This component is included in construction output.

Construction of the building and integral systems

Examples of this component are:

• land preparation, including site clearance

• professional fees, for example, for architects and engineers

• the building structure

• systems that are integral to the building, including electrics, plumbing, heating, ventilation, and water supply

This component is included in "other buildings and structures" in GFCF.

Information and communications technology (ICT) equipment

Examples of this component are:

• servers

• storage devices

• networking and telecommunications equipment

This component is included in "ICT equipment" in GFCF.

Other machinery and equipment

Examples of this component are:

• racks and cabinets, if standalone

• power distribution units

• back-up generators

• monitoring tools and sensors

This component is included in "other machinery and equipment" in GFCF.

Software and databases

Examples of this component are:

• operating software

• cloud platforms

• databases

• AI models

This component is included in "software" in GFCF.

Data centre services

Examples of this component are:

• hosting

• cloud computing

• co-location services

This component is included in output, gross value added (GVA), trade in services, and intermediate consumption.

Electricity and water usage

Examples of this component are:

• electricity used to power servers and other equipment

• water used in cooling systems

This component is included in intermediate consumption.

Wages and salaries

Examples of this component are:

• data centre engineers

• IT technicians

• network specialists

• facilities managers

• security staff

• administration employees

This component is included in compensation of employees (wages, salaries and employer social contributions).

Profits

Examples of this component are:

• operating profits earned from providing hosting, cloud computing, co-location, and related data centre services

• rental of elements, in part or full, of the data centre

This component is included in gross operating surplus of corporations.

Taxes associated with data centre activity

Examples of this component are:

• business rates

• energy taxes

• other production-related taxes

This component is included in taxes on production and products less subsidies.

Data centres and measuring GDP

Data centres contribute to GDP in different ways, at different stages of their lifecycle. During construction and fit-out, spending on buildings, equipment and software is mainly recorded as investment in the expenditure approach to GDP. Once operational, data centres contribute through the digital infrastructure services they provide, such as hosting, cloud computing and data storage, which are captured in the output approach. The wages paid to employees and profits earned by data centre operators are reflected in the income approach.

Data centres can also differ in their ownership structure. Investment associated with a single data centre may be undertaken by more than one business and recorded across multiple industries and asset types within the National Accounts, for example the co-location data centres. This means the various assets could have different economic owners. For example, one business may invest in the physical building and associated cooling and power infrastructure, while another invests in the servers, networking equipment and software housed within it.

Data centres may also be found in extended accounts that are beyond, but aligned, with the National Accounts, such as the environmental accounts. Further detail can be found in the Environmental considerations subsection of Section 3: How data centres appear in economic statstics.

The following sections describe how these components appear in economic statistics throughout the lifecycle of a data centre.

At each stage of a data centre’s establishment, its impact will be seen in our economic statistics, though data centres are not separately visible. Those stages are:

• identification and classification on the Inter Departmental Business Register

• financing for construction and operation

• pre-build

• construction of the data centre

• fitting out the data centre

• once the data centre is operational

• environmental considerations

Identification and classification of a data centre on the Inter Departmental Business Register

Data centres are either owned, constructed, or operated by multi-national companies or UK-based companies. The UK's Inter Departmental Business Register (IDBR) uses information from Value Added Tax (VAT), Pay As You Earn (PAYE), and Companies House records to include and classify UK businesses. This is used as the sampling frame for our surveys, where surveys will then select UK businesses for inclusion in their surveys if they meet the survey sample design criteria.

Typically, businesses above a specified threshold (for example, by employment or turnover) are fully enumerated in a survey. Smaller businesses are sampled, with the survey results weighted to represent those not selected. We supplement our monthly business survey data with VAT turnover for certain industries in short-term gross domestic product (GDP) measures. This data source will also include data centres.

When a business structure is created on the Inter Department Business Register (IDBR), a reporting unit is assigned. Reporting units are the practical units used for collecting business-level data in our surveys, which feed into the National Accounts. Data collected at reporting-unit level are classified, weighted and aggregated, alongside other sources, to contribute to wider industry- and economy-level estimates. A reporting unit may represent a single local unit (individual site) or multiple local units.

Each reporting unit is classified to a Standard Industrial Classification (SIC), based on its dominant activity. Data centres are most associated with UK SIC 2007 Section J (Information and Communication), specifically SIC 63110 (Data processing and hosting activities). However, a reporting unit will only be classified under SIC 63110 if that is its dominant activity. If the dominant activity is not in data processing and hosting activities, the reporting unit may be classified outside SIC 63110, even if the data centres (local units) it reports for are themselves classified to SIC 63110. For example, in the case of a real estate company owning data centres, though the data centres it reports for would themselves be classified to data processing and hosting, the reporting unit would be classified to real estate activity.

Survey respondents supply their data on a reporting-unit basis, so this means that the activity of the data centres would not be directly visible in SIC 63110. Consequently, investment associated with data centres may appear within industries such as information and communication, real estate, financial services, or other industries, depending on the economic owner of the assets involved.

The UK SIC has been updated from SIC 2007 to SIC 2026 to better reflect the modern UK economy. The SIC 2026 framework is not yet in operational use across our systems. The earliest planned use is in the Blue Book 2031 (UK National Accounts). However, this will not improve the visibility of data centres within the National Accounts, as their classification will still depend on the dominant activity of the reporting unit. For more detail, see our UK SIC 2026 revision process article.

Financing for construction and operation

In the UK, the construction and operation of a data centre can be financed in several ways, depending on ownership and corporate structure. A UK data centre operator may use funding from within its wider group, which could include funds from an overseas parent or related company. Alternatively, the UK entity may raise finance directly from domestic or even overseas sources, for example, by issuing debt or equity, or by borrowing from a financial institution.

How these flows are recorded in UK economic statistics depends on the financing arrangement. Within the UK National Accounts, financing flows are recorded in the UK sector accounts, including the financial account and balance sheets, under the relevant institutional sector and financial instrument. These data are brought together in the financial accounts and balance sheets of our UK Economic Accounts dataset.

Funding from overseas sources is recorded in the UK balance of payments and international investment position dataset. Subject to the nature of the relationship and the form of the financial instrument, funding is generally recorded as foreign direct investment, where the investor holds 10% or more of voting power, and portfolio investment, where it is less than 10%.

Pre-build

Before confirmation that investment in the data centre will go ahead, there will be spending by the business on architecture, surveyors' fees, legal costs, and more. If the data centre construction is confirmed, these expenditures will be capitalised and recorded as investment in "Other buildings and structures" under the wider "Other building and structures and transfer costs" within our Gross fixed capital formation (GFCF) – by sector and asset dataset. We can expect the "Information and Communication" industries (Section J) to show much of that investment, but this is dependent on the classification of the business or reporting unit making that investment.

For example, if a real estate company is the economic owner of a data centre, the associated investment would be recorded under that company's dominant activity. This illustrates a wider measurement challenge. Although data centres are often associated with data processing and hosting activities, the industry in which investment appears depends on the industry classification of the economic owner, for example, real estate.

Investment by industry can be seen in our Business investment by industry and asset dataset, though this investment will include investment in all assets, not only those relating to data centres. Business investment is a subcomponent of GFCF, which is whole-economy investment. More detailed analysis of investment in digital assets is available in our Redefining investment in digital infrastructure in the UK: 2026 article.

In practice, knowing expected capital expenditure can provide insights into future spend. We have no source for expected capital expenditure. However, construction industry intelligence provider Barbour ABI noted that there could be over £25 billion in investments over the next five years related to the data processing market in their Data centre construction market UK 2025 to 2029 report.

In terms of capacity, planned developments are expected to add around 6.2 Gigawatts (GW) of IT power capacity by 2030, more than double current levels, according to Oxford Economics's blog on the UK's data centre boom. This indicates a noteworthy increase in computing capacity in new and existing data centres and underlying hardware requirements. It also suggests that growth in demand for artificial intelligence (AI) and cloud-based services is contributing to more concentrated and hardware‑intensive investment in digital infrastructure. It should be noted, however, that proposal to build a data centre does not automatically translate into completion.

Construction of the data centre

When construction of the physical data centre building begins, this is reflected in our monthly Construction output in Great Britain bulletin under the category "private industrial new work", and in our Output in the construction industry dataset. This will include all private industrial new work activity, including, for example, factories and warehouses and not just that related to data centres alone, because they are not separately identifiable.

Any site improvements required before building work commences (such as land clearance or demolition of existing buildings), construction of the data centre building, and installation of electrical, ventilation and cooling systems are recorded as capital expenditure. All this activity appears within the "Other buildings and structures asset" in our Business investment by asset dataset.

Any necessary imports of construction materials for the data centre are captured in our monthly UK trade bulletin and our quarterly Balance of payments, UK bulletin.

Understanding conceptual and methodological differences between the UK's construction output and gross fixed capital formation

• Construction output focuses on construction industry activity, while GFCF focuses on asset ownership (for example, the economic owner and investment across the economy, not just the construction industry).

• Construction output measures the value of construction work carried out, regardless of who ultimately owns the asset, while GFCF records investment by the economic owner of the asset; for data centres, these may not necessarily be the same business.

• Construction output records the value of work done at the time it is carried out, while GFCF records the progress payments made by the economic owner, including when they are transferred; this timing misalignment can cause short-term divergence.

• Construction output estimates cover Great Britain, while GFCF estimates cover the UK, and so include Northern Ireland.

• There are also differences between the seasonal adjustment approaches for construction output and GFCF, which affects quarterly data and the deflators used to create the respective volume series.

Fitting out the data centre

Installation of the computing equipment required to run the data centre includes items such as servers, data storage, networking and telecommunications equipment, and items like racks to hold the computer hardware. These items fall within the broad asset groups "Information and Communications Technology (ICT) equipment" and "other machinery and equipment".

The economic owner of these assets may differ from the owner of the building itself. For example, one organisation may own the physical facility and associated cooling and power systems, while another may own the servers, storage equipment, networking equipment, and software housed within it.

If the equipment is leased to the operator under a financial lease arrangement, then it is recorded as investment by the operator. Equipment leased under an operational lease would be recorded by the lessor. Software and databases held under licence by the operator for more than a year would be included within Intellectual Property Products (IPPs). IPPs held for a year or less would be treated as intermediate consumption.

These asset groups appear within our quarterly Business investment by asset dataset. However, these cover all businesses and not just for data centres, which are not separately identifiable. Broad industry-level data are published within our Business investment by industry and asset dataset. Lower-level breakdowns by industry, such as ICT equipment, telecoms, hardware, other machinery and equipment, and software are published in our Annual GFCF by industry and asset dataset (Figure 1). Further breakdowns of digital assets by industry are available in our latest Redefining investment in digital infrastructure in the UK: 2026 article, which also looks to refine what is treated as infrastructure.

As with construction of the data centre building itself, any imported goods used for fitting out of the data centre will be captured in our monthly UK trade bulletin (Figure 2) and our quarterly Balance of payments, UK bulletin. However, these releases will not identify whether those imports are destined for data centres.

Relevant commodities would mainly be:

• CPA 26.2 Computers and peripherals

• CPA 26.3 Communication equipment

Once the data centre is operational

Once construction is complete, the data centre will look to generate income through the provision of services such as data storage and cloud computing. Any income will be captured for relevant industries, including:

• SIC 62 Computer programming, consultancy and related services
• SIC 63 Information services
• Section J Information and Communication
• any other affected industries

Income for these industries will be reported in both our Monthly Business Survey turnover of service industries dataset and our Index of Services, main components and sectors to four decimal places dataset.  

Businesses that build data centres to support their own businesses will not explicitly generate income. However, they should have lower costs and therefore, higher gross value added (GVA) because they do not have to pay another company to use their data centre.

Services exported by the data centre will be recorded according to the service being provided in our UK's Balance of payments dataset as:

• "Intellectual Property"

• "Telecommunication, computer and information services"

• "Other business services"

Any imports of services will also be recorded in our Balance of payments dataset. Further detail is also available in our Trade in services dataset.

Environmental considerations

Alongside many other industrial processes, data centres require electricity and water in the use phase to continually run servers, storage devices, networking equipment, and cooling systems that will be in continual operation.

The United Nations (UN) System of Environmental Economic Accounting (SEEA) follows the principles of the UN System of National Accounts.

The UK publishes data on electricity use in the Environmental Accounts, which are generally available at the two-digit SIC 2007 level. For data centres within SIC 63110 (Data processing and hosting activities), the most detailed level of data would be for SIC 63 (Information services). For businesses/data centres that are classified to an industry other than SIC 63110, these would be included in the relevant two-digit SIC 2007 code, as outlined in the Identification and classification of a data centre on the Inter Departmental Business Register subsection of this section.

Two-digit SIC data are available for both overall energy use and use of fossil fuels in the physical energy flow accounts (PEFA), which record the flows of energy from the environment to the economy, within the economy, and from the economy back to the environment. They are also available for air emissions, including greenhouse gas emissions and intensity, acid rain precursors, heavy metal and other pollutants, in the annual Environmental Accounts.

Intermediate consumption of energy and water by industry is also shown in the intermediate demand tables within the annual Supply use tables, which are also at a two-digit SIC level. In the case of SIC 63110 (Data processing and hosting activities), the lowest level of availability would be SIC 63 (Information services). The same visibility challenges as those described previously would apply, if a business is classified to an industry other than SIC 63110.

Data centre creation and construction phase activities would appear in the UK material flow accounts, which track the physical movement of natural resources and goods between economy, society and environment. These statistics show the materials needed to support goods and services used in the UK, and cover domestic extraction, physical imports and exports. These are available at the economy-wide level, but not by SIC code.

Environmental economy estimates continue to be developed. For example, implementing the updated SIC, which may help support further understanding of resource demand, use and impact of data centres. Additionally, the UN SEEA Central Framework is also currently under review, including its water-related aspects, and this is due to conclude in 2028, followed by national implementation.

Data centre case study

The following example illustrates how a typical data centre would appear across multiple components of the National Accounts framework.

Construction and fit-out

During the construction and fit‑out phase, the construction industry will record its construction as output of that industry in the output approach to measuring GDP. However, the economic owner's investment in the data centre will appear within GFCF in the expenditure approach to measuring GDP. Associated gross fixed capital formation will include:

• spending on the building itself and installation of integral systems

• spending on associated assets, such as machinery, ICT equipment and software, that are expected to be used in production of the data centre's services for more than one year

Estimates for these components are mainly based on a combination of construction statistics and business surveys, covering collecting capital expenditure data.

Some of the computer equipment used in data centres, such as servers and other technology, will likely be imported. As imports are deducted in the calculation of GDP, this can reduce the short-term impact of this investment on measured economic growth. However, this investment still contributes to the UK's capital stock, increasing the stock of assets available for production and supporting economic capacity over the longer term.

Once the data centre is operational

Once operational, data centres will typically contribute to economic output through the services they provide, such as data storage and cloud computing. This activity is recorded within gross value added (GVA) in the production measure of GDP, primarily within the Information and Communications Sector. The income generated from these activities is captured in the income measure of GDP, including profits (gross operating surplus of private non-financial corporations) and wages (compensation of employees).

Where a business has constructed its own data centre, the initial investment would have been captured as GFCF under the expenditure approach and the ongoing operation would appear as an operating cost.

Further detail on how these services are used across the economy, including their inputs such as energy use, is available in our Input-output supply and use tables.

Any major refurbishment to the data centre building to extend its life, increase its capacity, or improve its performance would be recorded as GFCF. Likewise, upgrading of ICT equipment or machinery would also be recorded as GFCF. Routine maintenance and repairs would be treated as intermediate consumption.

Over time, both the data centre building itself, and the computing and other equipment it contains, will depreciate. The data centre building would depreciate in line with the existing asset category "Buildings other than dwellings". Its computer hardware, for example, its servers, would depreciate in the same way as the "Computer hardware asset", inheriting the service life and depreciation assumptions of that existing asset category. For further information on the asset lives by type of asset used in capital stock estimation, please see our National Accounts article: Changes to the capital stock estimation methods for Blue Book 2019.

As investment in artificial intelligence (AI) accelerates, we need to ensure we are in the best position to provide understanding of how data centres contribute to economic activity, productivity growth and changes in labour demand across industries. Measuring these impacts presents challenges, particularly where the benefits and costs of AI will emerge over time or are captured within intangible assets and digital services, rather than traditional forms of capital.

These developments point towards several areas where further measurement and analysis will be required. As part of our Surveys and Economic Statistics Improvement Plan, we are developing an AI thematic account. This will provide one potential framework for tracking AI-related investment, production and use across the economy, while helping to assess the role of supporting infrastructure, such as data centres.

There is also a need to better understand the links between AI adoption, productivity performance and labour market outcomes, including how demand for skills may shift as businesses integrate AI technologies into their operations.

Taken together, these issues suggest a broader direction of travel towards more comprehensive measurement of the economic impacts of AI and the infrastructure that enables it.

Artificial intelligence

Capabilities of a computer program, or system controlled by a computer program, of recognition, reasoning, communication, and prediction that emulates human recognition, reasoning and communication.

Business investment

A component of gross fixed capital formation that reflects investment in fixed assets in the UK. It excludes investment by the government, and in dwellings and transfer cost assets.

Cloud computing

Technology that enables users to access computing resources, such as data storage, software, and related information and communications technology (ICT) services, remotely over a network.

Co-location data centre

Data centre that is typically owned by specialist operators but is used by multiple tenants who rent space to install their own servers, storage and networking equipment.

Data centre

The physical facilities, primarily dedicated buildings, that house computing infrastructure (servers, storage and networking equipment) typically used to deliver the memory and processing power for artificial intelligence (AI), cloud computing, data storage, and server hosting.

Economic owner

The institutional unit entitled to claim the benefits associated with the use of goods, services, natural resources, financial assets, and liabilities in the course of an economic activity, by virtue of accepting the associated risks.

Financial lease

Where the lessor, as legal owner of an asset, passes the economic ownership to the lessee, who then accepts the operating risks and receives the economic benefits from using the asset in a productive activity.

Gross domestic product

Measures the value of goods and services produced in the UK. It estimates the size of and growth in the economy.

Gross value added

The value generated by any unit engaged in production and the contributions of individual sectors or industries to gross domestic product (GDP). It is measured at basic prices, excluding taxes less subsidies on products.

Gross fixed capital formation

Whole economy investment in fixed assets, such as buildings, ICT equipment, machinery, software, dwellings, and transport equipment used in the production of goods and services for more than a year.

Hyperscale data centre

Large, purpose-built data centre campuses.

Inter Departmental Business Register

A comprehensive list of UK businesses used by government for statistical purposes.

Operational lease

A type of lease agreement where the legal owner of an asset is also the economic owner, meaning they retain the operating risks and receive the economic benefits from the asset by using it in a productive activity.

Standard industrial classification

The industrial classification applied to the collection and publication of a wide range of economic statistics.

Thematic account

Provides complementary data that allows for important insights into an activity or aspect of an economy that lacks visibility in the integrated framework of the national accounts.

A thematic account disaggregates and rearranges the items in the integrated framework of the System of National Accounts to provide that visibility, while remaining consistent with the principles of the core national accounts framework.

Thematic accounts were previously included under satellite accounts, alongside what are now referred to as extended accounts.