An A-Z guide to data centre terminology

A

Adiabatic cooling - A cooling system that uses a change in pressure to cool air, which is then used to lower the temperature of the IT equipment in the data centre. Importantly, adiabatic systems avoid conventional refrigerants and use less water than more traditional industrial systems such as cooling towers. They typically work by drawing warm air over pads containing water, which then evaporates, cooling the air. 

AI data centre - A data centre that focuses on supporting AI processes. These data centres will be designed to house the specialised IT equipment required to run and train AI models. 

See also: neocloud

AIGZ or AI Growth Zone - AIGZs were introduced in response to a recommendation in the UK Government's AI Opportunities Action Plan. They are designed to facilitate the "accelerated build out of AI data centres". The first AIGZ will be at Culham, Oxfordshire. As explained in section one of this briefing, the designation process requires local authorities to identify sites with adequate power, water and land availability together with strong digital connectivity, sufficient to support 500MW of AI Infrastructure. They are also expected to have existing planning consent or be able to demonstrate a viable pathway to full consent by 2028.  As such, the designation process should address many of the potential barriers to data centre development "upfront".  There has also been discussion of streamlined planning processes for AIGZs, but this would be likely to require additional measures.

C

CCA or Climate Change Agreements - A government scheme administered by the Environment Agency. CCAs are agreements between companies and the Environment Agency to reduce energy use and carbon dioxide emissions in return for a discount on the Climate Change Levy (an environmental tax charged on the energy used by a business).

Cloud computing/cloud services - This is an on-demand provision of IT resources, delivered immediately over the internet. Cloud services providers, which are typically hyperscalers, will operate their own data centres, with multiple customers "sharing" the same infrastructure through their use of the cloud services.

CNDCP or Climate Neutral Data Centre Pact - An initiative established in 2021 by data centre operators and trade associations. Parties to the CNDCP have pledged to meet specific sustainability goals, including annual energy efficiency targets, purchase of clean energy, and water usage effectiveness.

CNI or Critical National Infrastructure - Following amendments to the National Planning Policy Framework in December 2024, data centres are now classed as CNI and can also be opted into the Nationally Significant Infrastructure Projects (NSIP) regime, which means that planning applications are determined by the Secretary of State, rather than local planning authorities.

See also: NSIP regime.

Co-hosting data centre - Similar to a co-location data centre, but the third party also owns the IT equipment. Customers purchase storage and processing services from the owner.

See also: co-location data centre.

Co-location data centre - A data centre that is owned and operated by a third party, which will also provide auxiliary infrastructure such as power, cooling and security. Customers will rent space in these data centres to house their own IT equipment. These data centres can be designed so that customers may customise specific details, or come fully fitted with only the customer's equipment not being provided.

D

DCIE or Data Centre Infrastructure Efficiency - A calculation of data centre efficiency determined by dividing the energy used specifically by IT equipment by the energy used by the whole data centre.

Direct to chip cooling - A cooling system that directly targets the equipment that needs to be cooled by circulating a liquid coolant or water to cold plates that are attached directly to the equipment. This method allows the overall system to run at a higher temperature, therefore reducing the amount of energy used on cooling.

E

Edge data centre - A data centre that is smaller and located closer to end users, on the periphery of the network, to minimise latency. As AI is adopted more widely, it is expected that edge data centres will become more important to support the amount of data processing being carried out.

See also: latency.

ERF or Energy Reuse Factor - An indicator that measures how effectively a data centre recycles and reuses its waste energy measured by dividing the total amount of waste heat reused by the total energy consumption of the data centre.

See also: PUE or Power Usage Efficiency, REF or Renewable Energy Factor and WUE or Water Usage Efficiency

Enterprise data centre - A data centre that is owned and operated by a specific organisation for its own internal purposes. Being both the owner and operator allows the organisation to retain full control over the infrastructure, data and security.

EU CoC BP or European Code of Conduct for Data Centre Energy Efficiency Best Practices - The EU CoC BP was first released on 23 October 2023 by the EU Joint Research Centre (JRC). It provides a voluntary framework of best practices updated annually by industry experts. The CoC has evolved from a voluntary initiative into a key reference for regulatory compliance, including under the EU Energy Efficiency Directive (EU EED). The UK's Department for Environment, Food and Rural Affairs have publicly noted support of the EU CoC BP and have explained that government procurement requirements increasingly require either participation in the code or to meet specified code best practices. Efficiency measures set out in Climate Change Agreements were also aligned with the EU CoC BP.

See also: EU EED or EU Energy Efficiency Directive and Climate Change Agreements.

EU EED or EU Energy Efficiency Directive - Directive (EU) 2023/1791 of the European Parliament and of the Council of 13 September 2023 on energy efficiency and amending Regulation (EU) 2023/955 (recast) which came into force on 10 October 2023. The EU EED contains a number of national and private-company level energy efficiency measures, and includes separate reporting obligations for owners and operators of data centres in the EU (including non-EU based owners/ operators) with over 500kW of installed capacity to report certain key performance indicators (including PUE, ERF, REF and WUE) by 15 May each year.

See also: PUE or Power Usage Efficiency, ERF or Energy Reuse Factor, REF or Renewable Energy Factor and WUE or Water Usage Efficiency.

GPU or graphics processing unit - A computer chip that performs multiple calculations rapidly, widely used for high-powered AI processing.

G

GPUaaS or GPU-as-a-service - GPUaaS providers store GPUs (graphic processing units) in data centres and allow customers (which include AI developers) to access their processing power over the cloud instead of having to invest in the physical technology themselves.

See also: neocloud.

H

Heat network - A heat network (sometimes also referred to as "district heating") is a system of distributing heat generated from a centralised location to residential and commercial buildings through a network of insulated pipes, using heat generated from a low carbon source.  Waste heat from data centres can be used as a source, which can in turn assist with meeting sustainability objectives. An example would be the development at Old Oak and Park Royal in London, where waste heat from a data centre will be used to heat over 9000 homes.  For more detail, see our short guide to heat networks.

Heat rejection system - A method of removing, or "rejecting", the heat from a heat transfer system to an environment outside of the data centre. When paired with a heat transfer system, a complete data centre cooling system is formed.

Heat transfer system - A method of transferring heat away from IT equipment in a data centre. Heat transfer can be achieved using water or air. When paired with a heat rejection system, a complete data centre cooling system is formed.

Hyperscale data centre - A data centre of a much larger scale than a traditional data centre, both in physical size and computing capacity. Hyperscale data centres can cover millions of square feet of space, and store thousands of servers in additional to necessary auxiliary equipment. They have the power to provide cloud services at a massive, global scale, as well as support AI workloads.

Hyperscaler - A company that provides scalable cloud services by utilising the infrastructure in hyperscale data centres. Examples of hyperscalers include Amazon Web Services, Microsoft Azure, and Google Cloud.

I

In-house data centre - See enterprise data centre.

L

Latency - The delay between sending an instruction on a computer and receiving a response, caused by the time taken for the data to travel across a network and back again. Data centre operators are interested in reducing latency, thus resulting in a faster and smoother experience for end users. End users will often experience lower latency from a data centre that is closer to them, compared to one that is further away.

See also: edge data centre, ping

LPA or Local Planning Authority - LPAs are responsible for making planning decisions in the UK. This includes granting planning permission, which allows a data centre to be built. However, under the National Planning Policy Framework (NPPF), data centres are now classed as Critical National Infrastructure, and can therefore can be opted into the National Significant Infrastructure Projects regime and determined by the UK Secretary of State at national level.

See also: NPPF, CNI, NSIP

M

Managed infrastructure data centre - See co-hosting data centre.

N

NCSC or National Cybersecurity Security Centre - The UK's technical authority on cyber security, providing guidance to the public and private sectors.

NDCA or National Data Centres Academy - A training provider that delivers training on data centre skills, sustainable IT and energy-efficient data centre operations.

Neocloud - A GPUaaS provider that specialises in providing cloud-based access to GPUs for AI workloads. This focus on AI is what distinguishes neoclouds from other cloud service providers like hyperscalers, which may also offer GPUaaS as part of their wider offering

See also: GPUaaS

NESO or National Energy System Operator - Data centres wishing to connect to the UK electricity grid (as opposed to utilizing an existing connection) need to apply to the National Energy System Operator (NESO) for approval, as do suppliers of power, such as new wind or solar farms. Such suppliers are also key to data centre development, because most UK data centres will be looking to put Power Purchase Agreements in place; it follows that if such suppliers cannot access the grid, data centre projects risk being left without power sources that they can connect to. Historically, grid connections worked on a first come, first served basis, but with insufficient capacity to deal with all requests, NESO is now looking to give priority to projects for which there is strategic need and which are ready to use the grid.

NIS regime - Refers to the UK Network and Information Systems Regulations 2018, which requires in-scope UK-based businesses (including some data centres) to adopt appropriate security measures to protect against cyber threats. These measures include monitoring, auditing and testing, together with specific procedures to report and respond to security breaches.  The UK Government is planning to reform the regime via the Cyber Security and Resilience Bill, which will bring more UK data centres into scope (recognising their role as a form of critical national infrastructure).

NPPF or National Planning Policy Framework - The National Planning Policy Framework provides guidance to planning authorities in England (Scotland, Wales and Northern Ireland have their own equivalent policies). Whilst it does not have statutory force, planning authorities are required to consider it and in practice, it has considerable influence on planning decisions. In December 2024, the NPPF was adjusted to require local authorities to consider the need for data centres when setting local policies and deciding planning applications. Data centres are now classed as Critical National Infrastructure, and can also be opted into the Nationally Significant Infrastructure Projects regime, which means that planning applications are determined by the Secretary of State, rather than local planning authorities. For more detail, see our briefing on the changes to the NPPF.

NPSA or National Protective Security Agency - The UK Government's national technical authority for physical and personnel protective security (as distinct from cyber security, for example). It is part of MI5, the UK's internal security service. It has worked with the National Cyber Security Centre (NCSC) to produce guidance for data centre owners and users on data centre security.

See also: NCSC

NSIP regime or Nationally Significant Infrastructure Projects regime - Refers to an alternative UK planning application process introduced by the Planning Act 2008 that applies to Nationally Significant Infrastructure Projects (NSIPs), which are developments in the energy, transport, water, waste and data centre sectors that meet specified criteria. The purpose of the regime is to streamline the planning process for large projects of national importance. Broadly, developers of NSIPs do not need to apply for planning permission, but they must apply to the Planning Inspectorate for a development consent order in line with published National Policy Statements. However, data centre developers may choose to go through the local planning process instead.

See also: NPPF.

O

On-chip cooling - See direct to chip cooling.

On-premise data centre - See enterprise data centre.

P

Ping - Ping stands for Packet InterNet Groper, which is a program that can be used to test a network's latency. It works by sending an Internet Control Message Protocol echo request to another device on the network and measuring the time taken for the device to send a response.

See also: latency.

PPA or Power Purchase Agreement - A long-term agreement (typically 5-20+ years) between a major energy user (such as a data centre) and an energy provider (usually of electricity from renewable sources, such as wind or solar). The user usually agrees to purchase a set amount of electricity, at an agreed price, from a specific renewable source. This helps gives users long term certainty over price and source of supply, whilst giving energy providers the confidence that they will be able to make a long-term return on their investment in renewables.

See also: temporal matching, volumetric matching.

PUE or Power Usage Effectiveness - An indicator of a data centre's energy efficiency, calculated by dividing total energy use by the amount of energy used by IT equipment only. The closer a data centre's PUE is to 1, the more efficient it is.

R

REF or Renewable Energy Factor - Another indicator of a data centre's energy efficiency, calculated by dividing the total energy use of the data centre by the amount of renewable energy consumed (being all of the renewable energy owned and controlled by a data centre – e.g. energy generated on site, purchased under renewable certificates or otherwise certified by the utilities provider as renewable). A REF of 1 would mean that all energy consumption was from renewable energy sources.

See also: PUE or Power Usage Efficiency, ERF or Energy Reuse Factor and WUE or Water Usage Efficiency

REMA or Review of Electricity Market Arrangements - A review launched in 2022 to ensure that the UK's electricity market arrangements support a low carbon, cost-effective and secure electricity system fit for the future. Whilst the review remains ongoing, the Government has decided against zonal pricing (e.g. potentially cheaper electricity for users closer to significant sources of supply) and to retain a national wholesale pricing regime. Other key elements of the review include: (i) the Strategic Spatial Energy Plan, expected in 2026, which aims to map optimal zones for electricity and hydrogen infrastructure; and (ii) reforms to the Transmission Network Use of System (TNUoS) charges, which are the fees paid to cover the costs of maintaining and upgrading the UK's electricity grid.

S

Scalability - In the context of data centres, this is a data centre's ability to efficiently handle fluctuating demands and increased/decreased workloads without disrupting performance.

SDDC or Software Defined Data Centre - A service that provides virtualised versions of data centre resources. Instead of having physical equipment for networking, storage and computing, these services are provided and managed using software.

SMR or Small Modular Reactor - A nuclear reactor with a capacity under 300 MW. This is much smaller than conventional nuclear reactors, and can therefore be built at a lower cost. SMRs could be used to power data centres and would offer advantages over renewables such as solar and wind in terms of reliability of supply (whereas the amount of power available from solar and wind will vary).  However, the technology has not yet been shown to be commercially viable. The UK's SMR programme is being led by state-owned Great British Energy – Nuclear (GBE-N), working with Rolls Royce SMR as its preferred partner.  Wylfa on Anglesey in Wales has been confirmed as the site for the UK's first SMR.  GBE-N is looking to identify additional sites and is expected to report on potential locations later in 2026.

T

Temporal matching - A term that is sometimes used in Power Purchase Agreements (PPAs) in relation to energy obtained from renewable sources.  One of the key aims of a PPA (see above) is usually to enable the user to buy an amount of electricity from a renewable source that matches its energy usage. Whilst this can be measured over a period of say, one year (see volumetric matching below), critics of that approach argue that temporal matching – where the user purchases enough renewable energy to match its hourly consumption – is preferable. This is because it more accurately reflects the impact of variable output from renewable sources such as solar or wind. A PPA based on temporal matching will therefore typically also include energy storage technologies, such as batteries, and/or the ability to draw power from less variable low carbon power sources, such as nuclear, geothermal or tidal power. The downside from a user perspective is that temporal matching PPAs are typically more expensive, by around £10 per megawatt hour, as compared with PPAs based on volumetric matching.

See also: PPA, volumetric matching.

V

Volumetric matching - A term that is sometimes used in Power Purchase Agreements (PPAs) in relation to energy obtained from renewable sources.  One of the key aims of a PPA (see above) is usually to enable the user to buy an amount of electricity from a renewable source that matches its energy usage over a given period, such as year. If the figures align, the user can be said to have "matched" 100% of its power needs with clean energy sources on a volumetric basis. However, critics point out that the output of many renewable sources, such as wind or solar, is variable, whereas data centres usually need a consistent supply – and that the 100% figure is arguably misleading (in the sense that a data centre is likely to need to take power from other sources during periods when output from renewables fails to match demand).  The same critics often argue that temporal matching (see above) is preferable (although it is also normally more expensive).

See also: PPA, temporal matching.

W

Wholesale data centre - See co-location data centre.

WUE or water usage effectiveness - Water is used in a data centre for cooling, and WUE is a measure of how efficient the data centre is in its consumption. WUE will show the amount of water used per unit of IT equipment power consumption. Signatories of the Climate Neutral Data Centre Pact (CNDCP) have pledged that, by January 2025, new data centres at full capacity in cool climates that use potable water will be designed to have a maximum WUE of 0.4 litres per kilowatt hour in area with water stress.

See also: CNDCP, PUE or Power Usage Efficiency, ERF or Energy Reuse Factor, REF or Renewable Energy Factor and WUE or Water Usage Efficiency

For more on data centres, see:

• Making data centres happen: what issues should real estate investors bear in mind when considering investing in data centre assets?
• Infrastructure and Energy Spotlight – Autumn/Winter 2025 – see section 1