News & media The Hot Topic of Data Centres – Data Centres Demystified Series
A series exploring what exactly data centres are, their different types, what challenges we face as they scale and how to overcome them. What is fact and what is fiction when it comes to their environmental cost.
Article 1: Data Centres Demystified: The Infrastructure Powering our Digital Lives
Data centres are a hot topic, from the scale and growth of the sector to concerns around the environmental cost and power required to run more of them. 2025 reports place UK data centre electricity usage at about 2% of total demand and by the end of 2025, global data centres consumption is projected to reach as much as 23 gigawatts of power, twice the energy consumed by the Netherlands[1]. But what do we mean by a data centre? What are the challenges they pose and what needs debunking? Are there ways we can we use them responsibly, scaling with sustainability in mind? These are the questions this series sets out to answer.
From ENIAC to the Cloud: A brief history
Data centres are the physical backbone of our digital world. They are specialised buildings where vast numbers of computers, storage devices, and networking systems run the online services we use every day, and they’re not new. Data centres first emerged in the 1940s, with the first being established at the University of Pennsylvania in 1945 to house the Electronic Numerical Integrator And Computer (ENIAC) – the first programmable, general-purpose electronic digital computer[2]. By the 1980s, microcomputers took up less physical space, but the need for centralised management of computer systems drove the development of data centres. Demand then increased dramatically during the internet boom in 1990s and 2000s. From there, it was cloud computing that revolutionised the data centre landscape and led to the development of large-scale hyperscale centres.
Today, with the integration of mobile devices, social media, streaming services, AI and machine learning, more data is being created and consumed globally than ever before. Demand is the highest it has ever been and is still growing.
The different types of data centres
The term “data centres” is an umbrella term for many different types, most commonly colocation, enterprise or on-premise and hyperscale data centres. Colocation centres rent out space, power, cooling, and security to multiple customers, who bring and manage their own servers. These can be retail facilities hosting dozens or hundreds of clients in a shared environment, or wholesale sites leasing entire halls or buildings to a small number of large customers.
Enterprise or on-premise data centres, are owned and operated by a single organisation for its own needs, while hyperscale data centres run by companies like Amazon, Google, or Microsoft, power massive global platforms and cloud services.
Although they all provide the same basic function – to keep IT equipment operating securely and continuously – their scale, ownership model, and technical requirements vary widely, shaping everything from their energy use to their efficiency.
How data centres work: Power, cooling, and security
At the simplest level, every data centre has the same basic job, to keep IT systems, servers, storage, and networking running continuously, securely, and within precise environmental limits. That means providing a constant and reliable electricity supply, robust cooling systems to manage the large amounts of heat generated by equipment, and layers of physical and digital security.
This infrastructure needs to run 24/7, and is supported by backup generators, battery systems, and advanced monitoring to ensure uptime. Power Usage Effectiveness (PUE) is the ratio of the total facility’s energy use to the energy actually consumed by the IT equipment inside. It is one of the industry’s core sustainability KPIs, recognised by industry-led initiatives such as the Climate Neutral Data Centre Pact[3] or the EU Commission’s Delegated Regulation (EU) 2024/1364 on the first phase of the establishment of a common Union rating scheme for data centres[4]. A lower PUE means a larger share of the electricity is going directly to the IT equipment itself, with less needed on overheads like cooling and power distribution losses.
Who holds responsibility? Operators vs. Tenants
The way these functions are organised, and who owns what, varies sharply between data centre types. In a colocation facility, the operator owns and manages the building, power supply, cooling systems, and security, but not always the servers, which usually belong to customers. A single colocation site can host hundreds of tenants, each with their own IT equipment, workloads, and security policies. This creates a split in operational control: the operator can improve infrastructure efficiency but cannot dictate how efficiently each tenant’s server is. That limits the operator’s influence over the total energy and water footprint – for example, ICT hardware able to operate at higher temperatures can reduce the energy requirements for cooling, while older hardware can require more.
By contrast, enterprise data centres and hyperscale data centres are owned and run entirely by the company they support. Here, the same entity manages both infrastructure and IT, making it easier to coordinate efficiency measures across the whole operation. The important difference is that an enterprise data centre is a private, smaller-scale facility built to meet an organisation’s specific needs, while hyperscale facilities are run by global cloud giants such as Amazon, Google, and Microsoft. These massive sites host tens of thousands of servers and are designed for maximum efficiency at scale, often with in-house engineering teams for cloud computing platforms and big data storage. Often the term “data centre” evokes an image of hyperscale facilities alone, which is unsurprising given by the end of Q1 2025, hyperscale operators accounted for 44% of global data centre capacity, with 1,189 large facilities in operation[5]. However, in reality, different types of data centres are used and governed very differently.
Measuring efficiency
Sustainability performance is increasingly tracked not just by PUE but by other KPIs such as Water Usage Effectiveness (WUE), which measures litres of water consumed per kilowatt-hour of IT load, and Renewable Energy Factor (REF), which shows the proportion of energy coming from renewable sources3. These metrics help highlight the trade-offs between efficiency and resource use: for example, some cooling methods can improve PUE but increase water consumption, so WUE is needed to keep the picture balanced3.
This complexity means that in certain models, especially colocation, the operator’s ability to meet environmental, social, and governance (ESG) targets can be limited by factors outside their control. A colocation provider may invest heavily in renewable power (improving REF) and advanced cooling (lowering PUE) but still see suboptimal results if tenants’ servers are underutilised or run inefficient workloads.
For policymakers, investors and customers alike, understanding where responsibility lies and where it doesn’t, is key to setting fair expectations and designing effective sustainability strategies.
In Article 2, we’ll explore what challenges data centres truly present.
[1]https://assets.raconteur.net/assets/r/pdf/EAI_web.pdf?_gl=1*1bhhs9j*_gcl_au*MTgxNTM5OTA3NC4xNzUwMzQyMTU2
[2] https://penntoday.upenn.edu/news/worlds-first-general-purpose-computer-turns-75
[3] https://www.climateneutraldatacentre.net/wp-content/uploads/2025/04/CNDCP-A-Rating-Scheme-for-Data-Centres.pdf
[4] https://eur-lex.europa.eu/eli/reg_del/2024/1364/oj/eng
[5] https://www.datacenterknowledge.com/hyperscalers/hyperscalers-will-command-60-of-global-data-center-capacity-by-2030-report