Data centres and clean power
Energy The Economics of Climate Change

Data centres and clean power

Surging demand in the UK and Ireland poses a challenge for the grid, requiring a mix of solutions.

Hungry for power 

Data centres currently account for almost 2% of total global electricity demand, a proportion that is steadily rising: the IEA expects that this share could double by 2026.   

This demand is fuelled by society producing and consuming data at an exponential rate.  Cryptocurrency, 5G networks, cloud-based services and large-scale AI usage are behind the trend.

Driven by the financial and tech sectors, the UK already hosts more data centres in Europe than any other country except Germany.  They currently account for 2% of Britain’s electricity demand, a figure that is set to increase fourfold by 2030, according to the National Energy System Operator.  In Ireland, data centres play an even more significant role in the electricity system. They are expected to draw 30% of demand by 2030,  up from 21% in 2023 and just 5% in 2015.   

Clean power ambitions and increasing demand from data centres

Both the UK and Ireland aspire to decarbonise electricity. While the projected increase in demand for clean power over the rest of the decade will chiefly be driven by electrification of heating and transport, the explosive growth of data centres clearly presents a challenge to both countries’ ambitions. 

The UK has announced a Clean Power 2030 mission defining the pathway to its legally binding goal of net zero. Aspirations include doubling onshore wind capacity, tripling solar and quadrupling offshore wind by 2030. Ireland’s target is to halve emissions by 2030 and reach net zero no later than 2050. It has set a goal of increasing the share of renewable electricity to 80% by 2030; to that end it has set ambitious targets of generating 9GW from onshore wind projects, 8GW from solar power and at least 5GW from offshore wind.  

Data centres present a locational challenge

The increasing load of data centres, their fairly flat demand profile and the limited willingness of operators to tolerate outages have implications for decarbonisation and grid transformation on the road to net zero. The difficulty of ensuring enough low-carbon supply – already significant – is amplified by uncertainty over the demand for electricity, which will depend on the interaction of a complex set of factors. One of these factors is demand from data centres.

Nevertheless, while it is challenging to do so, policymakers already have tools to deploy low-carbon power at scale, to meet demand at a system level (and are refining these tools to ensure they remain fit for purpose).  Energy efficiency can help reduce the scale of this challenge. Data centres have made strides in improving energy efficiency in recent years and innovative approaches to dealing with data waste could help further.  

A significant challenge that data centres present has to do with their location: they are often built in places where the grid has not been planned for large loads. Expanding the grid to integrate data centres adds to the difficulty of building a network capable of accommodating increased electrification and the growing capacity of the variable renewable energy sources needed to achieve net zero. 

The challenge arises in part since impacts on the electricity system are only one of a range of factors operators need to consider in choosing where to locate a data centre:

        Power infrastructure: Data centres require secure, reliable power. Locations with a significant share of renewable energy are attractive for companies motivated by corporate social responsibility (CSR) goals. In an attempt to incentivise data centre operators to choose more efficient sites, Ireland has made a commitment as part of its Climate Action Plan to develop a framework for large energy users to report their emissions in greater detail. On the whole, the major operators have been supportive of this initiative. Some, including Google, have already established procedures for reporting on electricity usage and associated emissions from their data centres.

        Proximity to users: Particularly for cloud computing and real-time communication services where the ability to retrieve data instantaneously is paramount. Being close to major metropolitan areas and tech hubs enables access to high capacity / high interconnectivity / low latency data transmission networks. This is evident in the clustering of data centres around more densely populated areas, such as Dublin and south-east England. In Dublin data centres have placed further strain on an already highly congested part of the grid. Ireland’s Enterprise Strategy in 2022 acknowledged the ‘indispensable role of data centres in the economy and society’ but highlighted the need for infrastructure upgrades to overcome the limited capacity for further data centre development.

        Local conditions that support investment: Streamlined regulatory processes and attractiveness to investors are important considerations.

        District heating networks: In Denmark, for example, data centres are able to sell waste heat to local utilities.

The challenge may, however, be exacerbated by inadequate incentives and processes to ensure that data centre growth is aligned with impacts on the energy system. 

Ways to support growth and meet climate objectives

If designed appropriately and if operators are given sufficient incentives, existing data centres can offer a degree of flexibility that can help mitigate their impact on grid congestion. For example, data centres can be incentivised to shift non-critical tasks (e.g. software updates, AI model training) to a time where grid conditions are more favourable. Data centres may also feature behind-the-meter generation and backup energy storage that can provide ancillary services. For example, Microsoft’s data centre in Dublin is able to use its battery capacity to provide fast-frequency responses to the grid.  

It will also be important to consider how future data centre expansion can be enabled, while minimising impacts on the system. When planning new data centres, it should be possible to influence their location and the amount of energy demand flexibility (e.g. in relation to cooling), built in from the start. But it is also possible to influence grid plans to accommodate new loads. 

The challenge of securing renewable energy supplies in specific locations has led some regulators to encourage, and even mandate, self-sufficiency as part of a long-term strategy.  

In Ireland, a large energy users’ connection policy requires operators that are adding a new heavy load (such as a data centre) to ensure their own power generation.  This may be viable as a short-term resort, if power constraints are extreme and a project can command significant private investment. However, it is a solution that sacrifices efficiency by failing to take full advantage of the cost savings from being part of a wider interconnected system. 

In the longer term, the trade-offs between grid impacts and broader factors should be managed more efficiently. In principle, exposing investors to whole-system costs and benefits can inform their decisions on where to locate new data centres and how much flexibility to incorporate into their design. Alternatively, at the other end of the spectrum, investment choices could be informed by a central energy system planning approach. Potentially, given data centre location depends in part on regulatory decisions in other sectors (such as access to high capacity communication networks), strategic planning could extend beyond the energy sector. 

In most cases, and in particular given the long lead times involved in planning and developing new grid infrastructure, managing the trade-offs will require a combination of the two approaches. This will require increasingly close partnership between energy system planners and grid operators, telecoms network operators and data centre operators, to understand the evolution of data centre demand in specific locations and to find the solutions that help keep costs to a minimum.