Written submission from Friends of the Earth (England, Wales and Northern Ireland) (DCU0034)

Risks and opportunities to the sustainability of data centres in the UK

Response to the Environmental Audit Committee
April 2026

About us

1. Friends of the Earth England, Wales and Northern Ireland was established in 1971. We support over 200 local action groups (including a number already engaged in challenging hyperscale data centres in their communities) and are part of an international network of more than 70 national groups, counting over 2 million members and supporters globally.  Our vision is a fairer,  greener,  more just country where every community has the right to live in dignity and in harmony with nature.

2. We support strong environmental standards and the right for communities to have a say in the decisions that affect them. We have previously challenged the government’s climate plans through the courts, resulting in the publication of a new plan in October 2025. We are committed to ensuring the government sets and sticks to its carbon reduction targets, whilst protecting and restoring nature, and does this in a way that is fair.

Summary

3. The UK needs a coherent strategy for digital infrastructure development and data centres. One that is transparent, effectively monitored and aligned with national and global goals to cut emissions and promote equity. This technology is still emergent; there is still an opportunity for the UK to position itself as a global leader in sustainable AI and digital development.

4. However, the current policy is not fit for purpose. In particular, it is unclear how the government intends to square its climate and biodiversity goals with the de-regulatory and insufficiently strategic approach it is taking to data centre roll out. Without clear policy, data centre demand growth is unlikely be compatible with carbon budgets. As a first step, the Government must come forward with its promised National Policy Statement, which will set out the policy framework for decision-making.

5. Data centres must be required to source 100% renewable electricity, with stringent additionality criteria to ensure that only newly added renewable capacity counts. Data centres must not be allowed to compete for renewable energy or water with housing development, or other priority sectors for decarbonisation.

6. Strategic planning is hampered by the lack of consistent reporting from data centre developers and operators about the impact of their activities. There should be clear and mandatory requirements for sustainability reporting, coupled with strict operational requirements to maximise the efficient use of resources, including water and waste heat and to address the growing scourge of electronic waste.

7. Efficiency is only part of the story; we also urgently need an informed conversation about how much compute capacity we really need, and how this should be prioritised next to other sectors.

8. We are also deeply concerned about the current lack of democratic participation, and the potential for costs (both financial and non-monetary) to be passed to ordinary citizens, while the benefits primarily accrue to a small number of big tech multinationals.

Environmental impacts and climate goals

Question 2: What are the environmental impacts of different types of data centre currently having in the UK and what are the future impacts likely to be?

9. We set out our views on the current and future impacts of data centres in two reports published in 2025: ‘Harnessing AI for Environmental Justice: Principles and practices to guide climate justice and digital rights campaigners in the responsible use of AI’[1] and ‘Data centres - are our heads in the cloud?’[2] We argue that there are environmental impacts throughout the data centre lifecycle.[3]

These include:

      Impacts of mineral and metal mining and processing in manufacturing components;

      Impacts associated with construction materials (such as concrete and steel);

      Impacts of construction on green space and local environments, in particular impacts of construction traffic;

      Loss of green space and biodiversity impacts for construction in the green belt (for example around London);

      Energy use - both direct and indirect impacts (including the consequences of large new demand source in the context of a wider decarbonisation trajectory).

      Water impacts – for use in cooling.

Generation of high volumes of e-waste.

10. The lack of transparency in the sector and the absence of mandatory reporting makes it very hard to quantify these impacts with confidence. This is exacerbated for future impacts by the lack of strategic planning.

11. To take the example of water use, water companies and the Environment Agency have been warning the Government about pressures on water supplies. However, the 2025 National Framework for Water Resources report,[4] produced by the Environment Agency,  highlights that the lack of accurate information is creating a serious barrier to strategic planning.[5]  

12. Research by TechUK , in collaboration with the UK government, concluded, on the basis of a voluntary survey of 73 sites in England that ‘commercial data centres are not intensive water users.’ 51% of surveyed sites used waterless cooling systems and 64% used less than 10 million litres per year, although 4% used more than 10 times this. Worryingly, 5% of surveyed operators using water in their systems did not measure water use at all.[6]  However, this research is limited to a minority of sites as there are thought to be over 500 operational facilities in England. Total water usage by the sector in the UK is unknown as data centres do not routinely share data.[7] 
13. Analysis of EU-wide data has also found as data centres became larger, they use more water for each unit of power going towards computation. As more ‘hyperscale’ centres are proposed for the UK, this is a growing concern.[8]

14. We are also concerned about fairness, and the distribution of these impacts. For example, there is a risk that rising demand for energy and water may translate into higher bills for customers, many of whom may also be experiencing the consequences of AI-driven automation and job losses.

15. Environmental impacts are often considered through the lens of operational power and water usage. However, it is important to consider the direct and indirect impacts throughout the whole lifecycle and supply chain.

16. For example, data centres rely on a complex international supply chain to provide the inputs for semiconductors and power infrastructure – which themselves incorporate a number of minerals recognised for their criticality. The critical minerals supply chain is far from transparent and data is lacking to support traceability, allow monitoring of associated emissions or environmental impacts, and identify and mitigate the infringement of human rights.[9]  The 2025 UK Critical Minerals Strategy[10] does not directly describe this issue, but mentions several times the need to "promote" and "support" transparent supply chains.

17. In 2022, the UK generated approximately 1.65 million tonnes of electronic waste annually, making it the second-largest e-waste producer per capita in Europe at around 24.5 kg per person per year, according to the UN.[11] Much of this waste continues to be exported, often illegally, for example to West Africa, with knock-on harmful impacts for local people and the environment. The need for constant upgrades to support evolving AI capabilities is further shortening the lifespan of data centre devices and driving faster cycles equipment upgrades and waste. We have found no mention of e-waste in any of the Government’s AI strategy documents.

Questions 3, 4 and 6:

      What impact are data centres having on climate change and the Government’s Net Zero targets and how will this change in the short, medium and long term in the UK?

      To what extent will Artificial Intelligence (AI) accelerate the need for data centres and is this being adequately taken account of by the Government and relevant bodies?

      Has the Climate Change Committee adequately taken account of the impact of data centres, especially in its advice on the Seventh Carbon Budget?
 

18. With regard to operational energy usage and Net Zero targets, DESNZ officials have provided assurances that the Carbon Budget and Growth Delivery Plan (October 2025)[12] did allow for growth of data centres, in line with NESO forecasts.[13]

19. However, we feel that there are discrepancies in the figures used, coupled with a lack of consistency across Government. It may be that there are assumptions being made about the percentage of applications or expressions of interest that will turn out to be speculative. Even if so, there is a need for much more robust and transparent scenario planning. In our view there are grounds for serious concern about the achievability of Net Zero targets in the current AI policy context.

20. The National Energy System Operator (NESO) has made projections for energy use in TWh to 2030, and more speculative projections beyond (to 2050). We assume that the projections considered by DESNZ in developing the Carbon Budget and Growth Delivery Plan are those set out in NESO’s Clean Power Plan to 2030 (i.e. approximately 22 TWh of electricity usage by 2030). Meanwhile, DSIT’s Compute Road Map looks at connection capacity required, measured in GW, as opposed to energy use (measured in TWh). This means we have to look at range of scenarios to understand how the DSIT projections translate into actual energy use and how the two documents compare.[14]

21. The first question is how much of the maximum capacity data centres are using in practice (the ‘load factor’). Figures released by UK Power Networks and analysed by the Green Web Foundation suggest that existing data centres in the UK are currently operating at a load factor of around 25%.[15] However, industry data from the US suggests that, after considering a balance of use types (from training to inference), 70% power usage (load factor) is a reasonable assumption for ‘typical operation’.[16] This range already introduces a high degree of uncertainty.

22. Analysis of Government figures suggests that there is already a significant gap between the NESO projections and the 6GW baseline target capacity set out in DSIT’s compute road map. The table in the annex sets this out in more detail. DSIT’s 6GW figure equates to annual usage of 26.3TWh by 2030, even with an average load of only 50%, rising to 36.8TWh at 70% usage (or a gap of approximately 15TWh). Meanwhile, at the December 2025 meeting of the AI Energy Council, participants were asked to consider demand figures of 8GW and 10GW, without any reference to how these were compatible with clean power plans or the Carbon Budget and Growth Delivery Plan.  Moreover, according to Ofgem, 71 new data centres have already achieved ‘financial commitment’ to invest (i.e. have moved beyond the purely speculative stage). If built, these 71 centres would require around 20GW of electricity.[17] Even at the lowest plausible levels of annual power usage (25% of capacity), this still equates to double NESO’s usage projections.

23. Projections to 2050 are much more speculative. NESO has produced a range of scenarios in 2025, ranging from 30TWh to 71TWh.[18] In July 2025 Lord Vallance told peers that the ‘total amount of electricity use is due to go up from seven to 62TWh by 2050,’ as (briefly) set out in the 2024 Clean Power Plan and suggested that small modular reactors would be needed to meet this need.[19] Given this, it is particularly surprising that the Climate Change Committee has neither provided information in its advice to Government for the period 2038-2042 about which scenarios it has applied, nor chosen to disaggregate electricity demand for data centres from wider commercial demand.

24. The impact of these different scenarios on Net Zero targets depends directly on whether the supply of low carbon energy can keep up with demand and therefore how clean the overall energy supply is, as well as the extent to which the energy demand from data centres slows down decarbonisation in other sectors. Even DESNZ’s record-breaking auction of contracts to offshore wind, announced in January 2026, is only due to create an additional 8.4GW of capacity by 2030.[20] Nearly three quarters of this grid capacity would be absorbed by data centres by 2030 even on the baseline scenario. With regard to emissions, research by Carbon Brief has recently found that DSIT is assuming much lower emissions per unit of electricity than the estimates reported in planning applications for individual data centres made by Google and other companies, meaning that the overall climate impacts may be much higher in reality even in the immediate and less dramatic phase of growth.[21]

25. In short, we believe there are significant discrepancies between the Government’s own scenarios, and no clear and consistent picture of projected energy demand. It is likely that the impacts will be much more significant than the Government’s optimistic modelling suggests, on planned growth trajectories. In addition, the fact that the Climate Change Committee’s advice for the Seventh Carbon Budget does not make explicit reference to data centre growth, and considers commercial energy use only in aggregate makes it very hard to understand both the knock-on impacts for decarbonisation of other sectors and whether the Government is taking proper account of its own estimates for demand growth.

26. AI companies have also frequently made the argument that the benefits of additional AI capacity for the environment, for example for grid optimisation, will outweigh any environmental impacts. Independent research commissioned by a coalition including Friends of the Earth US has found that the overwhelming majority of these claims (74%) are unfounded, and that where there are benefits, these can be attributed to much less energy hungry ‘traditional’ AI (as opposed to generative AI, which is driving the data centre boom).[22]

Question 10: To what extent will the resource demands of data centres impact on other sectors with regard to competition for resources and decarbonisation?

27. It is clear that data centres are in direct competition with other projects and sectors for both raw materials and access to the grid.

28. Many of the resources required for data centres are the same (in terms of raw materials) as those needed to support the green transition, such as copper and electrical steel. In the words of a recent World Economic Forum briefing, ‘every added megawatt [for data centres] embeds tens of tonnes of metals and minerals.’[23] In the context of finite global resources, and seriously constricted supply chains, a boom in data centre construction inevitably comes at the expense of other infrastructure, including, for example electric vehicle manufacturing, or increasing battery storage capacity for renewable installations.

29. The UK Critical Minerals Intelligence Centre designates 34 minerals as ‘critical’.[24] Many of these are both vital for the green transition and in demand by the defence sector and technology companies. At current levels of demand, the world in 2040 is expected to need four times the amount of critical minerals for clean technologies than it does today The government should set out a policy approach which clearly differentiates between those supply chains which actively contribute to the UK’s climate ambitions and those with a use unrelated to net zero or which contribute to unsustainable mineral consumption – and prioritise the former.

30. We additionally note that the current UK Critical Minerals strategy commits to and relies upon an increasingly circular economy to deliver the resource security the UK needs – yet currently the England Circular Economy Strategy remains delayed and unpublished, while four-nation working to implement a cohesive approach to materials reuse and recycling is severely lacking.[25] Friends of the Earth research (available upon request) has found uneven levels of ambition join up between energy, critical mineral, industrial and circular economy strategies across the UK. We recommend that the UK government urgently publish England’s Circular Economy strategy and ensure that, taken together with relevant strategies from the rest of the UK, it provides a clear route to limit resource use and increase circularity in critical raw materials.

31. Data centres are also in competition with other users for new renewable energy. For example, research for Friends of the Earth Ireland by Professor Hannah Daly of University College Cork found that ‘between 2017 and 2023, all additional wind energy generation [in Ireland] was absorbed by data centres. As a result, renewables are not delivering net reductions in fossil fuels use in power generation.’[26]

32. We also know that in 2022, the UK’s Greater London Authority told housing developers in three west London boroughs that new housing projects could be blocked until 2035 , because despite the housing crisis, all the electricity capacity had been allocated to data centres.[27]  Similar experiences have been widely reported in the Netherlands.[28] Ofgem’s consultation on grid connections looks to find a strategic approach to resolving these tensions – but proposes to do so by prioritising viable data centre projects, rather than considering wider social value. [29]

Planning and impact on local environments

Question 5:  To what extent do existing policies, such as the Environmental Improvement Plan and the Planning and Infrastructure Act and associated policies, take account of the potential impact of data centres, particularly in terms of water use, nature and the environment?

33. It is important to note that there is no formal definition of a data centre in UK, nor are operators required to register with a regulator.[30] There is also no definition of a ‘hyperscale’ data centre.  This is a significant policy gap, as it means that there are no clear thresholds for identifying, for example, when more robust scrutiny or impact assessments should be mandatory, or when different planning regimes should apply. What follows should be considered with this gap in mind, and it should be addressed as a matter of urgency through the publication of a National Policy Statement.

34. Data centre applications are also often hard to identify in the early stages of the planning process, as they may be classified as ‘B8 Storage and Distribution’ or ‘sui generis’ (other). A stand-alone use class for data centres would improve transparency, allow more consistent treatment of data centres in the planning system and enable a cumulative picture to emerge.

35. Current  policy frameworks, including the 2025 Environmental Improvement Plan (EIP) and the planning system (notably the current National Planning Policy Framework, as well as recent associated legislative reforms under the Planning and Infrastructure Act 2025), provide broad environmental protections applicable to data centres, but ultimately fail to explicitly address the wider range of impacts.

36. The Environmental Improvement Plan neither recognises site-specific or cumulative pressures on water resources or wider ecosystems, nor provides sector-specific guidance on mitigating their impacts, including the cumulative effects of data centre clustering. However, in this respect their treatment is no different to other types of development antithetical to these objectives (such as oil and gas extraction, or other water-heavy uses). In general, the EIP does not consider threats to its outcomes from specific development types, so this gap, whilst significant, is unsurprising.

37. The lack of a proactive, strategic framework to manage the cumulative environmental effects of data centre development is a particular concern in areas that are already water-stressed. Nor do we see any integration with the Strategic Spatial Energy Planning currently being undertaken by the National Energy Scheme Operator.

38. The wider town-planning framework in England provides two overarching systems to evaluate potential harms:  i) the Town and Country Planning Act 1990 ( TCPA 1990) system and ii) the Nationally Significant Infrastructure Projects (NSIP) system. The TCPA 1990 system is principally dictated by the National Planning Policy Framework (NPPF). These are considered in turn.

Data Centres in the TCPA regime

39. The current NPPF (2024) is limited with regard to data centre specifics, simply requiring local plan makers to ‘make provision’ for data centre use, while decision makers should ‘recognise their locational requirements’. While existing NPPF policies will be relevant to water and environmental impacts, arguably these are more likely to be outweighed moving forward; especially with data centres recognised as Critical National Infrastructure.[31] 

40. The new draft NPPF (published in 2025 and coming into force in summer 2026) provides more  strengthened policy support for data centres, including ‘substantial weight’ in favour of AI Growth Zones and requirements for Local Planning Authorities to allocate specific sites in local plans.[32] Environmental policy considerations are also included, such as those requiring consideration of cumulative effects of pollution on the natural environment.[33] PM13 (Setting Standards) may also allow some flexibility for Local Authorities to tighten up water efficiency standards for certain uses beyond those in Building Regulations in “exceptional circumstances” and “in areas of serious water stress”. There is limited consideration in the new NPPF of cumulative water supply and usage.

41. We are also deeply concerned about the inclusion of Environmental Delivery Plans (EDPs) in the NPPF (as inserted by the Planning and Infrastructure Act 2025). Where an EDP is designated by Natural England - subject to Secretary of State approval - developers can offset impacts to even some of the most protected nature sites by simply paying into a pot (or a ‘Nature Restoration Fund’). [34] This constitutes a grave threat to nature, a position we have held throughout the Planning and Infrastructure Act’s path through Parliament. Although focused on accelerating house-building, the Act does not preclude the use of Environmental Delivery Plans for other development types, such as data centres. There is a high probability that data centres may be proposed within future Environmental Delivery Plans, with potentially very damaging impacts. In our view data centres (in particular ‘hyperscalers’) should be excluded from the EDP framework. 

42. There is also the more detailed but more discretionary Environmental Impact Assessment (EIA) framework in operation through the TCPA Regulations 2017. While EIA involves a more robust assessment of a development’s impact, including in a cumulative sense, it is not mandatory for data centres. This should change to support robust assessment, and data centres above an appropriate size threshold should be added to Schedule 1 of the EIA Regulations 2017 .[35]

43. Similarly, Strategic Environmental Assessment (SEA) or Sustainability Appraisal (SA) of plans also aims to ensure that policies and allocations strike the right balance in reaching positive environmental objectives. This is undertaken before a Local Plan is adopted, and should provide further consideration of data centre allocations and policies. However, with the new NPPF setting out the need to allocate sites we can expect future tensions between ensuring local policies and allocations deliver sustainable development and the national policy expectations.

Data Centres in the  Nationally Significant Infrastructure Projects Regime (NSIP)

44. The NSIP regime was brought in by the Planning Act (2008). Despite promises in Parliament, the Government is yet to produce a draft National Policy Statement (NPS) for data centres, which would set out key policy considerations as well as justify need for such land use.[36] Without the NPS, we cannot comment further on the impacts data centres consented under this system might have in terms of water use, nature and other environmental impacts.

45. We do know from other development contexts (such as roads, power stations and renewables) that Development Consent Orders (DCOs) – the actual permission given – are granted by the relevant Secretary of State, even where the Examining Authority has recommended refusal, including where nature reasons have been provided. Such a principal is likely to remain for data centres under the regime, especially due to their weighting as Critical National Infrastructure, underlining our concerns about a lack of public engagement and democratic oversight.  

46. We also note with concern that the Government has recently allowed a hyperscale data centre application in Buckinghamshire to be directed into the NSIP regime, despite the fact that the National Policy Statement has not even been published in draft form, let alone adopted in Parliament. [37] The Government should publish the NPS urgently, and before any further permissions are granted.

47. We are also concerned that the opportunity for public and stakeholder engagement will be further curtailed, on account of the watering down of the pre-application consultation stage for projects in the NSIP regime. The Planning and Infrastructure Act 2025 – subject to secondary legislation and further, yet unpublished, guidance - has effectively abolished the statutory pre-application period, replacing it with a more ‘proportionate’ approach, subject to ‘statutory guidance’. We await the Government’s response to the recent Consultation on Streamlining Infrastructure Planning to understand the implications for consultation on issues such as nature and water resources for data centres.[38]

48. Our concerns regarding Environmental Delivery Plans (EDPs) and the new ability of developers to pay a levy to bypass protected species licensing regimes also apply to the NSIP regime. Environmental Impact Assessments and Strategic Environmental Assessment are also required for the NSIP regime; due to the scale of projects considered under this regime they are however more likely to be carried out.

      How should the impact of data centres be factored into future policies, such as the Land Use Framework, regional planning, housebuilding and reform of the water sector?

49. The Land Use Framework is not a material consideration for planning decisions; nonetheless, there will be a requirement for future NPPFs to be informed by it. This could provide scope for future NPPF iterations to consider key strategic issues, such as water stress and nature recovery, and to set criteria for councils and combined authorities when allocating land for data centres, in both and site-based contexts.

50. Since 2025, Mayoral and Combined Authorities have had a duty to produce Spatial Development Strategies. This includes the need to identify strategic land allocations, such as land for AI Growth Zones. Well-evidenced strategic planning, subject to robust environmental assessment, will be essential to these strategies. Properly applied, these strategies could be a vehicle to enable much more strategic delivery of data centres and related infrastructure, with due consideration of the cumulative water and wider environmental impacts, than the current ad hoc and piecemeal approach of decision-making by individual planning authorities

      How important is the location of data centres and what factors should be considered for optimum siting of them?

51. Data centres raise a number of key land-use concerns, including water-stress, excessive energy use and greenhouse gas impacts (especially linked to back-up diesel power generation), as well as other issues such as impacts to amenity, the historic environment, outlook, townscape, air pollution etc.  Developers will also make the case for power and water availability, as well as the need for clustering in ‘availability zones’ and ‘latency’ as influencing overall siting. There is a balance of factors to consider. Below is a non-exhaustive list of the environmental factors we think should be considered.

      Water stress. This must be considered cumulatively, across all seasons, and balanced against other pressures (such as house-building).

      Availability of clean power.

      Nature impacts. Data centres should  be sited well away from protected nature and habitat sites of all types, whether local, county, national or international. We have already set out that they should be excluded from Environmental Delivery Plans (EDPs).  All site allocations (local or strategic) should undergo strict Strategic Environmental Assessment and Habitats Regulations Assessments to establish suitability.

      Feasibility of capturing excess heat.  All developments should be required to develop a strategy for the sustainable use of excess heat, and priority should be given to those where there is realistic and achievable possibility of connection to a district heating network or similar amenity project.

      Landscape value.

      Amenity (social) impacts. Data centres have significant potential to impact on local residents in the form of persistent light and noise pollution, construction traffic, overshadowing. At large scale, locations near to existing or future housing are likely to be highly detrimental to residents. [39]

52. Many of these considerations are even more acute for any proposals making use of onsite diesel generators (or potentially gas turbines) for back-up power (neither of which we support).

53. The issues around latency also require greater integration, and more specifically a public conversation about the purpose served by data centres. At the moment a data centre dedicated to serving up personalised advertising via social media, targeting vulnerable individuals for gambling or enabling the rapid generation and storage of deepfake videos is treated in exactly the same way in planning as a data centre whose primary clients will be medical or engineering science researchers, for example. Not all uses have the same social utility – many could indeed even be considered harmful or undermining to our values and institutions – and we need to question the assumption that all should be treated equally in the planning system, or have the same needs with regard to lag times (latency).

Opportunities for mitigation

Question 7: What existing and emerging technologies can be used to minimise the environmental and climate change impact of data centres?

      What specific role can renewable energy play in reducing the carbon footprint of data centres?

54. There is a lot of scope for greenwash in this area, and data centres must not be allowed to slow down progress towards net zero. In order to ensure genuine additionality, new data centres must secure their energy needs through Power Purchase Agreements that explicitly link to new renewable generating capacity and storage. They must not be allowed to compete for Contract for Difference renewables (to ensure they are not being subsidised to divert new renewable capacity away from other sectors). We also agree with the recommendations of the Green Web Foundation that data centres must be required to adopt carbon-aware scheduling to reduce consumption at times of high grid carbon intensity, and to require green energy claims to be validated on at least an hourly rather than annual basis.[40]

55. There are precedents. The Spanish Government has, for example, chosen to respond to the crisis in the Middle East by introducing (amongst other measures) a new sustainability obligation for data centres, requiring any data centre connecting to the grid to balance its new consumption with an equivalent amount of new renewable generation.[41]

Question 8: What opportunities do data centres offer in helping to power and heat local communities and amenities and what will be required to deliver benefits?

56. There are significant opportunities for waste heat capture, including from existing digital infrastructure, but there are currently very few incentives for developers to put the necessary systems in place. Germany has gone furthest in Europe.

57. The Government’s recent Warm Homes Plan, which we welcomed, sets a target of more than doubling the proportion of heat demand met by heat networks in England to 7% (27TWh) by 2035, with heat networks expected to supply around a fifth of all heat by 2050.[42] Industry-led research suggests growth in data centres could produce enough waste heat to warm 3.5 million homes in the UK by 2035.[43] There is a significant opportunity here. There are also well-known examples of waste heat being used to heat swimming pools; UK energy start-up Deep Green is a pioneer here.[44]

58. However, there a big gaps between the potential and the reality of delivery, and there is a lag in the development of heat infrastructure, even in the most ambitious cases. This is why it is essential that heat recovery should be designed into data centre from the outset, and considered in the context of site allocations. For example, Bristol City Leap — a 20-year partnership between Bristol City Council, Ameresco and Vattenfall — is one of the UK's most ambitious city-scale heat decarbonisation programmes. Vattenfall has committed to investing £475 million by 2030 to expand Bristol's heat network, with a goal of supplying the equivalent of 12,000 homes and has agreed a partnership with the University of Bristol's Enterprise Campus to export heat from the university's new quantum computer to the Bristol heat network.[45] However, elsewhere in the city (but in a different local authority area) the Isambard-AI supercomputer (the UK's most powerful, and one of the world’s most sustainable by other measures) is not connected, and the reuse of the heat remains a future aspiration without a clear timeline, despite significant housing growth in the immediate vicinity.[46] This example highlights the challenges even where many of the enabling conditions (such as a values-led developer and established infrastructure partnerships) are in place.

59. Germany has gone furthest in Europe in regulating for waste heat capture. Germany's Energy Efficiency Act, which came into force at the end of 2023, introduces binding requirements specifically for data centres. New data centres opening from July 2026 must achieve an Energy Reuse Factor of at least 10%, rising to 15% from July 2027 and 20% from July 2028, meaning operators must demonstrate that a proportion of their waste heat is being actively reused, whether on-site or fed into district heating networks. Operators are also required to provide detailed data on the waste heat they produce to a national register, and to make that information available to potential heat network operators on request.[47] There are alternatives to regulation (for example, Denmark has used a system of tax incentives); we cannot comment on their relative effectiveness.

Lesson from outside the UK

Question 9: Are there beneficial or precautionary lessons to learn from the impact of data centres outside the UK?

60. Friends of the Earth (England Wales and Northern Ireland) is part of international federation and as such we are well placed to draw on examples from our sister organisations around the world. The experience in Ireland is particularly instructive, as a case study in the cumulative impacts of unmanaged growth. Ireland has become a hub for data centres largely due to favourable tax incentives, a skilled workforce, and its strategic position for transatlantic data connectivity. Friends of the Earth Ireland has been campaigning on this issue for several years. There are concerning parallels between the UK and Irish Government’s approach to data centre development, particularly with regard to promoting the benefits to investors of a deregulatory environment.

61. In 2023, data centres accounted for 21% of Ireland's electricity consumption. As noted above, data centres absorbed 100% of Ireland’s additional wind generation capacity between 2017 and 2023.[48] As a consequence, Ireland’s progress towards decarbonisation is stagnating. There are significant knock-on implications for fuel poverty; in the same areas where Meta has bought the entire electricity output of solar farms,[49] over 60% of homes remain reliant on heating oil and solid fuels such as peat or coal.[50] Moreover, network constraints mean that data centres in Ireland are actually increasing fossil fuel consumption, through the use of on-site gas generation.[51] Friends of the Earth Ireland is currently taking legal action against Ireland’s energy regulator, alongside Friends of the Irish Environment and Client Earth, over its decision to allow fossil fuels to be used to power data centres.[52]
62. Despite the Irish example, there are beneficial lessons to learn from the EU’s approach, at least on paper.

63. In 2023 the Energy Efficiency Directive (EED) was introduced in Europe. This requires data centre operators (above a given size) to report on a number of relevant metrics. Broadly, this covers things like: annual power and water usage, efficiency, and how much of their power comes from renewable sources.[53] Member states must ensure that data centres with a total energy supply above 1 MW use waste heat or other recovery applications (unless it can be shown that it is ‘not technically or economically feasible’). The Commission is preparing a Data Centre Energy Efficiency Package planned for adoption later this year, which will introduce a rating scheme for data centres in Europe and launch work on minimum performance standards.[54]

64. There are issues with implementation and enforcement. The Directive has not been fully transposed in several countries, including Ireland, and there are still very low compliance rates, with limited or no consequences for non-compliance.[55] Nevertheless, the ambition of the Directive stands in marked contrast to the UK Government’s refusal to sign the 2025 Paris AI summit declaration on ‘Inclusive and Sustainable Artificial Intelligence for People and Planet,’[56] the only major attendee alongside the USA.[57]
    
    ENDS

For more information please contact:  

Kierra Box, trade and environmental regulation campaigner, Friends of the Earth

kierra.box@foe.co.uk  

Mary Stevens, innovation co-lead and AI campaigner, Friends of the Earth

mary.stevens@foe.co.uk

Annex 1

The table compares a range of scenarios from different Government planning documents, and seeks to enable cross-comparison by converting GW to TWh (based on different scenarios for usage).

Implied TWh/year at different load factors. Formula: TWh = GW × 8,760 hrs × average load factor.

Notes on load factor

25% = current UK utilisation, based on metered grid data (Green Web Foundation, Jan 2026, drawing on UKPN open data platform)

50% = illustrative mid-range scenario (estimate).

70% = illustrative upper scenario, based on US AI data (https://eta-publications.lbl.gov/sites/default/files/2024-12/lbnl-2024-united-states-data-center-energy-usage-report_1.pdf).

14