Smarter, greener extractivism: digital infrastructures and the harnessing of new resources

ABSTRACT

The past several years have seen increased scholarly attention to the concept of ‘extraction’ and ‘extractivism’ as critical frameworks in the humanities and social sciences. These are not only concepts and processes through which to understand material extraction but also expanded formations of extraction as requiring an assemblage of interlocking activities united under an umbrella of exploitative, material economic practices. This article demonstrates that digital data infrastructures, especially data centres, are acting as tools in which to manage the compounding contradictions of paired ‘green,’ digital growth in constrained electricity systems like Ireland. Building on theories of ‘green extractivism’ in the digital sphere and drawing upon fieldwork and policy analysis in data centre industry settings, this article argues that in the form of interconnected climate and digital infrastructures, tech capital is shoring up its role in green transformations, including in the grid systems that will need to adapt away from fossil fuel supply to intermittent energy sources and increasing demand from data centres. In this way, data centres are becoming technologies of green extractivism for overlapping projects of digital grid transformations and climate-friendly capitalism.

KEYWORDS:

• Media infrastructure
• data centres
• green extractivism
• environmental media
• critical data studies
• energy humanities

In Ireland, data centres have been a major feature of the landscape – and the country’s news cycles – since the early-2010s. In these early days, the discourse was exploratory and celebratory – Ireland was ‘“cool” for data centres,’ capitalising on a ‘natural’ climate and thriving tech ‘ecosystem’ to attract these profit-making multinational facilities (Lally, 2019). Similar to other northern European countries which were seeing growth in these infrastructures with the rise of smart devices, streaming, and cloud computing, from Iceland to Sweden to Norway to the Netherlands (Johnson, 2019; Rone, 2023; Upham et al., 2022; Vonderau, 2019), these industries were discursively tied by state boosters and industry lobbies to national prosperity, keys to development ‘progress’ in a post-industrial digital era (see Maguire & Winthereik, 2021).

At least in popular discourses, the connections between the ‘cloud,’ streaming, and the land, resources, and labours required to keep these digital industries running were broadly under negotiation in the public sphere in the early-2010s – while critical social scientists and humanists were drawing upon histories of media, science, and technology to conceptualise the materiality of the digital and its environmental implications (see Cubitt et al., 2011; Gabrys, 2011; Hu, 2015; Maxwell & Miller, 2012; Mosco, 2014; Parks & Starosielski, 2016; Starosielski, 2016), the dots connecting datafication and smart systems with the mounting extraction and harnessing of planetary resources had yet to be connected in systematic ways. As Ireland became a ‘data centre capital of Europe,’ with all of the compounding contradictions of digital growth metabolising through the country’s landscapes, big tech capital nested itself within the landscape of Irish infrastructural development through the enabling policies of FDI facilitation (Bresnihan & Brodie, 2024).

This was an intensive and extensive process by which multinational tech shored up its global hold and Governments adapted policies to the industry’s emerging economic dominance, and Ireland emerged as a place where the emerging climate strategies of big tech were tested against the regulatory frameworks being crafted to incentivise and manage decarbonisation. During these years, data centre companies, especially the US hyperscale data centre operators like Amazon Web Services (AWS), Google, and Microsoft, were getting ahead of the narratives that were emerging around the astronomical energy and resource usage of the ‘cloud.’ Discourses advertising ‘increased efficiency’ in spite of ‘unreliable metrics’ permeated the industrial research literature, arguing that voluntary ‘green’ initiatives and best practices were the way forward for this still under-regulated industry (Avgerinou et al., 2017; Host in Ireland & Bitpower, 2017); by the late-2010s, Google was promoting themselves as the single-most prolific private buyer of renewable energy in the world (Frangoul, 2018), and Microsoft and AWS were publicly buying up renewable energy contracts by the proverbial truckload (Day One Team, 2019; Microsoft News Center, 2019). The scientific research around ‘smart grids’ to manage intermittent renewable energy, and data centres’ role in their development, emphasised that data centres’ sustainability was not a political problem, but a technical one – the mechanics of grid flexibility could be adapted to suit data centre needs, which would in turn align with decarbonisation commitments (see BloombergNEF, 2021; Guo et al., 2021). The global and promotional climate calculations of companies that rely on near infinite and constant growth were coming up against increasing public and regulatory scepticism as well as the practical limits of twinned local grid challenges and decarbonisation and adapting strategies accordingly. These were not problems that could be resolved with better promotion. These companies started to adapt both their promotion and infrastructures. As an executive at a 2022 data centre industry webinar argued, ‘We’re no longer in a “do-good” phase, we’re really in a “we need to make an impact” phase’ (Iron Mountain, 2022). This coincides with the growing necessity of getting involved in the green infrastructure and energy space to manage the limits and contradictions of digital growth, reproducing technoscientific relationships to land, the environment, and people that facilitate – and even necessitate – expanded and more carefully managed networks of extractivism.

Such activities require a careful synergy between promotion and actual infrastructural build-out. AWS, Google, and Microsoft have each pledged commitments to 100% renewable energy and carbon neutrality through both energy and carbon calculations, networking their energy operations into new frontiers of value and the planetary ecologies required to harness them (see Amazon Press Center, 2023; Calma, 2021; Hogan, 2018; Pasek, 2019). When executed in the realm of sustainability commitments via ‘net zero’ and ‘carbon negative’ promises, what media scholar Mél Hogan refers to as ‘big data ecologies’ constitute the abstracted material dimensions of eco-mathematics which crunch complex political ecologies across territories into accounting ledgers that can be reported to industry regulators and promoted to the public (Hogan, 2018). All these goals are effectively offsets, in the sense that localised energy and carbon metrics are measured against global or remote purchasing, but they nonetheless have material effects in the ongoing growth of digital infrastructures and their energetic requirements. These calculative, technoscientific rationalities, by which climate mathematics act as justification for ongoing growth as long as the money is invested in ‘climate action’ and ‘sustainable energy,’ served to obscure the growing resource-intensivity of endless data collection for cloud, streaming, AI, and other digital services and the mechanisms of value extraction and creation these activities represented.

Through these value dynamics, new frontiers of resource-making were conditioned, and have gotten users and the public accustomed to big tech companies’ involvement in the global landscape of decarbonisation. This goes beyond using digital systems to more efficiently find and optimise ‘climate solutions’ via AI, machine-learning, and other big data-enabled ‘tools,’ as cloud companies are also becoming involved in by partnering with start-ups and research organisations (e.g., AWS, 2022; Matias, 2022; Microsoft AI, n.d.). As this article outlines, drawing from debates in media studies, human geography, environmental studies, and STS, big tech companies have increasingly become involved in energy systems via careful discourse management and infrastructural build-out. Employing discourse and policy analysis to understand the interface between corporate and state strategies for digital decarbonisation, this article analyses Ireland’s ‘data/energy nexus’ and green development strategies through the lens of green extractivism. Across generation, delivery, grid services, and through a number of other technical integrations with the grid that go beyond climate calculus, networks of green extractivism are being shored up through digital systems. From data extractivism to renewable energy and empowered by state policies towards digital/green transitions and ‘digitalisation and decarbonisation,’ the companies supplying the software of ‘industry 4.0’ are transforming the hardware of extractive capitalism and painting it ‘green’ – hardware which they also, largely and increasingly, have significant ownership and/or operational stakes.

From promotion to extraction

While such moves are frequently dismissed as ‘greenwashing’ (Miller, 2017), this extends beyond promotion, and requires a profound investment in regulatory and infrastructural transformations. Greenwashing is one way that capitalist organisations manage the PR around these mounting contradictions, but these discourses also have material analogues that sometimes exist in tandem or friction (see Lyon & Montgomery, 2015). Like all capitalist transformations encountering actually existing legal, regulatory, territorial, and environmental frontiers of extraction, ‘limits’ need to be managed, broken, and/or circumvented (Moore, 2022). The history of monopoly tech capital is also the history of regulatory struggle and adaptation (Rikap & Durand, n.d.), which occurs across territory, and the extent of ‘big tech’ power in contemporary global politics has led many orthodox ICT policy researchers to even extol the virtues of tech self-regulation and ‘corporate ethics’ as regulatory frameworks for faster uptake and reduction of harm via accelerating digital technologies (e.g., Arogyaswamy, 2020). But the saga of big tech dominance and the maintenance of its power also, in an era of complex global supply chains and patchwork technological development, is a story of messy and conflicting investments and resource commitments by actors at various scales and with various interests, especially across its infrastructures. Just like the state, corporate power is distributed and uneven, and operates in a field of tension and co-existence with public discourses and sentiments in particular places.

As the infrastructural backdrop of the digital economy, overwhelmingly controlled by the big tech ‘hyperscalers,’ the simultaneous management of this this image and underlying engineering has been crucial in painting data centres ‘green’ (see Carruth, 2014). These discursive politics are thus essential for understanding the resource imaginaries and physical infrastructures of ‘green extractivism’ (Bruna, 2023; Dunlap & Jakobsen, 2020; Voskoboynik & Andreucci, 2022), specifically in the interface between material energetic resources and the digital economy. The past several years have seen increased scholarly attention to the concept of ‘extraction’ and more broadly ‘extractivism’ as critical frameworks arising from histories of peasant, Indigenous, and agrarian struggle in the global south (see Chagnon et al., 2022; Gudynas, 2009; Riofrancos, 2020). In these analyses, extraction (and extractivism) are not only frameworks through which to understand material extraction – such as mining, oil drilling, and other activities of resource removal – but also expanded conceptions of extraction as requiring an assemblage of interlocking activities united under an umbrella of exploitative economic practices (see Gago & Mezzadra, 2017; Mezzadra & Neilson, 2015). Recognising the continuities and ruptures of these historical resource dynamics, scholars across disciplines have explored the concept of green extractivism, recognising that the market dynamics of climate adaptation are increasingly shaping relationships to extractive sites and practices, including especially through renewable energy (Allan et al., 2021; Dunlap & Correa Arce, 2022).

A crucial aspect of green extractivism is its mobilisation of discourses and imaginaries of climate action towards the landed expansion of extractive activities (Voskoboynik & Andreucci, 2022) – discursive justifications recognisable from existing models of extraction, but in a way that directly maps green capitalism onto historically uneven geographies. According to Daniel Macmillen Voskoboynik and Diego Andreucci, speaking about lithium extraction in the Atacama, rather than only a material difference, what actually differentiates green from ‘traditional’ extractivism is largely discursive and imaginative, aiming to ‘legitimise’ novel extractive enterprises in the name of sustainability, ‘climate-friendly development, ecological industrialisation, and techno-renewable futures’ (2022, p. 13). The discourses surrounding eco-modernisation and the ‘promises’ of industrial development associated mean that the infrastructures arising in green extractive regions are associated with the construction of a new, networked, green future on a supposed ‘greenfield’ or otherwise wasted territory, providing services and value for elsewhere at a time of planetary catastrophe while marginalising the possibility of truly ‘just transitions’ for workers and ways of life already existing there (see Mercier et al., 2020). For example, in the midlands of Ireland, data centres and renewable energy in the form of circular ‘Energy Parks’ have been proposed as ways to re-develop the post-extractive cutaway bogs of Ireland’s peat-driven energy modernity, a model which contracts the public just transition to multinational tech and energy suppliers; or in the western regions of the country, multinational wind parks contracted to provide energy to data centres are typically built on upland blanket bogs, otherwise ‘wasted’ land that can be mobilised for green and digital transitions. In each of these cases, community voices and projects are marginalised, and frequently bulldosed in favour of large-scale, multinational extractive enterprise – as the latter sections of this article will emphasise, specifically wind energy. Green extractivism as a framework, aligned with the expanded senses of extraction above, thus enables scholars to represent both a logic (epistemological) and practice (material) of governance that applies the calculative rationalities of ‘socially necessary extraction’ – in this case, to support decarbonisation to ostensibly maintain a liveable planet – to landscapes and populations across different socio-ecological regimes. Whether raw mineral extraction for the underlying components of digital technology and renewable energy, as has been well-represented in other literature (e.g., Arboleda, 2020; Voskoboynik & Andreucci, 2022), or expanding renewable energy infrastructure in Ireland and elsewhere to support the growth of data infrastructures, the violent and dispossessive processes of material extraction become not only necessary but justified by normative and moral knowledge systems, aligned with the reproduction of capital (Cirefice & Sullivan, 2019; Gómez-Barris, 2017). Following Richard Maxwell and Toby Miller, we have to ask if extractive expansion for data infrastructure is socially necessary when tethered to unsustainable (and profit-driven) corporate interest (2012, p. 7), whatever the specific resource-making process and imaginary.

Critical scholars of ‘renewable’ energy have long criticised the logics and practices by which large-scale industrial infrastructures in these sectors have been constructed across existing unequal geographies and environmental relations (Howe & Boyer, 2016). As Alexander Dunlap and Martín Correa Arce argue in the context of ‘renewable’ resource logics, extraction constitutes the ‘pulling, drawing out and harnessing (usually with special effort, skill or force) … as green extractivism suggests, kinetic energy from wind, solar, hydrological and bioenergy “resources”’ (2022). Key to this is the material and discursive form in which corporate power has infrastructurally managed the political ecologies of these ‘emergent’ resources in order to harness their value (Knuth et al., 2022), which requires the management and refinement of intermittent atmospheric energy sources into electricity that can be reliably and responsively distributed to sites of consumption. Mechanisms of ‘resource-making’ (Bridge, 2010; Kama, 2015) are thus affected by the tools and practices by which the resource is extracted and distributed. In the case of atmospheric energies, this is seen as ‘commons’ for the taking, circulating unhindered through the air but able to be harnessed via infrastructures on pieces of land or sea, typically through the enclosure of property and ownership (see Hughes, 2021), thus transformed and then distributed for-profit through privatised utilities systems. They represent a key overarching tension in discussions of climate action: between liberal frameworks of ‘sustainability’ and the capitalist mechanisms by which these regimes are both maintained and justified (Vaughan et al., 2023); and environmental justice movements writ large, such as anti-extractivist struggles against the resource regimes ‘sustained’ by such policies (Bresnihan & Milner, 2023; Cirefice & Sullivan, 2019). As the following sections will argue, data centres’ increasing utilisation as infrastructural managers of these complex and unruly energy geographies makes them crucial refineries of both data and energy – by making energy legible as and for aligned processes of datafication.

Green data extractivism

Many scholars of green extractivism have recognised the currency and importance of digital technologies and corporations in official justifications of green extractivism (see Arboleda, 2020), as it has become clear that new land grabs for geological materials like tech and energy minerals (e.g., lithium, called ‘critical minerals’ or ‘critical raw materials’ by states) and extensive acreage for atmospheric (like wind and solar), geological (like geothermal), and tidal harvesting are well underway by capitalist actors and powerful state partners (see After Oil Collective, 2022; Watts, 2018). Beyond the minerals making up the land, land-as-property is also increasingly central – as David McDermott Hughes admits, ‘To generate the same energy as a coalfield, solar panels would need roughly four times as much space, and wind turbines seven times’ (2021, p. 27; also Wade & Ellis, 2022). Across these geographically distributed facilities, and in their integration to the grid and the buildings it delivers to, forms of forecasting, modelling, and real-time management and delivery need to manage, frequently aligned, in Ireland, with times of data centre energy-intensiveness. Thus, the processes by which datafication is directly shaping, through infrastructures and policies, the landscape of decarbonisation beyond sites of overt extractivism require more expansive conceptual linkages across the various types of extraction that digital technologies facilitate and require (see Iyer et al., 2021; Martín & Valderrama Barragán, 2023).

But this is what makes the landscape of datafication such an important site for a comprehensive analysis of green extractivism. Much focus on the idea of ‘data extractivism’ has been on the typical industry adage(s) that ‘data is the new oil,’ or the ‘oxygen’ of the fourth industrial revolution (see Celis Bueno & Schultz, 2021; Prasad, 2022; Taffel, 2023). But these foci on the specificity of data as a resource with specific epistemological stakes (Ricaurte, 2019), a new ‘landgrab’ via territorial expansion of big tech business practices as Couldry and Mejias (2023), underplay the fact like for any resource, land – and its ownership – remains a central factor in the supply chains and networks of refinement required to make the resource manageable and productive. Like the atmospheric circulation of wind can only be harnessed for capital by stems and blades planted on privately-owned land (Wade & Ellis, 2022), data is collected and funnelled into infrastructural networks across land that is only differentially territorialised and governable, profited from by a select few while occupied by all. Entering this ‘atmospheric phase of capitalism,’ as Hughes defines it (2021), thus also means paying attention to the ways that land continues to shape the flows of digital and energy capital.

In the context of digital data’s extractive dimensions, a heuristic of extractivism through environmental politics is necessary to account for the centrality of the ownership and ability to accumulate capital through land and infrastructure in the expansion of digital economies through data centres. With such an extreme concentration of multinational data centres (see Bresnihan & Brodie, 2021, 2023; Brodie, 2020, 2023), in Ireland, they have also become facilities through which to manage the ongoing contradictions of extractive growth across digital and renewable energy systems. The planned simultaneity of extractive processes, made efficient by digitalisation and datafication, make them crucial real-time operators and managing facilities of extractive processes. With ongoing cloud expansion, data centres and the business models that they support, from streaming to AI training models, have increasingly and more effectively integrated the supply chains required to power them into a circular loop of efficiencies, which in turn facilitates and accelerates processes of material extraction supposedly ‘outside’ of its supply chains. As concentrations of infrastructures and resources, data centres metabolise the environment and the data of digital economies not only as monstrous machinic refineries of data, but as multiply extractive facilities requiring a multitude of extractive inputs – marshalling extractive supply chains in the form of energy-intensive, remote computing (Brodie, 2023).

The remainder of this article probes these resource imaginaries through the ways in which big tech companies are increasingly infrastructuralised via the delivery of energy and datafication as its own utility (see Steinhoff, 2018). The operation of infrastructural networks across private contractors and operators, let alone the owners of the resources and infrastructures themselves, increasingly distances resources and utilities from public and democratic control and accountability. Big tech companies infrastructuralise an ethos of green extractivism being recognised globally as a primary emerging logic of capitalism. The closed-loop systems emerging between energy and data represent where discourses and materialities converge to craft a world for green tech capital, managed within and through the material infrastructures of large-scale data centre facilities.

Territorial limits of efficiency

As has been much described, data centres use significant amounts of energy drawn from electrical grid connections. In 2021, data centres used 14% of Ireland’s grid capacity on a daily basis, as compared to the 12% used by all rural homes combined (see Carswell, 2022). In 2022, data centre electricity usage ballooned to 18% of Ireland's daily grid capacity, more than even all urban homes combined, and in spite of the de facto ‘moratorium’ implemented in Dublin by semi-state grid operator Eirgrid until at least 2028. For comparison, we can look at two of Europe’s other significant data centre markets: in the Netherlands (which has also temporarily banned data centres) these facilities use around 2-3% of the country’s electricity; in Denmark the number may reach 7% in 2025, which has already prompted concern. Outside of Europe, in Singapore, which also implemented a moratorium in 2019, data centres drew 7% of the country’s electricity daily in 2020.

Concerns about the overwhelming of the Irish grid has led to calls for a moratorium on data centre development by left-wing parties like People before Profit and environmental groups like Futureproof Clare and Not Here Not Anywhere. These groups argue that data centres present an unnecessary hurdle rather than solution to Ireland’s climate targets, justifying the expansion of ‘transition fuels’ like natural gas, short-term solutions locking in new kinds of fossil fuel extraction and burning to support ongoing economic expansion (see Corr, 2022; Not Here Not Anywhere, n.d.). These calls for limiting or halting data centre development were only heeded by state agencies once the Commission for the Regulation of Utilities (CRU) and Eirgrid engineers flagged the very real possibilities of blackouts in 2021 (Curran, 2021). After years of reports indicating that the growth of data centres in Ireland was coming up against the practical limits of the national grid, CRU released evidence and recommendations that made the situation abundantly clear: the strategies for data centre development in Ireland had to change, and as soon as possible (McLay, 2022). The report presented three potential avenues: 1) do nothing, 2) moratorium, and 3) significant mitigations, including compelling data centre providers to provide on-site energy and/or locate in rural Ireland further from constrained capacity. It was effective in both communicating technical realities and materialising public concerns, while also demonstrating the alignments between technoscience and enterprise in presenting ‘solutions’ to a crisis that had long been recognised by left politicians, journalists, and social scientists.

In spite of the evident necessity of curbing data centre development in the short-term, the Fianna Fáil/Fine Gael coalition Government and influential semi-state development organs – especially IDA Ireland, the state’s FDI facilitator who have spoken publicly about their opposition to the mitigations (Allen, 2022) – have been reluctant to implement full restrictions. The Government’s ‘Principles for Sustainable Data Centre Development’ released amidst these regulatory recommendations in 2022 espoused an expanded capacity for data centre development, stopping short of ‘curbs,’ arguing that new developments need to demonstrate economic impact, responsiveness to grid capacity and efficiency (including provision of grid services), additionality of renewables, co-location/proximity to future-proof energy supply (for example by private wires), decarbonisation by design, and SME access and community benefit (Government of Ireland, 2022, pp. 20–21). While admitting the need for short-term measures, the Government has remained supportive of the data centre industry in the long term as a key supporter of the ‘“twin transitions” of digitalisation and decarbonisation of our economy and society. These transitions can – and must be – complementary’ (see Government of Ireland, 2022, p. 2). Eirgrid followed the third CRU recommendation not to halt data centre growth entirely, but to limit connections in Dublin as of 2021, and as of late 2022 there were 21 new data centres in planning outside of Dublin (Moss, 2022b). Data centres are still being approved in Dublin, finding workarounds like connecting to the Dublin gas network to power the facilities through on-site fossil fuel generation or demonstrating other proprietary solutions (Hosford, 2022). According to the Bitpower dashboard, an energy solutions-cum-cloud advocacy company in Ireland, in June 2023 there were 82 data centres in Ireland utilising 1,261 MW of power (an average of 15 MW per facility), with 14 more under construction (averaging 25 MW per facility), 40 approved for construction (averaging 32.15 MW per facility), and 12 in various stages of planning (averaging 32.5 MW per facility) (Bitpower, 2023). The majority of these remain in the Dublin region.

Frustrated data centre industry speakers at a 2022 event in Dublin argued that Eirgrid was aware of the strains that data centres would put on the Irish grid as early as 2014, and so the industry was being scapegoated by the state and victimised by popular discourse (Foxe, 2022). Industry boosters formed advocacy and lobby groups, including Cloud Infrastructure Ireland, to protest these regulatory limits, with solutions primarily being their greater involvement in the energy system to manage efficiencies (e.g., Cloud Infrastructure Ireland, 2021). Industry practitioners frequently argue that efficiency is key – cloud computing has made business activity more energy efficient (compared to filing cabinets and server rooms) and has become more energy efficient generally as technologies have advanced. Companies are publicly concerned with their energy management systems as the years have gone on, by necessity employing experts trained in cloud infrastructure and its energy engineering. As widely cited by the industrial and technical literature, efficiency gains have offset the astronomical energy growth that might have been expected with the exponentially increased computing power available over the past half a century (IEA, 2023; Masanet et al., 2020; Pasek, 2023).

But even in the analysis by data centre industry actors, the issue is not necessarily about what is done with data and where it goes. The issue is then, in my analysis, about infrastructural growth associated with data-driven technologies when tethered to unequal systems of capital accumulation and its necessarily extractive networks, whether multinational data centres or the similarly private energy systems that they are increasingly shaping and depending on. How much more ‘efficiency’ can be tolerated in other sectors when it requires these continuously expanding infrastructures and material inputs? As efficiency reaches its functional limit under current computing technology, which by most accounts it is (Pasek, 2023), the efficiencies gained by datafication via AI, machine-learning, and these other extremely data-intensive procedures will be hamstrung by the functional capacity of the infrastructure and the ongoing availability of mineral and energetic inputs. As efficient as any computing system is actually able to be, and as much as AI is leading to efficiencies in other sectors, as long as data and its energetic requirements keep growing, computing eventually reaches the peak of its efficiency and needs even more infrastructural build-out at a global scale in the form of large-scale processing, storage, and distribution facilities and its supporting networks.

Thus, beyond the territorial expansion of data colonialism via business models of data-driven capital, new landed frontiers for telecommunications, energy, and extractive infrastructures multiply (see Arboleda, 2020; Mezzadra & Neilson, 2013). For example, big tech companies are entering direct contracts with renewable energy providers via corporate power purchase agreements (CPPAs) to offset their energy use in Dublin with demand-led development of renewable power in the rural north, west, and south – AWS’ contracts with multinational wind farms in Antrim, Donegal, Galway, and Kerry have been variously controversial, with facilities in Donegal and Galway undergoing ongoing and ardent community objection at the time of announcement and Donegal being the site of a catastrophic peatslide in 2020 (see Ortar et al., 2023). Microsoft has also reportedly purchased 900 MW of renewable energy through these agreements, 28% of the state’s goal for 2030 (Moss, 2022a). The mostly rural places these wind and solar facilities are built frequently experience significant fuel and energy poverty, and are often ecologically sensitive landscapes like peat boglands, meaning the inequalities of multinational energy infrastructures providing ‘green’ offsets to digital corporations are starkly experienced and widely protested, with minimal access to the green, digital promises of these smarter energy systems (Brodie, 2023). Thus, in spite of the democratic exclusion of rural communities facing extractivist developments, amidst the changing landscape of climate adaptation and expanding digital capital, techno-solutions to the limits encountered by digital growth continue to shore up big tech’s control over infrastructural networks, as data centres are increasingly key to creating ‘the smarter, more flexible energy systems needed to get to net-zero emissions’ (Varro & Kamiya, 2021) – from energy generation to households and most points in between.

Infrastructural ecosystems of atmospheric refinement

The lack of material connection and recognition of frictions between discourses around energy, data, and resource requirements of big tech, and even an argument that digital technology under current regimes of growth will be essential to scale down emissions and meet climate targets, is part of the discursive construction of ‘ecosystemic’ relations which imagine a manufactured holism across society, state, capital, and the environment. They coalesce into a set of naturalised relations which, according to Thomas Patrick Pringle in his essay ‘The Tech Ecosystem and the Colony,’ ‘indicate the segregated character of social organisation, which may be the critical definition required for considering how tech ecosystems exacerbate standing radical inequality and neglect the material destruction of the environment that premises the digital economy’ (Pringle, 2021). Data infrastructure as a privatised utility to be managed within an apparently functioning tech ecosystem at ‘equilibrium’ with decarbonisation stands in stark contrast to the visible, and contestedly regulated, technical interactions between the ‘tech ecosystem’ and the grid, or the ‘tech ecosystem’ and the environment. In these operations, in their local and global character, we can see what Pringle identifies as the valuation of ‘the action of the whole at the expense of the marginal’ (2021), except that the ‘marginal,’ whether people (especially rural communities) or environments, are partially and imperfectly presented as undifferentiated beneficiaries of rather than sacrifices to imagined tech equilibrium in assembling a decarbonised market. As this article argues, data centres are deployed by tech industry boosters – materially and discursively – to manage and maintain this balance towards ongoing digital growth, acting as resource refineries at an increasingly operationalised data/energy nexus.

Emerging exchanges between data centres’ multinational infrastructures and local instantiations of their operations, which occur across public infrastructures and through publicly managed utilities, have a privileged status due to the green imperative of state energy policies being aligned with economic imperatives of digitalisation. What is crucial in Ireland, beyond the specific mechanisms of resource interoperability as described in this article, is the conjunction of data and renewable energy as discursively connected to a national development framework through FDI. Ireland’s resource base, at this point, is imagined to be entirely the result of FDI companies locating here – at least, this is the narrative promoted by figureheads of the data centre industry. As Garry Connolly, CEO of Host in Ireland, a preeminent lobbying force for the data centre industry in Ireland, emphasised in a webinar from 2023 entitled ‘Digital Infrastructure: An Irish Export Success,’ Ireland’s biggest export in 1973 was live cattle, whereas in 2023, 27% of Ireland’s exports are ICT-related services in 2023. In this progression, as he puts it, ‘from a spud economy to a chip economy,’ a growing industry of data and technology practitioners had made Ireland a global hub for data, but also a site where its contradictions needed to be worked out based on national energy constraints. Amidst the growing ‘resource nationalism’ of Ireland’s data economy – the result of these overlapping forces of multinational investment in tech – we see similar arguments arising around the country’s offshore wind potential, with industry and state figures proclaiming the island could be the ‘Saudi Arabia of wind energy.’ However, this will only happen if ‘all stakeholders work together,’ including the data centre industry, to quickly actualise the infrastructural and industrial environment needed to materialise this promise (Dunphy, 2022). Connolly reiterated the urgency and importance of digital infrastructure and FDI for Ireland’s development model through these specific resource imaginaries, in reference to Ireland’s wind energy potential:

We don’t have any other choice. We’ve no gold. We’ve no oil. We’ve none of these other natural resources to date. But what we do have coming through is 15x more access to electrons than we will ever need. What’s data created from? Electrons. So, we’re going to get over this hump very soon. And then people will be saying, ‘What can we make with all of these excess electrons that we can export?’ Skills, services, and data. (Host in Ireland, 2023)

In this framing, data centres marshal the ‘electrons’ harnessed from Ireland’s renewable energy resources, connecting the national resource of wind with how that can be refined and manufactured into a product to be exported – which is also framed a resource for extraction.

Similar slippages between the discursive and infrastructural dynamics of data centres’ green extractivist ethos occur across the technical literature of industry. Cloud Infrastructure Ireland, for example, commissioned a 2022 report, with industry partners Baringa and Bitpower, called Green Data: A Vision for Sustainable Data Centres in Ireland, which proclaims:

Looking ahead, Ireland has massive potential renewable energy resources which exceed Irish demand by over 300%. As these are developed, the electricity they produce can either be exported ‘raw’ to overseas markets via interconnectors, or it can be ‘refined’ domestically, including in data centres, and exported as computer services. This latter option creates much more value in the Irish economy. (Baringa & Bitpower, 2022, p. 7)

The language in this report demonstrates not only the increasing concern with becoming integrated into energy systems rather than simply using them, but also the discursive management of what data centres actually do – the idea that they refine not only data but energy itself. Cloud services, in the form of infrastructure and software, make data circulating around and for energy systems viable for greater efficiencies via data extraction, but only with the proper balancing of these energies across generation, refinement, and delivery (which describes both what energy systems and data systems do through the data centre itself).

Data centres thus, in this imaginary, act to ‘balance’ a haywire data/energy ecosystem during the transition to renewables as extractive inputs need to be managed in more closely aligned tandem. If the grid is not currently up to the task of handling these energy sources, especially with the growing need for electricity from data centres (among other sectors like EVs), data centres can thus act as service providers while also benefitting from more efficient management. As a 2021 engineering research article on the possibilities of data centre grid solutions by ‘Cloud Service Providers,’ the top three in global market share, AWS, Microsoft, and Google, each host flagship facilities in Ireland, argues,

using cloud systems as the information infrastructure for the next-generation grid to deliver multi-level information services (i.e., Infrastructure as a Service, Platform as a Service, and Software as a Service) to the stakeholders in smart grids. The work also demonstrates how different kinds of smart grid applications, including compute- intensive, data-intensive, and multi-party collaborative applications, can be efficiently supported by data center backboned cloud platforms. (Guo et al., 2021)

These multiple points of entry of cloud companies into energy provision and delivery represents an even more front-end version of their involvement in energy sourcing, demonstrating greater influence at more points in the energy supply chain, and in the material function of refining energy into data – and vice-versa.

This is an emerging extractive landscape requiring serious attention from media scholars and critical resource geographers. On the one hand, for example in Dunlap and Arce’s framing (2022), there is minimal difference between the energetic resources of lithium and other critical minerals with existing and familiar forms of extraction like metal and coal mining. Wind, solar, or geothermal, on the other hand, are novel resource frameworks in that they require harnessing and mastery of earth movement and systems via real-time processing through digitalised infrastructures. Discussions of smart metres, datafied heating, and digital grids are of course essential for understanding the emerging utility landscape, increasingly populated by big tech companies and aligned digital interests (Sadowski & Levenda, 2020; Sareen & Müller, 2023; Velkova et al., 2022). For example, ESB in 2019 partnered with Siemens for the rollout of pilot smart metre programmes (Siemens, 2019), a company with significant investment, technical and financial, in Ireland’s energy/data generation landscape and specifically enclosed ‘energy parks’ combining data centres, renewable energy, and battery storage in the midlands (Siemens, 2023). Connections between distribution networks and sites of extraction are being collapsed, and pathways between companies shortened. Digital technologies collapse and condense the process of energetic transformation from resource to infrastructure to utility. This shortened supply chain is of great interest to the industry as they try to push the regulatory framework of ‘private wires,’ direct lines of electricity bypassing the public grid from sites of generation, to re-localise their energy supplies (see Bresnihan & Brodie, 2023). But the difference is now that the product of energy is more rapidly available as a service, as the resource-making is enacted by and through digital efficiencies. What arises, then, is a friction with the existing technological affordances and capabilities of grids and household energy systems in the atmospheric constitution of data and energy resources. Real-time coordination of extraction, refinement, and delivery of data/energy is increasingly operationalised across these interlinked systems, and this must be both discursively managed as a tool of decarbonisation and infrastructurally built-out as a tool of security and crisis adaptation in a messy and sometimes non-interoperable infrastructural landscape.

While the above presents evidence that data centre providers are increasingly concerned with how they can become essential to grid transformation while contributing to energy challenges during decarbonisation, this is not to suggest that this control is primarily profit-driven (for now). Data centres are being proposed as grid solutions providers due to their on-site energy infrastructures, including battery storage and uninterruptible power supply mechanisms, in order to absorb excess intermittent renewable energy capacity – ensuring ‘grid flexibility’ by absorbing power and giving it back to the grid in times of reduced capacity, acting in service of grid decarbonisation (Paananen & Nasr, 2021). Microsoft in particular has tested this model in Ireland and plans to export it elsewhere in their data centre fleet (Roach, 2022), meaning the potential for commercialisation is there. On the other hand, AWS, like in other territories (see Velkova, 2021), has invested in district heating as a means to demonstrate their value in delivering circularity via utilising waste heat from their data centres, already the product of excess energy consumption, across a grid managed by South Dublin County Council, trading as the non-profit ‘Heatworks,’ in Tallaght, Dublin in coalition with various other public bodies (see Swinhoe, 2021). Similarly, through a partnership with an organisation called Energy Cloud, AWS cloud solutions have been enabling the delivery of ‘free tanks of hot water’ to residents in affordable housing developments at times of excess renewable energy generation (see Energy Cloud, 2023). While in 2023 these pilot projects were still delivered on a voluntary basis, demonstrating corporate responsibility and green commitments rather than generating value, they are at the same time posed as future techno-solutions to the infrastructural problems and inequalities raised and compounded by corporate-driven digital growth in the form of data centres. When you also bear in mind the cloud-based technologies for forecasting, monitoring, and measuring energy availability and use along the supply chain of the grid, by companies like Siemens and GE, the implications of big tech involvement in the operation and delivery of these systems introduces difficult questions about the administration of public utilities by multinationals and the public dependence upon their proprietary infrastructures. This is true whether considering the potential for a tech downturn, or simply in the extractive systems designed to maintain the energy status quo via techno-solutionism.

But common here is how data centres can act in service of a more resilient grid, or a greener power supply, or more efficient energy use, or even direct utilities in the form of heating – these are public services that will increasingly entail the expansion of the infrastructural footprint of privately-owned and run digital technologies, whether in terms of materials required to sustain and support them or their spread into more and more facets of infrastructure and utilities and our lives. Because at the same time, as an executive at a 2023 data centre industry conference in Dublin emphasised, data centre providers must resist attempts to classify – and thus regulate – their operations as ‘public’ utilities in and of themselves. If that happens, not only does it threaten the viability of the industry in Ireland – considering the ability and willingness of hyperscale data centre providers to move elsewhere, as the executive warned – it risks setting regulatory precedent that the industry should resist. Thus, at the moment, these programmes appear a practical response to the challenges of contemporary energy provision and its regulation, especially in Ireland, where these systems are in disarray due to years of underinvestment and already-existing privatisation leading to a patchy, unequal, and inflexible energy grid. This situation uniquely suits companies trying to extract value via infrastructural expansion, contracted to supply utility-adjacent services to an FDI-hungry and public-spending-averse state environment, while avoiding what they see as unwelcome regulatory scrutiny faced by public utilities. But the question that is perhaps most important remains: if these companies leave Ireland, as they are liable to do based on changing business priorities and regulatory circumstances, what are the consequences of locking them into public energy transition plans and infrastructures? Smarter, greener extractivism is emerging across the energetic networks of increasingly infrastructuralised digital capital. While for now it is only possible to speculate what this will look like across ostensibly public utilities, the points of action are undoubtedly at sites where multinational tech are becoming increasingly integrated into energy systems and the public imperatives of decarbonising energy systems.

Conclusion

What the ‘green’ transition is compelling decision-makers to realise at all scales is that energy systems, along with the vast infrastructural networks and industrial activities that they power, are interdependent across patchwork infrastructural, land, and utilities ownership, but not interoperable. The current landscape of techno-solutions proposes digital data as a tool by which the increasingly messy rhythms of atmospheric renewable energies and their dispersed global supply chains can be managed. Joined-up thinking is of course important, and climate solutions are frequently written as independent rather than interdependent, individual nodes through which to manage different transitions – energy (wind, solar, hydrogen, wave, hydroelectric), digital (cloud, AI, streaming), transport (trains, buses, cars), minerals (critical and rare earth), etc. – rather than part of a vast, and expanding, infrastructural network of energy and climate solutions increasingly entangled with interrelated technological systems. These transformations traced above about managing a transition away from carbon-intensive activities and fossil fuels. These are turbulent shifts, and the terrain of struggle is complex and not intuitive by any classical dynamics available to scholars and activists focused on ‘climate’ as an encompassing heuristic. But the expansion of capitalist activities, however ‘green,’ will always require sacrifices. In this turbulent environment of climate change, the infrastructures being designed must not only manage the messy, patchwork territories of ‘supply chain capitalism,’ itself a system that thrives on instability and precarity (see Tsing, 2015). They are also an embodiment of a logic of contemporary and future governance for which this turbulence can only intensify, and will require increasing mechanisms of territorial and environmental control – from policing of conservation to subduing dissent for large-scale projects to geoengineering (see Buck, 2019). Green and data extractivism are not only mechanisms of harnessing value and accumulating capital, they are also managerial symptoms of managing turbulence – atmospheric, infrastructural, and otherwise.

As systems being designed to manage the infrastructural turbulence of climate adaptation, ‘green’ technologies are being faced with increasingly tricky geopolitical and planetary circumstances, to the degree that their era might be one of what Sebastián Lehuedé has called ‘planetary data turbulence’ (2022). Ireland was, after all, once selected as Europe’s data centre haven because of its stable economic, political, climatological, and geological conditions (see Brodie, 2020). But climate factors influence geology – after all, the ‘Anthropocene’ (whatever its terminological problems and conceptual clarifications) is always geological timeline (Yusoff, 2018). Geology and climate themselves are central to energy systems, as well as to the ultimately land-based resource politics that dictate the digital economy and its atmospheric energies, infrastructural networks, and distributed technological devices; and the histories tied up in these technoscientific approaches to land and resources are inextricable from how these networked extractive systems function. Data, as Lehuedé notes through Paul Edwards, does not flow freely (2022), much like supply chains in general and especially those of energy. These are lumpy systems and services constantly dealing with blockages and frictions, and this includes infrastructural affordances and the geopolitical circumstances behind their existence. Capitalist solutions, then, create new sites through which these lumpy terrains must be navigated.

Data centres, and their increasingly synchronous integration of digital/resource/utility systems – from extraction to refinement to consumption – require an infrastructural assemblage that can reconcile and make these sites and systems interoperable. However, their energy security mechanisms built to deal with uncertainty – grid flexibility services, private wires, microgrids, and software/hardware to make demand more supply-responsive to intermittent renewable energy across national grids and worldwide data centre networks – are not a unified object without the business models that necessitate these facilities in the first place. Thus, beyond implicating Ireland and other territories gambling with a corporate tech-driven future within the compounding planetary inequalities of ‘green’ digital capitalism built on extractivism and exploitation across multiple territories, the intensifying links with digital capitalism’s extractive supply chains within infrastructural delivery risks institutionalising business models designed to ruthlessly manage turbulence not by building better and more equitably – but by structuring precarity within their very modes of operation, valuing ‘the action of the [planetary] whole at the expense of the marginal’ (Pringle, 2021). This reveals the fundamental imbalance of the systems that data centres are designed to support, where ongoing struggles for access, rights, and equity on the terms dictated by data-driven tech expansion frequently promote technical fixes to market mechanisms which maintain the structural inequities of capital. There are no more data centres without compounding digital growth, no matter what kind of ‘efficiency’ these data-driven economies facilitate. In Ireland as elsewhere, a development model centred on the non-stop expansion of these types of infrastructures, to the degree that they become essential private enablers of public utilities delivery, crafts a decarbonised future in the image of big tech-driven capital. At the same time, it also risks stranding entire areas and infrastructure dependent upon a precarious economic resource – private digital data and its energetic facilities.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by an Ad Astra start-up grant at University College Dublin.

Notes on contributors

Patrick Brodie

Patrick Brodie is an Assistant Professor and Ad Astra Fellow in the School of Information and Communication Studies at University College Dublin. His research focuses on the environmental implications of digital media infrastructures, with a particular focus on energy systems and geographies of extraction [email: [email protected]].