A recent article by the CEO of the Artificial Intelligence (AI) company Anthropic heralds the teleological acceleration towards an AI ‘superintelligence’ as the solution to the crises of our world today. Dario Amodei, in Machines of Loving Grace theorizes how the acceleration towards superintelligence could be utilized to solve a wide range of our most pressing crises, including slowing down climate change by innovating new solutions like atmospheric carbon removal. Amodei’s promise for the future is at odds with the concerns of today as exemplified by a Goldman Sachs study which revealed that a ChatGPT query needs nearly 10 times as much electricity to process as a Google search. With the boom of generative AI, this figure is projected to increase at levels that will quickly exceed the available supply of renewable energy. In his piece he outlines how these breakthroughs from strong AI could be limited. He specifically mentions two limiting factors which could stop the development of powerful AI: the lack of data on which AI can be trained, and ”constraints from humans”, which Amodei outlines as misplaced fears and legislation that dampen the potential of the technology. In a section on tilting AI in the ‘right direction’ to ensure the continuation of a liberal democracy, he shows his hand by outlining the “great sacrifice and commitment” required to ensure the terms on which powerful AI should be built: “securing [AI’s] supply chain, scaling quickly, and blocking or delaying adversaries’ access to key resources like chips and semiconductor equipment.”
Amodei transforms socio-political turmoil into a technological problem, one which can be solved in large part by increasing resources towards the development of AI. This perspective is not unique; tech billionaires envision our society’s most pressing problems as testing grounds for their innovations, therefore justifying their actions as necessary to find solutions. This belief in generative AI’s inevitability, and importantly, necessity to be the solution, is not fringe. A recent IBM study found that 64% of CEOs invested in AI solutions because “the risk of falling behind [drove] them to invest in some technologies before they [had] a clear understanding of the value they bring to the organization”. The all-encompassing claims to problem-solving sparking its urgent necessity, have trickled their way into our workplaces. As states prioritize corporate self-governance and corporate capture becomes increasingly codified, Amodei’s solution of quick scaling seems to be rearing its head in our policy direction as well.
The uncritical hype to adopt better and faster generative AI–the quick scaling and the securing of supply chains required for these solutions–naturalizes the material cost of AI solutionism.
The uncritical hype to adopt better and faster generative AI–the quick scaling, the securing of supply chains required for these solutions–naturalizes the material cost of AI solutionism. By the purely digital interface most users associate with generative AI, there is a perception of it being “everywhere and nowhere.” This narrative imbues AI with an ephemerality that allows it a slipperiness in the face of critique. It also “conceals the pervasive infrastructures that are built in order to generate [the] data” upon which AI sits. The computational infrastructures that undergird the project of generative AI are everywhere: data centers, transoceanic or submarine cables, fibre-optic networks, and internet exchange points, to name a few. These infrastructures impact and are impacted by their geographies. An examination of the data center, therefore, presents the opportunity to understand how AI solutionism impacts our political ecologies and what is required in order to imagine an alternative to this Big Tech-led narrative.
Data Centers: What are they?
Physically, data centers are ‘server warehouses’, stacked with hardware components like servers, switches, routers, and storage devices that support the interactions being carried out on platforms. Due to the critical role of servers to provide computational power, data centers also house infrastructure to ensure efficient server use: power and cooling systems, air distribution systems, and uninterruptible power supplies.
However, the data center is not an isolated physical structure. To understand how it fundamentally shapes the planetary inequities of AI expansion, its material infrastructure must be examined discursively as a node in larger networks of extant unequal social, legal, and environmental geographies, where states actively codify policies to accelerate extraction.
As the generative AI hype is steered by the techno-solutionist rhetoric of Silicon Valley technocrats, data centers are increasingly molded by the same logic, too. At the heart of this rhetoric is the subsidization of energy, natural resources, and labor for private gain:
Energy, striking incentives for the use of power:
Multinationals in Ireland were reportedly charged a fraction of what ordinary consumers paid for bulk electricity during the 2022–2023 energy crisis, where data centers, drawing 14% of Ireland’s daily energy in 2021–projected to reach 28% by 2031–secure discounted access to public resources.
Labor power, relaxed under corporate self-certification:
In India, states like Uttar Pradesh have recommended exempting data centers from labor inspections under key acts (Factories, Minimum Wages, Maternity Benefits), outsourcing enforcement to firms’ self-reporting—a deregulatory gift to capital that renders workers vulnerable to wage theft and unsafe conditions for the profits of the data center.
Natural resources, siphoned for private infrastructure:
Policies like Uttar Pradesh’s 2021 draft data center policy mandate 24/7 uninterrupted water supply for data centers, with treatment plants built as ‘common infrastructure’—effectively privatizing public utilities to sustain the very systems deepening ecological strain.
The data center emerges not as a neutral container of computation but as a sociopolitical artifact, its material operations enabled by discursive constructs like ‘efficiency,’ ‘innovation,’ ‘ease of doing business’ that mask its role in entrenching colonial patterns of resource and labor extraction.
Data centers as Sites of Extractive Capitalism
The data center, as an expanding network of technological development, represents the material implications of digital technologies, beyond how they are used, consumed, and optimized. It operates as a site of the choices made by the Silicon Valley technocrats who wish to “move fast and break things”, privatize profits in the hands of a few, and socialize losses among the majority. While this paints a relatively bleak picture, the data center also shines a light on the material costs of this vision. Once picture-perfect, this vision of a fully automated market-first society has begun to crack.
Data centers, as a growing body of literature highlights, operate as energy siloes. They have an insatiable thirst for energy, water, and mineral resources. As Sam Altman pointed out, “We do need way more energy in the world than I think we thought we needed before, and I think we still don’t appreciate the energy needs of this technology.” With Big Tech’s unrelenting pursuit of scale and growth, the data center reveals a broader set of material relationships between the human and non-human actors of a data center. The inevitable increase in the number of data centers will lead to a rise in ecological costs. This cost will be borne by the people, while Big Tech siphons off local communities’ resources to maximize profits.
The inevitable increase in the number of data centers will lead to a rise in ecological costs. This cost will be borne by the people, while Big Tech siphons off local communities’ resources to maximize profits.
Digital capitalism, led by the aforementioned technocrats, reflects “centuries of extraction of the world’s natural resources and raw materials, and of mass corporate and individual wealth accumulation generated by ruthless business tactics.” Underpinned by violent social transformations, the data center represents infrastructural conduits that are eerily similar to colonial practices of the past. For instance, rural Mayo, Ireland, receives little to no cell phone service despite its proximity to a transatlantic fibre optic cable owned by Big Tech. They, therefore, represent exacerbating unevenness of the ongoing global structures of colonialism. This shatters the illusion of Big Tech-led technology acting as a civilising and universalising mission for the collective. It paints a picture of planetary expansion that “fails to account for the unequal relations sustained by and within [technologies] uneven reach into the patchwork territories of the earth.”
Another plane of extraction is that of human labor. The promise of automation relies on invisibilizing large-scale physical infrastructures, like the data center, and the labor required for generative AI to function. It relies on the “ghost work” of human annotators and moderators, lifting the veil to reveal the false promise of automation or ‘fauxtomation’, “the process that renders invisible human labor to maintain the illusion that machines and systems are smarter than they are”. The work is focused on training large swathes of data through data labelling or even algorithmic supervision. Notably, approx. 154 million to 435 million data workers come from the majority world. Operating outside the traditional employer-employee relationship, these workers often earn less than minimum wage and without social protections, such as paid sick leave or health insurance.
Big Tech’s tendency to pillage and impoverish the vulnerable has introduced another site of extraction—our bodies and daily intimacies. Once again, the extractive infrastructure is concealed and geared towards transforming the entire human experience into a source of profit, inextricably linked to ubiquitous Big Tech infrastructures, such as the cloud. This transformation or ‘datafication’ represents capitalism’s turn “inwards” to penetrate more layers of human life itself.
The data center represents a site of violent social transformation that is deliberately concealed to maintain the sheen of automation.
The data center represents a site of violent social transformation that is deliberately concealed to maintain the sheen of automation. It conceals the violence involved in the extraction of critical resources, from the mining of rare earth minerals to dumping industrial waste in community-held water bodies. People are now reduced to datasets and consumers of digital services, rather than right-holders. In an attempt to embrace inevitable technological progress, datafication and the ubiquitous cloud have permeated public service delivery. Eager not to leave any stone unturned, Big Tech’s relentless pursuit of wealth extraction comes at the cost of the most vulnerable, who are systematically impoverished of the resources required to imagine alternatives of digital, environmental, and collective justice.
The legacy of dispossession by colonialism continues. It ties into the technocratic practice of shifting the locus of power away from the collective into the hands of the few; one that has been years in the making. Sam Altman, the CEO of Open AI, illustrates this approach:
“AI is going to need a lot of energy, but it will force us to invest more in the technologies that can deliver this, none of which are burning the carbon … I still expect [that] unfortunately the world is on a path where we’re going to have to do something dramatic with climate, like geoengineering as a band-aid, as a stop gap, but I think we do now see a path to the long-term solution.”
Altman, like Amodei, reveals that Big Tech is geared to further increase its own power and scale, whatever the planetary cost may be. He tells an all-too-familiar tale, that the ecological crisis poses a technological challenge, and not a deeply political practice of extraction and exploitation. The carefully curated hype around generative AI is used, once again, to shift focus away from environmentally just practices. Microsoft unveiled its carbon moonshot in 2020, responding to the generative AI hype by shifting its goal to be carbon negative by 2030, more than five times away. Similarly, Google announced a 48% increase in emissions over the last five years. Evidently, there is a collective mission to address the growing ecological crisis associated with the data centers’ insatiable hunger and thirst. Altman reveals the technocracy’s approach—the “solution” or stop-gap measure, must create enabling conditions for Big Tech’s planetary expansion.
Evidently, there is a collective mission to address the growing ecological crisis associated with the data centers’ insatiable hunger and thirst.
“All we need to do as a public is have a little patience as our saviors in the tech industry perfect their techno-fixes so they can deliver us a digital utopia — which, it probably doesn’t need to be said, never actually arrives as those deeper issues just keep getting worse.” – Paris Marx, Tech Won’t Save Us
A discursive exploration of the data center represents the steady and continuous erosion of public autonomy to pave the way for seemingly apolitical solutions. The ecological salvation that AI claims to offer relies on its invisibility and deliberately disregards the broader set of material relationships associated with such pervasive infrastructures. Its seemingly magical and ephemeral quality is strategically leveraged to ensure there is little resistance to Big Tech’s techno-fixes.
Data centers as Visibilizing Infrastructures and Sites of Resistance
The metabolic rift that data centers exacerbate exposes how generative AI’s extractivism is not unprecedented. It remains “enmeshed in the colonial relations of power”, the accumulation of capital through unequal exchange of the colony. The same imperative of growth-through-dispossession now undergirds the logic of AI accelerationism put forth by tech billionaires, escalating from land and labor theft to the modern day enclosure of the data about our very selves within the cloud and consequently, the hyperscale data center of large tech corporations.
To examine the materiality of data centers is to visibilize the true costs of Big Tech’s stranglehold on capitalism, particularly its planetary consequences.
To examine the materiality of data centers is to visibilize the true costs of Big Tech’s stranglehold on capitalism, particularly its planetary consequences. By grounding our critique in the data centers that sustain AI, we expose the political ecologies hidden beneath the generative AI hype. The water crises, the deregulated labor zones, the lithium mines draining Indigenous lands are not ‘externalities’ but the logical endpoints of a system built on colonial logics of endless extraction.
The convenience bubble has burst. The harms are no longer distant. They are at the doorsteps of communities on the periphery of the colonial imagination, and they are coming for the rest of us, too. This is not a bug in generative AI’s development, but the core effect of continued dispossession. Therefore, without a critical appraisal of how the generative AI hype cycle further reinforces the “colonial matrix of power”, policy and research aimed at re-orienting access, governance, or development of AI will sit uncomfortably atop the uneven infrastructure which sustains it.
However, a better future is possible. Uruguayan campaigners in Canelones made clear how the intensification of data infrastructures is impacting socio-ecological webs of relations, drawing connections between the increased water shortages and the establishment of the data center with slogans like “No es sequia, es saqueo!” (It’s not drought, it’s pillage!) Once we recognize that the vision of AI solutionism, with its attendant harms, is not the only way forward, we can fight for a future where the boons of AI are equitably distributed, and the data center is a great place to start.
