articles
issues
support
about
The AI-Energy-Fascism Nexus On the value of knowing your enemy through their alliances
Anne Pasek oct 2026 · essay · issue 02
Honoré Daumier, "Gargantua". Public domain.

Too often, our tech criticism gets stuck in a form of analysis we might describe as the assholes critique.

Here, strategy and organising get sidelined in favour of gawking at the ideological freak show. We might start by asking: why are there so many damn data centres popping up right now? How do we stop the worst such projects from taking root in our communities? But this easily morphs into a different set of questions: Who in their right mind would need to build this much compute? Have you heard about these tech billionaires’ odd views on AI surpassing humans, about their enthusiasm for eugenic arbitrage, or their bizarre beliefs about the Anti-Christ?1 Thereafter — in group chats, on online platforms, in organising calls — things devolve into swapping links about TESCREAL philosophies, Elon Musk’s assorted perversities, and the remarkable stupidity of singularity bros.2

The assholes critique is certainly compelling. The worldviews gathering force amongst the billionaire classes are obviously important factors shaping the worlds we all share. There is real value in understanding why our enemies think the way they do, what their influences are, and the focal points of their desires. There are also surely libidinal satisfactions to be had in uncovering these at-times-cryptic motivations.

There are, however, limits to this approach. For one, it focuses a bit too much on individuals and ideas, jumping over the ways that even billionaires need to build coalitions across differences and manage conflicting goals in order to act on the world outside their homes. Typically, this involves the boards of publicly traded companies, the matryoshkas of political jurisdictions and regulators, and a public that at least nominally must be convinced not to follow through on those jokey guillotine memes.

Knowing what an individual believes doesn’t tell us much about what they can do. This is the key difference between biography and strategy. This is of particular importance when trying to account for the irrational exuberance of market actors engaging in what is almost certainly a destructive economic bubble.3 While a deep-pocketed individual might be able to gamble their ill-gotten wealth on profoundly under-elaborated bids to build a robot god, this is a lot harder for corporate board members, who have a fiduciary responsibility to maximise value for shareholders (at least from quarter to quarter). If they piss away Microsoft or Amazon’s stock value, shareholders can and will sue them to hell and back. Regulators, too, may well step in to prevent them from crashing the whole economy (at least election cycle to election cycle). What’s more, tech billionaires explicitly require the state to coordinate the conditions that allow them to keep importing and installing GPUs, sucking up more and more energy on increasingly unstable grids; and to continue to appropriate lands from a variety of political constituencies and biomes, in order to build their giant concrete server farms.

What the tech-freak CEO thus needs is less conviction in his own beliefs than a bridge between them and those of his allies. He needs to propose a bargain in which enough powerful interests see at least a temporary reason to join in the pursuit of AGI and data centre expansion. This is a tall order: the AGI project is obviously foolish, and the bubble is doomed to end badly.4 But perhaps there are just enough reasons for just enough partners in industry and government to form a coalition with this tech idiot, at least for part of the way.

Understanding these aligned rationales helps us better understand our enemies as they cohere across divergent goals and worldviews, all nonetheless opposed to left imaginings of the future. Many of them are indeed assholes, but their efficacy comes less from certainty in their own rotten accounts of social life than from their ability to build alliances and wield power.

When it comes to the data centre boom in particular, I have taken to calling these emerging alliances the AI-energy-fascism nexus, borrowing from the industry’s own terminology for the time-to-power chokepoint that curtails much of its ambitions, and adding the frequently obscured role of the far-right state. This tripartite alliance is a mutually legitimating one: each party gains something by associating with the other two and advancing the goal of data centre expansion, even if their broader ambitions are far from uniformly aligned.

This alliance marks a key pivot, both in material and discursive terms, in how the sector has tried to shape the norms and expectations for companies operating on an increasingly warming planet. Following these actors, this essay will focus heavily on the United States. The ramifications of this nexus, however, are global. We all share a stake in better understanding how the nexus works and how to thwart its integrity. May we do so as quickly as possible.

Firstly, it is important to remember that the interests of capitalists and fascists do not always automatically align. Indeed, it was the tension between the two that characterised much of the first Trump administration, when the tech sector was largely on team #resistance — playing a visible role in shaking their figurative heads at the breaking of norms, but also, to their credit, doing some modest work to build alternative forms of coordinated action in the wake of the federal government’s departure from the neoliberal consensus.

Energy and climate change are two interlinked areas where this was especially obvious. When the Trump administration signalled that it would withdraw from the Paris Agreement in 2019, big American tech companies were a large and vocal part of the We Are Still In coalition: an effort on the part of private businesses and local governments to meet U.S. climate commitments in spite of the federal government. Microsoft, Amazon, Apple, and Google also joined the fight when Trump cancelled the Clean Power Plan, a stalled Obama-era effort to decarbonise US electricity generation. They spent resources and took political risks speaking out for federal climate targets: this included public statements and drafting a shared amicus brief to the Supreme Court. They also poured millions of dollars into election financing and lobbying efforts to support the Inflation Reduction Act, electrification tax credits, and downballot climate initiatives, often lining up against fossil fuel funding.5 The tech sector also went out of its way to establish and support clean energy and carbon removal markets; this included a spectrum of variably credible financial actions, but it would be a mistake to dismiss it all as mere greenwashing.6 They were essential to spreading the concept of carbon neutrality in the 2010s and net zero in the 2020s. Meta, Amazon, Google, and Microsoft are far and above the largest corporate purchasers of solar and wind credits today, and have intermittently participated in civil society efforts to ratchet up standards and ambitions in these markets — often at the cost of their own procurement budgets and ahead of the appetites of corporate actors in other sectors.7

Why? A lot of discussion in this space gets caught up in the morality tales of sustainable business subjects and their associated language games, or the role that perceived green virtue plays in mediating access to tech talent.8 This is undoubtedly part of the picture, but a wider political-economic view suggests that vertical integration and platform monopolies may play a more decisive role.9 Tech companies are increasingly becoming energy companies, both in the internal structure of budgets and reporting lines, and in their acquisitions of new customers and subsidiaries.10 And unlike traditional industrial manufacturers, they are better equipped to make the leap to clean electricity and capture value by establishing the standards and “smart” infrastructures of that massive societal transition.

A year ago, I was convinced that we urgently needed a book on the formation of green data capitalism, accounting for the ways that the tech sector was set to be an ambivalent ally to the climate movement: sometimes adding its considerable lobbying muscle to struggles around decarbonization and climate adaptation, and sometimes setting up data-heavy solutions and platform monopolies that would compromise the movement’s urgency and equity concerns.11 The trick would lie in negotiating the terms of this wary alliance.

Would that this was the problem we are confronted with today. Instead, since day one of the second Trump administration, we’ve witnessed an abrupt about-face. The tech sector has been a highly visible part of his coalition, and millions of dollars have been donated to his proxies in ongoing moments of political theatre and grift.12 The sector itself, moreover, has failed spectacularly to keep up with its own climate targets. Instead of reducing emissions toward common net-zero goals, they are sprinting in the opposite direction. Microsoft is up 23% this year relative to its 2020 baseline; Google is up by 25%; and Amazon, which already has its own eccentricities in reporting, is up by 16%.13 These companies have also paused more ambitious forms of renewable energy and carbon credit procurements and have successfully lobbied standards organisations to water down criteria for everyone, after years of demanding the opposite course.14

Clearly, tech capital and authoritarian governments are no longer at cross purposes on the climate, at least in the ways that matter in marshalling political and economic power. This again merits explanation.

Why did their positions shift? The pivot to AI, I argue, goes quite a ways towards explaining things internally. Tech CEOs describe the AI race as an existential concern for their businesses, instituting a perceived gold rush in the production and use of parallel compute. The bigger-is-better ethos of OpenAI and Anthropic has also encouraged large cloud companies, looking to both compete with and supply frontier AI companies, to mortgage their climate commitments for an unparalleled feat of capital expenditure.15 This surely poses reputational challenges for the sector, which will struggle to ever position itself as a credible climate leader again. However, this does not necessarily mean that the sector can’t be wildly hypocritical in its business strategies, nor does it explain the visible and sudden fascist ring-kissing.16 Instead, all of this speaks to a realignment of business strategy and, importantly, to the alliances that the tech sector needs to execute that strategy successfully.

The crux of my AI-energy-fascism nexus theory, then, is that the AI energy rush has instigated a shift in the calculus of the tech sector as a whole; given new life to oil and gas interests facing existential challenges; and bolstered a far-right political coalition that desperately needs to shore up signs of economic prosperity. This association is mutually legitimating: all three parties get something from the other two and, in turn, capacitate them. We can think of this as a triangle — or perhaps a tensegrity structure if you’re an enthusiast for spatial metaphors — where all three components pull in different but complementary directions, supporting the reach and stability of the system as a whole. This rearrangement of political friends and forces is additive, transformational, and reinforcing, meaning that a deep understanding of any singular part can only get us so far.

The bigger question to ask is: what does each part do for the other? In the remainder of this essay, I will provide a preliminary answer across the many vectors in this alliance.

AI needs fossil capital. The compute needed for AI model development and inference is largely reducible to access to chips and energy. I’ll bracket the question of chips, which don’t seem to be currently throttling the sector’s ambitions, and focus exclusively on energy here.17

The story of frontier model development is one where more is more: a brute force approach to scale and complexity remains the dominant (albeit still unsuccessful) roadmap to commercially viable systems. This sets up the terms of the “‘AI race”: companies compete to add more compute to the pot until, eventually, AGI happens. Financially, this is a rather painful course, given that it now costs hundreds of millions of dollars in compute time alone to train a cutting-edge model. Nevertheless, the strategy succeeds at shaping the market in ways that advantage incumbent, well-capitalised players. In other words, we don’t necessarily have a race to build the best technology for a given task; we have a race to capture a market with a good enough tool for as many people as possible, and to ensure the continuity of monopoly platforms in doing so.18

As a consequence, the sector’s electricity demand is soaring. According to the International Energy Agency, global data centre energy usage is set to double in the next four years.19 Even this figure is a bit of a guess, however; we simply do not have a clear picture of how high AI energy demand theoretically is, since AI companies face ongoing challenges securing energy contracts in virtually all energy markets. Put differently, these companies want to buy even more energy to power even more data centres than they are currently able to. Energy is the big bottleneck in the AI race.20

This means that AI companies are not too discerning about where they get their megawatts; they just want them delivered ASAP. The urgency here stems in part from the perceived geopolitical competition of the AI race — if China builds the robot god first, it will worsen the American century of humiliation. More pragmatically, it also stems from the fact that GPUs have both higher energy requirements and a far shorter working life than conventional chips do.21 As soon as companies get their hands on them, they want to run them constantly in order to get a shot at making back their investment.

All of this implies a higher persistent demand for energy, coinciding with a flattening of the price differential between different kinds of generation sources on conventional markets. The cause of this is that all sources of cheap energy have been snatched up across the United States, and also increasingly in other parts of the world.22 This implies minor wins for renewables, but the overall benefactor here is methane gas, a phenomenon that was already visible in 2024 and 2025, in which producers saw record-breaking growth in U.S. markets, and which is predicted to continue.23 Demand has spiked to the extent that there is now a global shortage of new gas turbines.

Meanwhile, older fossil infrastructure is also finding new life: multiple reports detail how coal and gas plants have had their retirements deferred in order to meet new power demand directly linked to data centres.24 With characteristic impatience, the tech sector’s approach to building new facilities now routinely includes adding additional, behind-the-meter gas generators to meet demand that can’t otherwise be serviced by the existing grid.25 What began as a startling violation of environmental law and racial inequity, with xAI’s use of mobile gas turbines in a historically Black and environmentally burdened community in Memphis, has since become common practice.26 This has also altered the geography of AI data centres, with developers looking to more remote sites in other countries — such as Indigenous territories in Canada — where they can build both data centres and power plants directly on top of gas fields.27

Fossil capital needs AI. As one might expect, this surge in demand is also helpful to the fossil fuel companies that now play an ever-greater role in AI supply chains. In an era of diminishing possibilities for growth because of the seemingly imminent transition away from fossil fuels, data centres are a new business case for oil and gas majors, helping these firms build a stronger lease on the future.28 It is also an opportunity for the fossil sector to push back on environmental regulations and to score some narrative points within a larger culture war that they are always keen to keep kindled.29

For example, consider the Western Energy Alliance (WEA), a regional industry advocacy group that works on behalf of oil and gas producers in the interior U.S. Whilst oil majors are usually strategically quiet on all but investor returns, these smaller PR shops can be surprisingly candid commentators. Indeed, WEA’s account of the AI/energy crunch conveys both a smug sense of anti-environmentalist validation and a transparent effort to link the importance of the AI race to the future of fossil gas.30 Its VP, quoting investor projections, argues that in an era of justly expanding AI data centre demand, efforts to further decarbonise the growing grid will threaten the overall reliability and affordability of energy for everyone. If we don’t have a surplus of 24/7 power from sources like methane, that is, we’re going to see rising prices and brownouts.31 This ignores the fact that, in energy markets where data centre development is rushing ahead alongside parallel growth in fossil fuel generation, prices are rising anyways.32 This narrative attempts to wed the two industries’ interests together, such that seemingly urgent AI development can only be achieved through a robust fossil energy sector.

This realignment is also an opportunity to adjudicate the tech sector’s prior opposition to fossil fuel futures, recasting those efforts as naïve and insincere: “Publicly they push for renewables, but privately they’re locking in natural gas deals. Is reality setting in that renewables are not up to the task?”33 Here, the question of disconnected public and private speech mirrors critiques of “woke” culture — the imagined gap between what you’re pressured to say in public vs. what you really think in private. It’s also an opportunity to subtly reinforce anti-environmental, gender-reactionary frames about the deficient virility of solar and wind: only masculine, “firm” power sources can get the job done.

Fossil capital and the far-right need each other. In turn, these latent libidinal currents index the long-standing alliance between the energy sector and conservative — and increasingly revanchist — political movements. This vector in the nexus stretches back to the very beginning of climate denialism,34 and unsurprisingly played a constitutive role in fundraising for and staffing the second Trump administration.35 The resulting damage to climate action today is profound, almost certainly greater than credence in any given moment.36

This is also all a clear signal that fossil fuel interests are, more than ever, unable to operate competitively without the active participation of the state. They need a strongman to force through a regulatory environment and incentive structures that can artificially prolong their role in our energy mix.

This alliance and playbook, of course, predates Trump and the AI race. However, the AI race has a powerful role to play here in reciprocating some of these favours. Trump’s AI executive orders have been a vehicle through which to open up federal lands for both AI data centres and their accompanying energy infrastructures (so long as they are not powered by solar or wind).37 Announcements for data centre projects have likewise accompanied Republican campaign events in energy-heavy swing states like Pennsylvania.38 Rhetoric around “energy dominance” is now wed to American geopolitical imaginings of victory in their Sinophobic scaling contest with China: as an example, legal efforts to prevent xAI from continuing to illegally burn methane in Memphis have been denied on the basis that such data centres are national defence concerns.39

Energy companies have in turn been quick to legitimise the Republican narrative that global U.S. dominance in AI is both possible and desirable. Chevron’s CEO, for instance, describes how the company is “proud to play our part in bringing to fruition President Trump’s vision for a new American golden age, powered by our enormous energy resources”, and how “President Trump’s pro-American energy policies and commitment to energy and AI dominance give us the confidence to invest in projects that will create American jobs and strengthen our national security.”40 These comments, made in response to the release of the July 2025 AI Action Plan, are in turn hosted on the White House’s website as a legitimating nod from industry.

Perhaps the greatest rhetorical conjunction appears elsewhere in the AI Action Plan itself. Echoing Sarah Palin’s drill, baby, drill — a call to action on Arctic oil drilling — the White House today asserts, “simply put, we need to ‘Build, Baby, Build!’”41

The far-right needs AI. In such executive announcements and frameworks, the Trump administration accrues political benefits from being seen as the broker of new AI/energy partnerships, and thus a guarantor of national security and jobs. Yet AI is doing far more for the state than mere narrative work at the moment. It may, in fact, be disguising a recession caused by the government’s punitive tariff policies.

For a party that ran on inflation and economic growth as key issues, it has been exceedingly convenient that the efflorescence of AI capex came when it did. In 2025, American GDP was up 4%—1% above most predictions—amid a chaotic trade war. The steady state of aggregate economic growth amidst the demolition of long-standing geopolitical alliances appeased many financial actors and has prevented a widespread loss of confidence in the American dollar, at least so far.42 Yet these aggregate numbers disguise the fact that the outsize role of spending on data centres, chips, and energy all compensated for significant losses elsewhere in the economy. AI capex exceeded all consumer spending in the country during some financial quarters this year, papering over an economy in decline. Pithily summarised, “The US economy really is the AI economy now.”43

Beyond these macroeconomic favours, AI also plays a range of useful social and aesthetic functions for authoritarian leaders. Some of this comes down to creating a fig leaf for purposefully blunt and destructive actions, such as DOGE’s glorified keyword-search grant cuts, sanctified in part through the fiction that “AI” was doing it. In other instances, such as the integration of so-called AI into military decision structures, this amounts more directly to what Dan Davies calls an “accountability sink”: structures where responsibility for outcomes, or even the authorship of key decisions, gets intentionally obscured.44 This is immensely convenient when you are committing war crimes.45 And generative AI’s role in reinforcing the libidinal economies of fascism also cannot be understated.46 Increasing information pollution, a widening break from establishment media sources, a shifting aesthetic sensibility based on homophily,47 and a disdain for creative labour performed in the use and recirculation of generative outputs all play a role here.48

AI needs the far-right. Conversely, the many ways tech companies have benefited from having reactionary figures in power are clear by now. Nationally, the U.S. has seen a number of executive orders that aim to prevent individual states from passing regulations on the informational character or environmental performance of AI systems. This is likely illegal and unenforceable, but it helps slow efforts that might restrict the pace of the AI race and its related expenditures. Consider also how, within the first 24 hours of his presidency, Trump repealed a (purely voluntary) framework that had been set up by Biden for tech companies to disclose the environmental impacts of different AI models, allowing them to be more directly compared.49 This is a clear reversal of the green data self-governance model that tech companies had previously championed. The speed suggests this was delivery on a promise made to donors panicking about how to reconcile prior environmental commitments with the now seemingly infinite need to scale compute as quickly as possible. This is not deregulation; it is authoritarianism: these actions create discretionary and volatile policy environments that work to secure preemptive assent and performative collaboration.50

Internationally, too, it is clear that the Trump administration has been keen to exert political pressure on the EU to soften its approach to regulating AI companies, and their access to personal and nonconsensually gathered data for training AI models.51 This includes threats to levy more tariffs, punishing national economies for modest efforts to slow American model development and business cases.

This all exceeds the kinds of pressure that tech companies can exert alone, as well as what a conventional government might regard as a rational division between private benefits and public interest. But this has been true of AI tools for a long time now. The mass adoption of LLMs, sometimes discussed as industrial “diffusion”, increasingly relies on state interventions in public workplaces and private markets.52 Genuine enthusiasm for these tools, especially at prices that reflect their true cost, is minimal. Government mandates for agencies to incorporate AI into their workflows, often in contexts where the accuracy and privacy they offer are wildly mismatched to the use case, can be seen as a vanguardist effort. Continuing public subsidies for energy and land also further hold up the industry, all the while inflating the stakes of the bubble.

That bubble will eventually pop. The AI-energy-fascism nexus will not endure for decades. It may not even outlast the Trump administration. For all that each point in the triangle benefits from the other, there are also other forces and events that may break this alliance. For one, fundamentalist Christians are deeply distrustful of AI. Unprofitable data centre development projects, too, cannot continue to juice the economy indefinitely, particularly in the face of increasing local and international ire. Board members and lenders will eventually buck against deferred profitability. Unions are pushing back on the forced adoption of ill-fitting tools. The Luddites are on the rise. Forests are burning, heat waves are cooking us, and the climate continues on its deadly trajectory.

And yes, this is all indeed the work of assholes. Our understanding of these threats and opportunities in our organising is enriched by a deeper account of the ideological extremism of the individuals behind the AI race. However, our greatest points of leverage will ultimately follow from an analysis of AI as a cross-sectoral political-economic nexus — and of the many contradictions and fault lines that stretch across its vectors.

The task ahead of us now involves a lot of coalition mapping and making. Some friends we will make solely through our ideologically righteous accounts of our enemies’ moral failings. Others will come from more tentative alliances. Pope Leo’s encyclical on AI, for example, has been very helpful in reaching conservative Catholic councilmembers weighing votes on data centre moratoria in my corner of Canada. Concerns about data centre noise pollution, radiation, and fertility impacts are often exaggerated, but have nevertheless proven to be quite persuasive in many peri-urban communities.53 Farmers generally resent the loss of agricultural lands and landscapes familiar to rural life. Indigenous water protectors are mustering to the cause, taking data centres as the latest iteration of oil pipeline threats. Parents are highly anxious about chatbots causing cognitive declines and social corruption in their children. We have a nexus of our own to build.

While I’ve focused almost entirely on the American context here, there is a lot of localisation to be done in the way that AI articulates to fascist narratives and energy economies in other contexts. In Canada, for instance, this includes regional energy antagonisms: if Alberta can’t get oil to tidewaters, then it can at least “export intelligence” and stand-up markets for local gas by providing concierge services to data centre developers.54 It also builds on fears around national sovereignty, driving an ineffectual decoupling of compute from American territory, if not companies or chips.55 In other parts of the world, the AI-energy-fascism nexus will take different pathways, and may even involve greater participation from renewables, in different markets.56

In all cases, however, the near-term prospects of the AI race ultimately depend more on aligning capital and political interests than being ideologically persuasive. If the opposite were true, things would not have gotten nearly as far as they have. The longer this race is run, the more harm it will do, both in its progression and in its moment of violent dissolution. Let us therefore know our enemies not just as individuals or a sector, but through the breakable bargains they strike with others.

Notes

  1. 1. The AI term is vexingly imprecise and represents a barrier to precision in many conversations. However, alternatives also come with their communicative costs. I am begrudgingly using the term here, with assurances that we’re all generally aware that we’re discussing generative transformer tools/large language models (LLMs) most of the time, that none of this is actually intelligent, and that the scientists who just want to do machine learning on geophysical datasets can safely be left out of it.

  2. 2. Timnit Gebru and Émile P. Torres, “The TESCREAL Bundle: Eugenics and the Promise of Utopia through Artificial General Intelligence”, First Monday, 14 April 2024.

  3. 3. Luciano Floridi, “Why the AI Hype Is Another Tech Bubble”, Philosophy & Technology, 2024.

  4. 4. Servaas Storm, “The U.S. Is Betting the Economy on ‘Scaling’ AI: Where Is the Intelligence When One Needs It?”, International Journal of Political Economy, 2025.

  5. 5. Although they do have a track record of being outspent and frequently out-organized by fossil energy companies. See: “Big Tech and Climate Policy”, InfluenceMap, 2021; Lisa Stiffler, “Amazon, Microsoft Back Campaign against Initiative That Would Defund Climate Efforts in Washington”, GeekWire, 17 April 2024.

  6. 6. Anne Pasek, “Managing Carbon and Data Flows: Fungible Forms of Mediation in the Cloud”, Culture Machine, 2019.

  7. 7. Emily West, “Decoding Amazon’s Climate Pledge”. In Paul Smith, Alexander Monea, and Maillim Santiago eds. Amazon: At the Intersection of Culture and Capital, 2022; Emily West, “Performing and Directing Sustainability Theater: Amazon and Google’s Influence in Voluntary Carbon Markets”, Platforms & Society, 2027 (forthcoming); Michael Leggett and Peggy Kellen, “The Debate over Renewable Energy Certificates (RECs)”. On David Roberts, Volts, 10 September 2025.

  8. 8. Hunter Vaughan et al., “ICT Environmentalism and the Sustainability Game”, Journal of Language and Politics, 2023.

  9. 9. Silvia Weko, “New Sites of Accumulation? Why Intangible Assets Matter for Energy Transitions”, Review of Political Economy, 2025; Silvia Weko, “Digital Dependencies: How Google, Amazon, and Microsoft Reshape the Geoeconomics of Energy”, Globalizations, 202.

  10. 10. Anne Pasek, “From Atoms to Electrons: An Energy History and Future of Computing”, Digital Energetics, 2023.

  11. 11. For early thoughts to this end, see: Meg Weisser et al., “Green Data Capitalism and Its Rural Extractions”. In Patrick Brodie and Darin Barney, eds. Media Ruralities, 2026.

  12. 12. Diana Nerozzi and Javob Wendler, “Tech, Crypto, Tobacco, Other Companies Fund Trump’s White House Ballroom”, Politico, 23 October 2025; Clare Duffy and Alex Leed Matthew, “These Tech Leaders Donated to Trump. Now They’re out Billions of Dollars”, CNN, 9 April 2025.

  13. 13. Blake Montgomery, Nick Robins-Early, and Dara Kerr, “Big Tech’s Lofty Climate Goals Wrecked by Energy-Hungry AI”, The Guardian, 7 July 2026.

  14. 14. Gaye Taylor, “Big Tech Lobbying Watered Down SBTi’s Corporate Climate Standard, Critics Say”, The Energy Mix, 16 June 2026.

  15. 15. Nick Srnicek, Silicon Empires: The Fight for the Future of AI, 2026.

  16. 16. This hypocrisy on climate has arguably been enduringly true. See: Zero Cool, “Oil Is the New Data,” Logic Magazine, 7 December 2019.

  17. 17. See Ed Zitron, “Where Are All The Data Centers?”, Where’s Your Ed At, 12 May 2026.

  18. 18. Srnicek, Silicon Empires.

  19. 19. Laura Cozzi et al., “Key Questions on Energy and AI”, International Energy Agency, 2026.

  20. 20. Stephen Lacey and Nicola Phillips, “Energy Is Now the ‘Primary Bottleneck’ for AI”, Latitude Media, 22 May 2024.

  21. 21. Victor Tangermann, “Top Economist Warns That AI Data Center Investments Are ‘Digital Lettuce’ That’s Already Starting to Wilt”, Futurism, 21 November 2025.

  22. 22. Stephen Lacey, Jigar Shah, and Katherine Hamilton, “The Data Center Boom: ‘All the Cheap Power Is Gone”, Open Circuit, 21 February 2025.

  23. 23. Thomas Spencer et al., “Energy and AI”, International Energy Agency, 2025.

  24. 24. Valerie Volcovici and Laila Kearney, “Data-Center Reliance on Fossil Fuels May Delay Clean-Energy Transition”, Reuters, 26 November 2024.

  25. 25. These generators alone could soon produce as much in greenhouse gas emissions as the entire nation of France. See: Valerie Volcovici, “Gas Plants for US Data Centers to Be Major Source of Climate Change-Linked Emissions, Report Says”, Reuters, 1 July 2026.

  26. 26. Dara Kerr, “Elon Musk’s xAI Datacenter Generating Extra Electricity Illegally, Regulator Rules”, The Guardian, 16 January 2026.

  27. 27. Emma Zhao, “First Nation in Court to Challenge Proposed Wonder Valley AI Data Centre Project in Northern Alberta”, CBC News, 10 June 2026.

  28. 28. International Energy Agency, “Electricity 2024: Analysis and Forecast to 2026”, 24 January 2024.

  29. 29. Jordan B. Kinder, Petroturfing: Refining Canadian Oil through Social Media, 2024.

  30. 30. “The AI Boom: Why America Needs Natural Gas”, Western Energy Alliance, 10 March 2025.

  31. 31. “The AI Boom”.

  32. 32. Simply building the new grid transmission and substation infrastructures for AI facilities is a cost shouldered by all ratepayers in most markets; coal plant refurbishments are also highly cost-ineffective. See: Miguel Yañez-Barnuevo, “Data Center Power Demands Are Contributing to Higher Energy Bills”, Environmental and Energy Study Institute, 24 February 2026.

  33. 33. “The AI Boom”.

  34. 34. Riley Dunlap and Aaron McCright, “Climate Change Denial: Sources, Actors and Strategies”. In Constance Lever-Tracy, ed. Routledge Handbook of Climate Change and Society, 2010.

  35. 35. Owen Bacskai, “Fossil Fuel Industry Donors See Major Returns in Trump’s Policies”, Brennan Center for Justice, 21 July 2025; Alan Zibel and Toni Aguilar Rosenthal, “Trump’s Polluter Playground: Fossil Fuel Insiders & Ideologues Prop Up Dirty Energy & Derail Clean Power”, Public Citizen, 6 October 2025.

  36. 36. Nichola Groom, “A Timeline of Trump’s Moves to Dismantle the US Wind and Solar Energy Industries”, Reuters, 27 August 2025.

  37. 37. Michael J. Kratsios, David O. Sacks, and Marco A. Rubio, “Winning the Race: America’s AI Action Plan”, The White House, 2025.

  38. 38. Ben Geman, “Five Key Takeaways from the AI-Energy Summit with Trump”, Axios, 16 July 2025.

  39. 39. Tim De Chant, “DOJ Claims xAI’s Unpermitted Gas Turbines Are a Matter of ‘National, Economic, and Energy Security’”, TechCrunch, 16 June 2026.

  40. 40. “White House Unveils America’s AI Action Plan”, The White House, 23 July 2025.

  41. 41. Kratsios et al., “Winning the Race”.

  42. 42. Jim Edwards, “A huge chunk of U.S. GDP growth is being kept alive by AI spending ‘with no guaranteed return,’ Deutsche Bank says”, Fortune, 23 December 2025.

  43. 43. Jake Conley, “The AI Economy: Business Investment Overtakes Consumer Spending as the Biggest Driver of GDP Growth”, Yahoo Finance, 30 April 2026.

  44. 44. Dan Davies, The Unaccountability Machine: Why Big Systems Make Terrible Decisions—and How the World Lost Its Mind, 2025.

  45. 45. This is of course an important point for AI critics to keep in mind. See: Kevin T. Baker, “AI Got the Blame for the Iran School Bombing. The Truth Is Far More Worrying”, The Guardian, 26 March 2026.

  46. 46. See: Roland Meyer, “Echte Emotionen. Generative KI und rechte Weltbilder”, Geschichte der Gegenwart, 2 February 2025.. Of course, this influence goes both ways, with the Trump government awarding large contracts to AI companies willing to make ‘anti-woke’ products (which is to say, those that accord with the reactionary gender and race politics of the far-right). This inflates a market for what might otherwise be a very niche product. Unsurprisingly, xAI is a clear early benefactor of this policy.

  47. 47. Wendy Hui Kyong Chun, Discriminating Data: Correlation, Neighborhoods, and the New Politics of Recognition, 2024.

  48. 48. Gareth Watkins, “AI: The New Aesthetics of Fascism”, New Socialist, 9 February 2025.

  49. 49. The White House, “Initial Rescissions Of Harmful Executive Orders And Actions”, 20 January 2025.

  50. 50. Alondra Nelson, “The Mirage of AI Deregulation”, Science, 2026.

  51. 51. Jennifer Rankin, “EU Could Water down AI Act amid Pressure from Trump and Big Tech”,The Guardian, 7 November 2025; Eliza Gkritsi and Jacob Wendler, “Trump Threatens ‘Substantial’ New Tariffs against Countries with ‘Discriminatory’ Digital Rules”, Politico, 25 August 2025.

  52. 52. Matt Seybold, “‘You Have to Use It. You Have to Trust It.’: Forced Adoption of AI Is The Subtext of Davos”, The American Vandal, 23 January 2026.

  53. 53. This is certainly another example of eco-fascism on the rise. I aim to write more in the future about the ways ecological concerns are an often-imperfect container for a wider range of AI refusals.

  54. 54. Nate Glubbish, “Data Centres in Alberta”. On Tim Henley, Alberta at Noon, 16 July 2026.

  55. 55. “Canada Finally Has a National AI Strategy. Experts Hate It”, The Walrus, 5 June 2026.

  56. 56. Examples of this include connections between the BJP and wind developers in India. See: David Singh, “‘This Is All Waste’: Emptying, Cleaning and Clearing Land for Renewable Energy Dispossession in Borderland India”, Contemporary South Asia, 2022. I thank Rahul Mukerjee for this point.