Over the years, a subject that was once confined to specialized discussions about the infrastructure of artificial intelligence has spread throughout society. If, over the past five years, the debate was, in a certain and fair sense, almost invisible to most people, today it has made its way into the news, influencers’ posts and comments, and even into the “real world,” at family dinners and in everyday social conversations.

One element that is worth highlighting for the purposes of the argument I intend to develop here is the specific discussion around the “materiality of the cloud in the AI economy.” This discussion, like the broader debate on infrastructure, is not limited to the economy of artificial intelligence, but it is within AI that it reaches a point of inflection and crisis.

Authors such as Kate Crawford (2021), Sasha Luccioni (2026), and others have long warned that the metaphor of the “cloud” serves very clear strategic purposes. The main one is to conceal the materiality of the digital economy, particularly that of artificial intelligence, which takes the harmful consequences of the exploitation of nature and people to an extreme.

From the precarious labor of data labelers in Latin America, Africa, and other countries in the Global South, all the way down to the roots of the extractive processes involving minerals and rare earth elements that underpin the industrial apparatus of data centers, the cloud turns out to have weight, substance, and blood.

If, as the “technology evangelists” proclaim, the cloud is a central infrastructure and essentially an ecological instrument with a light environmental footprint, examples from around the world abound to expose this as a fantastical and deliberately spectral discourse – one that conceals yet another episode of a process of colonization, now digital, imposing a new order of lords and subalterns upon the global economy – or, perhaps, reinforcing the old order under a new guise.

The denunciation of the “immateriality of AI” is justified. At IP.rec, we have already examined in detail the mechanics of this strategic discourse and what it deliberately leaves in a state of opacity. The global struggle, particularly over the past two years, has made this increasingly evident and brought it into popular discourse: AI operates on top of a massive infrastructure that extracts from nature while offering no clear social returns in exchange. Here, things unfold less as a method than as a kind of “tale”: looking at the spectacular posts on social media from profiles committed to private-sector lobbying, we see non-specialists reacting with concern and raising alarms about data centers.

The dilemmas surrounding energy consumption, the predation of water sources that should be prioritized for the population and for the production of social wealth, the emergence of heat islands around these facilities, and the noise and infrasound that harm wildlife, vegetation, and surrounding communities – all of this is already on the table and will become increasingly evident as the body of evidence accumulates.

Having said that, I would like to take this seriously and, so to speak, approach this extractive project from the other side: I want to talk about the superstructure, and about how discourses and ideologies drive a process of devastation of nature and territories.

I do so in order to show that discussing the worldview that runs through the world of big tech and the tech bros is not frivolous, and that this worldview mobilizes money and resources while corroding democracies around the planet through virulent lobbying apparatuses that deliberately undermine processes of building trust in our societies.

If denouncing the materiality of AI is necessary and always welcome, we must also remember the denunciation of the ideology behind it. And that ideology begins with the conception of the immateriality of AI as a mirror of the supposed immateriality of the mind – the metaphor of the “mind as a computer” (LAKOFF, JOHNSON, 1999).

This idea has been present in the field of AI since its origins and acquired more radical contours through the development of the so-called “cognitive sciences,” particularly in its first generation, known as computationalism. In my forthcoming dissertation, I analyze how authors in the technical field repeatedly flirt with this metaphor – and how legal scholars subsequently import this worldview uncritically (FERNANDES, 2026).

I do not intend to explore this subject in depth here, as it is both arid and time-consuming. But I want to state, as a premise, that the people involved in building artificial intelligence technologies believe – despite what they may say to the contrary – that what exists in a “neural network” is somehow analogous to what happens in the brain, within its neuronal web; that what exists in computational systems somehow corresponds to what happens in the mind; and that complex systems with emergent behaviors exist, as phenomena, in much the same way that human beings do – or almost do.

Words are treacherous, revealing at every turn what their users attempt to deny and evade. This is so true that the field itself has been unable to seriously and scientifically grapple with the body of metaphors it relies upon to operationalize its explanations.

The final stage of this process has become candy on the tongues of irresponsible children – the AI marketers: AGI, artificial general intelligence, and, beyond it, the singularity. And with them comes an entire field of ideologues who permeate the debate, from the more serious and committed figures, such as Marvin Minsky, to the more reckless ones, such as Nick Bostrom and Ray Kurzweil.

And it is only natural that, if they believe they can “create” a mind in computers, they should take the next step and imagine that they can “create life.” This, in turn, becomes linked to a genuine argumentative abuse that sustains two propositions: (1) that every effort toward this goal is legitimized a priori; and (2) that when they claim that the infrastructure economy of AI is clean, light, and “cloud-like,” this somehow becomes true.

Consider the comparisons with other fields: steel, oil, cars, aviation… all undeniably heavy. AI? Light, like the cloud and the mind.

They go even further: after all, humans themselves are heavier than AI, which is merely software – and this is where its proponents commit a banal, egregious error: arguing that there is “spirit in the software, whose body is the hardware.” This is where we encounter claims that it is less energy-intensive, from an energy perspective, to train a large model such as ChatGPT than what a human being consumes in order to learn something.

It is unsettling to see how these arguments spread, increasingly detached from reality. And yet they have a system, and a political purpose. Gebru and Torres (2024) have even given a name to this tangled, mysterious, ghostly, and spectral web of underlying ideologies: “TESCREAL” – Transhumanism, Extropianism, Singularitarianism, Cosmism, Rationalists, Effective Altruism, and Longtermism.

For the “evangelists” of this ideological web, there is a clear teleological roadmap: (i) reduce “intelligence” to verifiable performance, using a robust version of the Turing test as a (false, I say) test of functional humanity; (ii) identify the decisive algorithm for hierarchical pattern recognition through the reverse engineering of the neocortex; and (iii) promise the fusion of human and machine as the political and ontological culmination of technical progress.

The promise of controlling, overcoming, and recreating the human being as a machine – by the AI Overlords – is a driving force that mobilizes very concrete actions, and it is often overlooked even by those who rightly denounce the materiality of AI. In this case, ideology is not mere window dressing. It orchestrates a worldview: a particular conception of knowledge and of what can be done to the human being and to the post-human.

If the goal is divine, and if one believes that the Holy Grail of absolute control is within reach, then it becomes possible to discuss “a price to be paid.” For compulsive liars, people with highly volatile ethics and an overwhelming desire to have an empire of their own (HAO, 2025), the destruction of democratic society and ecological balance may seem like a perfectly reasonable price to pay.

The conversation here is about business and money, but it is also about ideas that are sold through doublethink – simultaneously truth and falsehood, belief and marketing – making the lives of all of us, day by day, a little more miserable.

REFERENCES

CRAWFORD, Kate. Atlas of AI: power, politics, and the planetary costs of artificial intelligence. New Haven: Yale University Press, 2021. 

FERNANDES, André Lucas. A tensão entre o conceito técnico de “inteligência artificial” e o conceito jurídico de “pessoa” : perspectivas a partir da história dos conceitos e das ciências cognitivas / André Lucas Fernandes, 2026.

GEBRU, Timnit; TORRES, Émile Phil. (2024). The TESCREAL bundle: Eugenics and the promise of utopia through artificial general intelligence. First Monday. v. 29, n. 4. https://doi.org/10.5210/fm.v29i4.13636. 

HAO, Karen. Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI. New York: Penguin Press, 2025.

LAMBERT, Katherine; LUCCIONI, Sasha. From cradle to cloud: a life cycle review of AI’s environmental footprint. arXiv, 2026. DOI: https://doi.org/10.48550/arXiv.2605.05416  

LAKOFF, George; JOHNSON, Mark. Philosophy in the flesh: the embodied mind and its challenge to western thought. New York: Basic Books, 1999.

Featured image: Gloria Mendoza / betterimagesofai.org / creativecommons.org/licenses/by/4.0/

André Fernandes