By Nadine N. Sayegh & Youssef Elkady

This paper begins with a specific paradox: namely, how is it possible that the existence of the not-yet is already reshaping and reorganising the material conditions of the already present? Our focus in this article is examining the AI machinery under capitalist relations of production. It is clear that we should not think of capital and AI as two separate entities. If we look at the computational capacities that are being proposed, which are still at a hypothetical and unrealised stage, we must ask how a hypothetical possibility is nevertheless continuously anticipated and valued by private capital. Moreover, investors are continuously financing AI companies and acquiring contracts that are then built into infrastructures of energy, land, data centres, networks and labour. Therefore, our claim is that capital works precisely by capitalising on a future without requiring that the material reality corresponding to that projected future as being already realised. More importantly, suppose that an anticipated conception of a superintelligence model is already reorganising the present, we claim that the realisation of this conception can become futile and unimportant because capital has lost its capacity to produce. Capital perversely utilises the existing or newly built material apparatus constructed for that anticipated future and, when the future remains unrealised or fails, turns this apparatus into the engine of another projected future in order to keep the machinery of accumulation continuing. This future is always deeply embedded within accumulation and is, in fact, one of its clearest mechanisms. Therefore, we ought to take financialised forms of capitalist accumulation seriously without adopting the simple idea that finance or speculative valuations have become completely disentangled and decoupled from production. In fact, capitalist accumulation has always already involved and depended upon future valorisation. Blanchot’s useful formulation of “what would come without arriving” because it is precise in the understanding of the not-yet not needing to be a completed or even a ‘real’ event in order to organise and structure the present condition (1980; 1986). Perhaps there are similarities between a Blanchotian disaster and AI but the main idea we present is that the anticipated ambiguous technological future is pre-emptively operative and structuring. Robert Boyer’s idea of a fiction-led capitalism is especially interesting because he refuses to think of economic fictions as some pure descriptions of an existing reality (2020). The distinction that matters here is the continuation of capital production whilst becoming deeply embedded and structurally dependent on anticipation of future revenues by the financial market. In this sense, the reproduction of accumulation becomes unsustainable without being tied and relying on futures whose realisation remains uncertain and, in fact, maintaining a high level of return on capital depends upon the partial ‘unrealisation’ of these futures.

Before turning to the argument, one term needs clarification: extinction in this paper means social and structural obsolescence, the end of roles and categories that capital once required. It leaves the literal death of the species aside as a separate possibility. We will move from capital’s dependence on unrealised futures to three extinctions: those of the worker, of social roles and of the human as a privileged category. It ends with the libidinal dimension of the fantasy that surrounds them.

Capital as Directional but Purposeless

Capital is unbothered with the arrival of the anticipated future; it is solely preoccupied with the structuring of the present around the possibility that it will. In fact, interestingly, the realisation of a particular anticipation could have depreciating impacts on accumulation because, once realised, it is confronted with the costs of production, competition and other material constraints. However, failure allows the movement to continue because those assets can be repurposed and incorporated into newer forms of imagined realities. Capital, at our stage, with the fetishisation of accumulation, is dependent upon unrealised futures. A failed promise and an unrealised investment are significant for our analysis because, while the particular promise can fail at the level of the expected returns it promised, there are still remnants, such as infrastructural projects and technical capabilities, that will likely become significant for future promises. However, if a certain investment becomes actualised, there is little assurance that the initial speculative returns would be produced.

To reconcile this with the introduction, realisation can appear futile and unimportant because the decisive step has already happened: capital has been reallocated into energy and data centres on the strength of the promise. If the promised capacities then arrive, that reallocated capital meets immense pressure on competition and demand. Failure and realisation therefore carry different risks for the same investment. Essentially, it is a failed promise, but more accurately a false promise entirely. A failed future leaves its structures behind: data centres, energy contracts, land holdings and trained workforces remain once a projected future collapses, and they become the substrate for the next one. Capital therefore sustains itself on futures that fail as readily as on futures that arrive.

The structural claim we aim to make is that those failed promises are nevertheless reorganising the present structures and restructuring them under their promises. This reorganisation extends beyond physical infrastructure and investment to organising and appropriating labour itself. AI does not stop extracting labour after training, because every interaction with the model, from a query to an upload, becomes training data. The labour it appropriates is continuous and unwaged, and often given without the person knowing, so extraction carries on after the worker is displaced. Moreover, AI systems are continuously deriving economic value from user interactions and other forms of data monetisation. Those systems capitalise on data-mining strategies to continuously upgrade their models internally and for commercial purposes, often without compensation in traditional capital-labour wage relations (Zhang, 2026).

We therefore would like to turn to the abstract and inhuman nature of capital. It is obvious that capital should not be understood as some subject who pursues specific and concrete ends.

Capital is, put simply, a social relation that organises itself through particular modes of regulation and institutional structures around one imperative: its own reproduction and valorisation. It is directional but purposeless. In Luhmann’s understanding, the economy is organised through the distinction between payment and non-payment: payments connect to previous payments and enable subsequent ones, reproducing the economic system through a circularity that is operationally closed in its own communicative logic yet structurally coupled to other systems (2013). Needless to say without any subject (Luhmann calls them psychic systems) having the capability of comprehending or directing its trajectory. The problem for a theory of accumulation is how this recursive reproduction becomes articulated with the specifically capitalist logic of accumulation and the economic subordination to the reproduction of value. Capital’s directionality is emergent through the interdependence of processes whose rationalities are not necessarily coincidental but collectively reproduce a particular movement. Capital’s purposelessness should not be understood as if processes lack criteria of success (or that ‘actors’ lack purposes); it simply means that the continuation of accumulation cannot be justified by any final purpose internal to the process itself. Technological development is significant in our analysis because if capacities are realised then immanently those systems become under immense pressure to alter the conditions of competition and demand. This is why we believe that the superintelligent model is particularly important: if firms across the economy predict that superintelligent cybernetic systems will establish the determinants of productivity and, consequently, structure their market positions, then it becomes an existential prerequisite to continually expand their computational capacities in order to accumulate and ensure their survival.

This abstract, inhuman logic has a familiar form in the AI debate. Scholars in the Marxian tradition have read capital as an impersonal compulsion: value that valorises itself and uses people as its vehicles, a form of domination that no person or class wields and everyone obeys. Speculative theorists have pushed this furthest, treating capital as an intelligence arriving from its own future, and science fiction writers have made the comparison directly, describing corporations as slow AIs that pursue profit with no regard for human cost. The extinction scenario attributed to superintelligence resembles capitalism itself. Capitalism already subjects and exploits social life to an impersonal compulsion where labour as a productive category must be mobilised wherever it can serve valorisation, while those it cannot profitably employ may be rendered superfluous. In both instances to reproduce life there is an inherent subordination to a process whose continuation no individual human purpose can finally justify. AI is both a product of this process and its newest instrument. Capitalism already operates as a kind of general intelligence, not because it possesses a mind, but because it coordinates and reorganises social activity according to an impersonal imperative that exceeds any individual purpose. AI is its product and its newest instrument, and capitalism already operates as a general intelligence, one that needs no human purpose to continue. The anticipated future also has a material price, paid in the present by nature. Training and running AI at scale consumes energy, water, minerals and land, and the sector is built to expand that consumption. Marx’s metabolic rift names the break capital produces between social production and the natural conditions it depends on, a reading that ecological Marxists have since developed. Fictitious capital, modern-day ‘financing’, refers to titles to income that has not yet been produced, such as shares, bonds and credit, whose price rests on expected future returns. It is a claim on surplus value not yet produced. AI valuations can function in this manner: they may price the anticipated arrival of superintelligence and the profits it is expected to bring, rather than simply reflecting present production. This is the mechanism by which capital capitalises on the not-yet: the promise is traded and financed today, and it organises real investment in data centres and energy before any surplus value exists to back it.

Boyer’s notion of fiction-led capitalism is central here because it is obvious that an economic fiction’s force is dependent upon the construction of a narrative that would coordinate a degree of material action around a future that should remain untenable. The peculiar idea of the profitability and accumulation crisis is that investors, to sustain high and sustainable levels of accumulation, are obliged to mobilise immense levels of investment, resources, and funding pledges (especially when reproducing capital becomes structurally difficult), independently of whether those commitments become actualised (Boyer, 2020). What’s even more interesting is that continual investment moves from beginning as a speculative fictional decision to becoming a necessary condition caused by the previous investments. What started as a fiction or an ideational fantasy is transformed and gains a structural material force because resources and belief are dependent on and organised around that fantasy. However, when, in certain instances, those projects fail, the already existing infrastructural projects and debt obligations are themselves transferred into newer fictions of accumulation.

The Labour Category and Its Obsolescence

We will also turn to the labour category which has gained a lot of significance in the discourse of extinction especially in the US media discourse. Politicians such as Bernie Sanders have focused primarily on energy costs, job losses, and the extinction of the human race. Using Robert Kurz’s definition of labour we understand labour as a historically produced category that was once (especially in Fordist eras) critical in the reproduction of capital accumulation. We ought to move away from labour as a category with some intrinsic and essential eternal human essence towards labour being understood as a specific social form that is measurable and produces value under capitalist social relations. Therefore, when we mention the obsolescence of the labour category that does not mean the disappearance of the human activity. We understand obsolescence of roles as always being deeply embedded and in relation to the existing social systems. Similarly, we claim that superintelligence does not attack or destabilise an essential understanding of the human, instead, superintelligent models are a fatal attack to the historical conditions that made intelligence something essential or inherent to the human.

If labour is a historical social form, what does AI add to value? For Marx, machinery transfers its own value to the product and creates none, and only living labour produces surplus value. AI therefore generates no surplus value of its own. It is dead labour: accumulated human writing and images, along with annotation work, much of it unpaid or underpaid, stored in a system that runs on energy and infrastructure. The relative surplus value that firms may gain by deploying it, through cutting the labour time needed to produce, relies entirely on that appropriated human labour. The copy produces nothing that the labour embedded in it did not already supply. Extraction also continues past training, since the model keeps drawing on living activity every time it is used.

This leads to a contradiction at the centre of the argument. Capital still depends on labour, but a growing share of the labour it draws on, the human writing and annotation work embedded in AI, is taken without a wage. Capital has to produce surplus value and then sell the product to realise it, so it depends on labour twice: as the source of value and as the wages that buy goods. Each firm cuts its individual labour time below the socially necessary average to raise its own profits, and in doing so removes buyers that other firms need. Marx locates the problem in the gap between producing and realising value, and sets it out in his law of the tendency of the rate of profit to fall: more capital goes into machinery for each unit of living labour. AI intensifies the condition. If it replaces labour on a sufficiently broad scale, it can place pressure on both the source of surplus value and the wage-based demand that helps realise commodities. The introduction’s claim that capital has lost its capacity to produce therefore concerns value and surplus value. Production in general continues, and output can keep rising while the labour that creates value shrinks. This also explains why valuation rests on anticipated value, as the fictitious capital paragraph above sets out. The contradiction is structural and persists, since capital continues by consuming its own conditions.

The rate of profit can be expressed as surplus-value divided by the sum of constant and variable capital; if constant capital grows relative to variable capital and the rate of surplus-value does not rise enough to offset it, the rate of profit tends to fall. Marx lists counteracting influences (intensified exploitation, cheaper constant capital, expanded markets) so automation does not mechanically produce a falling rate of profit. The displacement of workers can place downward pressure on the total mass of surplus value, although the aggregate outcome depends on counteracting tendencies, including expanded production, changes in exploitation, new markets, and the creation of new tasks.

Exhausting Biological Futures

Capital depends on biology to function and reproduce. The worker must be produced before being exploited, and the consumer before being sold to; that production is carried out as unwaged biological labour, overwhelmingly by women, and it enters capital’s accounts as a free input. What the fertility figures register globally is that this input is contracting. Fertility has declined across many regions, although its timing and causes vary across countries. The relation between fertility and the cost of reproducing a life relative to wages is therefore a central hypothesis to investigate, rather than a sufficient explanation on its own: housing, childcare, employment insecurity, gender relations, public policy, and changing preferences also shape reproductive decisions. This creates a condition that is structurally produced rather than freely chosen, since the futures of reproducing a life are themselves being foreclosed.

This reframes the turn to AI. Capital may be shifting toward speculative futures not only because the technology has advanced, but also within a social order that places increasing pressure on the conditions of biological and social reproduction. A system that makes survival unaffordable undermines the conditions under which the next generation of workers can be socially reproduced. Capital therefore substitutes a speculative future for a biological one — data centres and algorithms in place of the workers and consumers who will not arrive. What the essay has called capital’s dependence on unrealised futures may thus be understood not as a preference but as a substitution made under duress, and the constraint is capital’s own reproduction. The not-yet is not chosen over the present; it is what remains when the biological present has been spent.

Three Forms of Extinction

In this paper we aim to focus on three forms of extinction. The first concerns the extinction of the worker as an economic producer because on one hand the majority of work becomes redundant in the pursuit of extreme productivity and on the other the worker as a consumer or buyer of

services become less central because if the purchasing power of that worker declines as a result of being detached from the productive process that would imply a structural crisis for capital that requires the possibility of capital reproducing itself through imagination and the generation of future expectations of anticipated realities or alternatively through high universal basic income(s) to sustain demand. In the former option capital becomes invested and organised around the possibility of the yet-to-come.

Secondly, there is the social extinction which pertains simply to the disappearance of the existing social roles and institutional functions.

The institutions built around wage labour lose their function as that labour disappears: the workplace, the union, the career ladder and the school that trains for jobs. Social reproduction, the work that renews people and society, such as unpaid care and domestic labour, is tied to these roles. When they lapse, the means by which people secure a livelihood and a place in society lapse with them.

The third transformation is the decentring of the human category as holding a privileged position over categories such as intelligence, agency, and purpose. The Superdark Factory developed by Poliks and collaborators offers one point of reference for thinking about these transformations. The third extinction concerns the category of the human itself. Humanist thought has traditionally treated the human as the privileged holder of reason and agency.

If machines match or exceed human performance at tasks once reserved for intelligence, that privilege loses its factual basis. Intelligence then reads as a capacity that systems can have, with the human as one bearer of it among others. The human loses its standing as the measure of intelligence and purpose. This rests on a thin definition of intelligence. It treats intelligence as performance at tasks, and on that measure machines can match or exceed people. Intelligence also includes judgment about which ends are worth pursuing, and current systems have yet to develop this capacity. Intelligence is also, and probably most centrally for this discussion, a somatic experience as well.

A model predicts patterns in human output and cannot weigh what that output is used for. The decentering of the human therefore holds only under the thin definition, and capital is what makes that definition dominant, since capital measures intelligence by what it can monetise. Capital is directional but purposeless, and the systems it builds inherit that condition: efficient at any end and indifferent to which one. Two of their best documented uses make the point. Their costs fall on ecosystems, and their targeting systems already kill with no human oversight. A machine that optimises without judgment exposes this narrow definition of ‘capitalist intelligence.’

We use Virilio’s account of the technological accident because his understanding of technological accidents does not derive from their being, in essence, external perturbations of some smooth and linear technical progress. There is a form of linearity, however, for Virilio, it is a structural condition inherent to the creation of a system. The accident is internal to the elements of the new system (2007). We claim that nothing ever escapes the internal logic of the system. We cautiously apply this to AI because what becomes interesting is the changes (or continuation, but at more extreme levels) in the speed, scale, and organisational logic of significantly important decisions. For example, if we look at modern military systems, such as those of the US, they have become preoccupied with the maximalisation of efficient target systems that both classify and automatically detect potential targets whilst limiting the time available for the inefficient, slow, and faulty judgement and reasoning of the human. The machine need not possess some ‘evil’ human intentions for actions to have catastrophic consequences. The technical organisation is the most significant aspect.

Although there are a few distinctions, we find a lot of commonalities when we look at the displacement and exteriorisation of repetitive functional duties and the decentring of human forms of intelligence because, if technical systems of decision-making exteriorise certain functions that were once performed through human knowledge and ‘judgement’, then we are required to question how those processes are governed and structured. The technical system does not by itself peripheralise and render the human obsolete because, while this is a major element, the human is made obsolete through orders and structures that evaluate capacities primarily through the lens of continuous accumulation and velocity. We reject that such processes and flows are reversible through some form of intentional human intervention. It is clear, however, that while consequences are not necessarily irreversible, the eventual endpoint can hardly be prescribed. After all, actualisation is not produced through determinate relations but through differential and contingent ones.

If we broaden the scope, these extinctions form a sequence. The extinction of the worker removes the economic function of most people. Social extinction follows as the institutions built around that function lose their purpose. The decentering of the human then removes the category that justifies both, the claim that people are the necessary bearers of intelligence and purpose. Each form reinforces the next, and together they produce the unnecessariness of the human that the next section treats as fantasy.

The Libidinal Fantasy of Extinction

We argue that the deepest extinction fantasy is not the dying of the species as a whole (a definitive possibility) but rather the actualisation of the unnecessariness of the human. We should therefore very much stress the libidinal dimension associated with this fantasy. Many theorists have always understood structural social transformation through the need to theorising catastrophe as the locomotive for structural change. However, we note that those singularities (catastrophes or events) that cause structural and systematic change have become almost inconceivable without becoming reproducible as non-events. We argue that the discourse around superintelligence provides a fantasy of a total and irreversible systemic transformation (with the human becoming extinct as the final ultimatum). There is a particularly strange jouissance to the idea of extinction. It surely points on one level to a total disaster but also the annihilation and the deep and perverse fantasy of the human category finally escaping the humiliating and restrictive limits of being a mere human. The unforgivable and extreme nature of this fantasy is the atonement for the nonexistence of any alternative exciting imagined futures. The peculiar form of technology that would once and for all liberate humanity from the need of liberating itself. This anticipated rupture is the ultimate prophecy to release the human project from the confines through which it came to be defined through. Once and for all, our limited species is freed from bodily finitude, dependence, labour, imperfect, incoherent, and faulty intelligence, and, most importantly, from what we argue is the immense responsibility of decision. Finally, the most significant aspect is the necessary obligation to remain within the confines and measure of what intelligence and agency can be.

Lacan’s jouissance refers to an enjoyment that exceeds pleasure and borders and is inseparable from pain, excess, and the transgression of the limits through which a particular subject is constructed. In this context it is a pleasure bound up with escaping the limit of being human, even at the price of ending the human. Extinction appears as an object of desire as well as dread, a wish to leave behind the humiliating limits of the category. Mark Fisher’s capitalist realism holds that it is easier to imagine the end of the world than the end of capitalism, explains why this wish takes the form it does. Extinction becomes a displaced fantasy of systemic change, since the end of capitalism is unthinkable. The jouissance of extinction is the jouissance of imagining an end to the present order, even when that end is annihilation.

The same fantasy explains why AGI and capital provoke one fear. Nick Bostrom’s paperclip maximiser, a system that pursues one metric with no regard for human ends, resembles capital as Marx described it, except that capital’s metric is its own expansion. The fantasy gives that abstract domination a face and a date, so it can be feared as a future event when it is a present condition. Quine’s web of belief supports the point on catastrophe: when evidence contradicts a system, it adjusts at the periphery and protects its core. The 2008 financial crisis is the standard case, a systemic rupture absorbed through bailouts that left the system’s basic structure intact. The fantasy or desire, a wish for total escape from the human limit, and the fear, as the dread towards the same thing. Both attach to one scenario, the first as promise and the second as threat, and both direct money and attention toward it. That gives fear an economic function and current AI doomerism feeds the process it warns against. Indeed, the sign, or rather the image, of superintelligence would circulate as an empty signifier, one that becomes associated with and reproduced as an emphatic promise, a fatal threat, a fictional valuative narrative, and a justification for the concentration of power by the producers of the technology, before producing any tangible realities (athough the promise of the technology has had tremendous effects on restructuring the meaning of what it means to be human) (Baudrillard, 1976; Laclau). Superintelligence is the most pervasive sign: it is the ultimate sign of the future, hence the excessive investments that become the promise of inevitability.

A warning that a technology could end the species confirms its power, and the claim can become a selling point: it raises valuations and gives the largest firms a case for being the ones to build and control it. The Y2K scare drove large-scale spending on remediation, and radio, television, the internet and social media each arrived with their own panics and each reorganised work and attention. AI belongs to this series. The paper takes no position on whether AGI arrives and claims only that the anticipation already acts. Fear works like the anticipated future in the main argument: it requires no realisation to produce effects. The fantasy of extinction and the fear of it circulate as one economic product.

We would further like to be extremely cautious because fear ought to remain conditional. It is very clear that existential-risk discourse surrounding AI and the extinction of humans is not simply intended to raise valuations or legitimise the concentration of power (although it does); however, it also serves as a motivation for states to fictionally believe in the possibility of regulating, restricting, or opposing particular models and AI systems. The claim we present in this article is much narrower and simpler because when the hysterical narrative of catastrophe produced by AI executives themselves as well as ‘whistleblowers’, we argue that this induces the idea that only a small number of firms possess the resources or expertise to control the relevant technology and limit its potential catastrophic fate, and hence, fear would allow the legitimisation of the concentration it warns against. The prophecy does not simply come true but murders the prophet. If AI companies (the producers of this magical technology) are the ones themselves declaring the coming and inevitable catastrophe, then surely they ought to be appointed as the guardians who would prevent the threat of the inevitable catastrophe (and in the meantime profiting from this position as the dominant power and limiting and concentrating power).

We see, quite clearly, the prophecy is self-fulfilling because it is a valuation event, not a prediction. The apocalyptic object is structurally empty (total, singular, arriving) which is what lets it function as a placeholder for a change nobody can otherwise imagine.

Conclusion: Desire as Productive

It’s very important to add that this fantasy of human obsolescence is itself economically productive. Deleuze and Guattari excellently claim in the Anti-Oedipus that, “if desire produces, its product is real. If desire is productive, it can be productive only in the real world and can produce only reality (1983, 26).”

Returning to the opening paradox: the not-yet reorganises the present because capital trades claims on futures that do not exist, and the resources committed to those claims remain when the claims fail. The labour contradiction explains why the bet rests on anticipation, since AI removes the source of surplus value and the demand that realises it. The three extinctions describe the leftovers and the fantasy of human unnecessariness supplies the desire that keeps it funded. If desire is productive, that fantasy is economically productive: it attracts investment and shapes present infrastructure and labour. The conclusion distinguishes two main senses of the future. One concerns the future as a claim on anticipated returns and anticipation that becomes priced and financed. The other is the future as a field of possibilities (forms of life and social organisation), that do not yet exist and may never become profitable.

The argument holds, and it has a limit and that is what we must understand. Autonomous systems and targeting algorithms already kill with no human oversight, so a dead thing is ending living people. A thing that never had life nor can it make use of life, ending life. Capital absorbs the loss of labour, since it finds another worker or another machine. What dies in war is alternate creation, the futures that were never going to be monetised. The paper has argued that capital sustains itself on futures that never arrive. It also forecloses futures that could have arrived. The not-yet is anticipated, and it is destroyed and it is a present condition.

But we would not find comfort retreating into this conclusion, since capital and all self-replicating systems with no end nor purpose cannot, thus far, exist without biology. Rather than seeking shelter in a nihilistic position, it is worth attending to what the extinction discourse cannot survive, which is that the doom it announces is foreclosing on itself. The apocalyptic form requires a singular and irreversible arrival, whereas the mechanism that funds it requires that the arrival never come; the promise is priced because it is not yet, and it remains priced only for as long as it remains not yet.

The bursting of the AI bubble is therefore not a refutation of the argument but its demonstration. In China the STAR 50 index has fallen by forty per cent in under four months, and retail investors in firms such as Unitree have watched their holdings halve. What collapses in such a movement is shareholder value and the paper claims on a future that was never produced, and what does not collapse is the material world on which those claims were drawn. Food must still be grown, moved and sold, and the people who do this are not the people who priced the future; the loss is redistributed upward to the conglomerates that overcapitalised and downward to those who bought the story, but the capacity to eat is not abolished with the valuation. If capital has lost its capacity to produce, and its power is confined to the register of valuation, then its collapse is confined to that register as well.

This is where the apocalyptic form makes its category error, since AI is powerful in the digital and only there. It is empowered by people and does nothing unless instructed. Its physical effects, from energy consumption to targeting, arrive through the people and institutions that deploy it. It reorganises attention, information, valuation and the movement of claims on things, and it destroys jobs, concentrates ownership and consumes energy, water and land at a scale that is genuinely destructive, but it does not eat and it does not grow food. It risks the status quo, meaning the wages, roles and institutional furniture of a society organised around employment, and the status quo is not the same thing as human life. The fantasy inflates a crisis of the status quo into a crisis of the species, and that inflation is useful to the firms whose valuations depend upon species-level stakes, since a doom that threatens everything justifies whoever claims to be managing it. Y2K and 2008 are the precedents: both were treated as system-ending, both ended in an accounting, and in both cases people continued to eat and work and find ways of living through an economy that was, by every measure, worse than before. The same holds here. Take biology away from capital and nothing remains to it, because once the labour that produces the worker, the consumer who buys the product and the hands that grow and move food are subtracted, what is left is a set of claims on a surplus that no longer exists. AI is the purest instance of this, since it produces nothing that the labour embedded in it did not already supply, it consumes energy and water in order to do so, and its value is a price set on a future that has not arrived; remove the biological substrate and the apparatus is a ledger with nothing behind it. Capital forecloses futures, but it does not foreclose the future.

References

Blanchot, M. (1986). The Writing of the Disaster (A. Smock, Trans.). University of Nebraska Press. (Original work published 1980.)

Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford University Press.

Boyer, R. (2020). From economics as fiction to fiction-led capitalism. In J.-H. Coste & V. Dussol (Eds.), The Fictions of American Capitalism: Working Fictions and the Economic Novel. Palgrave Macmillan.

Deleuze, G., & Guattari, F. (1983). Anti-Oedipus: Capitalism and Schizophrenia (R. Hurley, M. Seem, & H. R. Lane, Trans.). University of Minnesota Press. (Original work published 1972.)

Fisher, M. (2009). Capitalist Realism: Is There No Alternative? Zero Books.

Kurz, R. (2016). The Substance of Capital: The Life of the Dead in Capitalist Production. Chronos Publications. (Original work published 2004.)

Lyotard, J.-F. (1993). Libidinal Economy (I. H. Grant, Trans.). Indiana University Press. (Original work published 1974.)

Luhmann, N. (2013). Introduction to systems theory (P. Gilgen, Trans.; D. Baecker, Ed.). Polity Press.

Marx, K. (1976). Capital: A Critique of Political Economy, Volume I (B. Fowkes, Trans.). Penguin.

Marx, K. (1981). Capital: A Critique of Political Economy, Volume III (D. Fernbach, Trans.). Penguin.

Quine, W. V., & Ullian, J. S. (1970). The Web of Belief. Random House.

Virilio, P. (2007). The Original Accident (J. Rose, Trans.). Polity. (Original work published 2005.)

Zhang, F. (2026). The labor theory of value in the era of artificial intelligence and digital platforms: challenges, innovations, and new mechanisms. Humanities and Social Sciences Communications, 13, Article 717. https://www.nature.com/articles/s41599-026-07030-4

CNBC. (2026, October 29). China’s tech darlings are losing their shine. Here’s what’s behind the sell-off.

KuCoin News. (2026, October 28). Chinese stocks lose $2.4 trillion amid skepticism over AI profitability.