Beyond techno-feudalism: A monopoly-capital analysis of artificial intelligence investment

Introduction: Artificial intelligence (AI) as an artificial investment

This article argues that the contemporary AI boom is best understood not as evidence that capitalism has been superseded by “techno-feudalism” but as a historically specific expression of monopoly-capitalist accumulation under conditions of chronic overaccumulation and profitability strain. Against the claim that platform “cloud capital” signals the emergence of a qualitatively distinct mode of production structured primarily by rent extraction and extra-economic domination, I argue instead that contemporary AI investment is best understood within the immanent dynamics of capitalism itself. By AI investment, I refer to a composite process encompassing: (1) venture capital and private equity financing of AI firms; (2) the equity capitalization of AI-intensive corporations on public markets; (3) large-scale fixed capital expenditures in data centers, semiconductor fabrication, and energy infrastructures; and (4) state procurement, subsidy, and strategic-industrial policy flows directed toward AI development.

Taken together, these interlocking circuits of finance, production, and state support operate chiefly as a surplus-absorption mechanism. They constitute a speculative, state-scaffolded outlet for surplus capital seeking renewed valorization under conditions in which sufficiently profitable opportunities in directly productive accumulation remain structurally constrained. In this sense, AI investment does not mark a rupture beyond capitalism but rather exemplifies the adaptive strategies through which monopoly-finance capitalism provisionally stabilizes overaccumulation by displacing it into capital-intensive, expectation-driven technological enclaves.

As such, I conclude that monopoly capitalism, as theorized by Paul Baran and Paul Sweezy (1966), provides a more precise and historically grounded framework for explaining why AI has become a privileged investment object, why its valuations can detach from realized productivity, and why the resulting bubble dynamics are systemic rather than anomalous.

The argument unfolds in a staged progression. It begins by (1) reconstructing the techno-feudalist diagnosis of contemporary capitalism, clarifying its central claims concerning rent, enclosure, and platform power, before engaging the principal Marxist objections.

Next, (2) the analysis returns to Marx’s law of the tendency of the rate of profit to fall and the associated problem of overaccumulation, stressing the distinction between the rate and the mass of profit and emphasizing that capitalist crisis tendencies are mediated through countervailing forces and displacement strategies rather than unfolding in a linear or mechanical fashion. On this basis, the article develops its central explanatory framework by demonstrating how Baran and Sweezy’s theory of monopoly capital provides a more adequate account of the contemporary AI boom. In particular, it shows how AI investment functions through three historically specific surplus-absorption mechanisms: (a) class-mediated consumption and the expanding “sales effort,” (b) investment demand driven by the imperatives of accumulation itself rather than demonstrated productive necessity, and (c) disproportionate sectoral expansion that concentrates accumulation within a narrow technological enclave.

Finally, (3) the argument then returns to techno-feudalism to show that, despite its rhetorical appeal, the framework is analytically misleading insofar as it mistakes capitalist crisis-management strategies for evidence of qualitative transcendence, thereby obscuring the underlying contradictions of accumulation that generate speculative excess and crisis displacement in the first place. Ultimately, this analysis concludes by situating these arguments within the concrete dynamics of the contemporary AI bubble, clarifying what it means to treat AI simultaneously as a productive force and as a surplus sink.

The political stakes follow directly: if the AI boom is structurally rooted in monopoly-finance capitalism, its likely “resolution” will take the familiar form of devaluation, consolidation, and intensified class power, rather than the emergence of a postcapitalist order.

From markets to fiefdoms: The techno-feudalist diagnosis

The increasingly fashionable thesis of techno-feudalism – and similar concepts alt-circulated under varying nomenclatures including exo-capitalism (Poliks and Alonso Trillo, 2025) and techno-rentierism (Birch et al., 2022: 7) – purports to diagnose a paradigmatic rupture in the contemporary mode of production whereby capitalist market relations are progressively displaced by quasi-feudal forms of domination (Varoufakis, 2023). On this account, value extraction is no longer primarily organized through competitive production and price-mediated exchange but through proprietary control over digital platforms, infrastructures, and ecosystems. Indeed, as the architect of this argument, Yanis Varoufakis proclaims “capitalism is dead,” having been superseded by a regime in which ownership of cloud capital (platforms, algorithms, data architectures, and networked digital environments) predicates itself on its capacity to extract rents rather than profits, thereby short-circuiting competition and subordinating market exchange to infrastructural command (Cadwalladr, 2023: interview with Varoufakis).

Interestingly, this argument is (in a way) implicitly Hegelian, insofar as it invokes one of dialectics’ foundational laws (later systematized by Engels, 1960), namely, that “quantity changes into quality, therefore an advance, when it reaches a certain size” (n.p.). The core claim, in effect, is that the quantitative escalation of technological throughput, data extraction, and computational scale has precipitated a qualitative rupture at the level of the mode of production itself.

On this account, proponents maintain that users and dependent firms no longer confront markets in any meaningful sense, but rather digitally constituted lords. Access to platform ecosystems is conditional and revocable, governed by unilateral rule-setting rather than reciprocal exchange; value is appropriated not primarily through commodity circulation, but through tolls, surveillance, and structurally enforced dependence embedded within proprietary infrastructures. As such, this soi-disant “technologically advanced form of feudalism” names a purported displacement of exploitation mediated by the wage relation by forms of domination mediated through digital enclosure(s), in which “markets are being replaced by digital fiefdoms” and economic power increasingly assumes the form of vassalage rather than contract (Singh, 2025: n.p.).

That said, the techno-feudalism thesis has been met with sustained and often sharp criticism from Marxist political economists.

Critics argue, first, that the techno-feudalism thesis conflates historically specific forms of capitalist domination with a transition beyond capitalism itself. From a Marxist standpoint, capitalism is defined not by the presence of competitive markets but by generalized commodity production, wage labor, and accumulation as the dominant social logic. As Ellen Meiksins Wood emphasizes, capitalism is characterized by the “imperatives of competition, accumulation, and profit-maximization,” imperatives that digital platforms do not abolish but reorganize and intensify (Wood, 1998: 97).

A second, closely related Marxist objection concerns rent. Against the claim that rent has displaced profit as the dominant form of surplus extraction, Marxist theory insists that rent (whether ground, monopoly, or financial) is a derivative distribution of surplus value rather than an alternative to exploitation at the point of production. ¹ As Brenner (2021) emphasizes, monopoly power does not negate capitalism but emerges from competitive accumulation and reorganizes it; the proliferation of rent thus signals advanced capitalist concentration and ongoing accumulation by dispossession, not a regression to feudal relations. Indeed, the contemporary process of digital enclosure identified by techno-feudalist advocates (through which previously open or weakly regulated spaces, practices, and resources are appropriated and governed via digital infrastructures) closely parallels the classical enclosures Marx analyzes as central to capitalist primitive accumulation of which advances through ostensibly lawful acts, such as the “acts for enclosures of Commons,” by which landlords effectively “grant themselves the people’s land as private property” (Marx, 1976: ch. 27). In other words, contra-techno-feudalist theses, rent remains a capitalist category, not a feudal one (Desai and Sachs, 2025).

A third Marxist objection similarly targets the claim that data constitutes an autonomous source of value. To sustain such a position would require succumbing to a paradigmatically Marxian form of fetishism, one in which the socially necessary labor time embedded in the production of data is rendered invisible and adopts a “mystical veil” (Marx, 1976: ch. 1). From a Marxist value-theoretical standpoint, data does not generate value autonomously. Its production, classification, and interpretation presuppose living labor, while its processing and deployment depend on energy-intensive infrastructures and large-scale fixed capital, themselves the historical sediment of exploitation. Thus, this material foundation is the product of “whole generations . . . oppressed and ill-treated by their masters, and worn out by toil” (Kropotkin, 1906: ch. 1), which have accumulated the conditions upon which contemporary production rests. Data, therefore, function analogously to constant capital insofar as it may enhance productivity, but it does not itself create value. As Christian Fuchs (2014) argues, digital capitalism extends exploitation into domains of communication and social cooperation while remaining fundamentally dependent on labor as the sole source of value. Thus, in this respect, platform capitalism intensifies rather than transcends exploitation.

Following, recent scholarship has considerably sharpened our understanding of digital capitalism, and any serious account of the contemporary conjuncture must engage this expanding body of work. Brett Christophers’ (2020) analysis of rentier capitalism, for example, demonstrates with precision how asset ownership and rent extraction have assumed heightened centrality in advanced capitalist economies, particularly through the juridical architecture of intellectual property and the infrastructural dominance of digital platforms. Cédric Durand (2017) extends this diagnosis by situating platform power within the broader dynamics of financialization, showing how digital firms increasingly operate through the capitalization of anticipated future income streams and the expansion of fictitious capital. In a different but complementary register, Jodi Dean (2005) foregrounds communicative enclosure, tracing how platforms capture, structure, and monetize social interaction itself, thereby reorganizing the political form of domination under conditions of networked circulation. Meanwhile, Nicholas Gane (2025), in a forceful critique of techno-feudalism, revisits Marx’s categories of rent and profit alongside Baran and Sweezy’s theory of monopoly capitalism, reminding us that “under competitive capitalism the individual is a ‘price taker’, while under monopoly capitalism the big corporation is a ‘price maker’” (366). Platform dominance, on this reading, reflects not a civilizational rupture but the intensification of monopoly power internal to capitalism.

Taken together, these interventions undermine the claim that digital platforms herald a transition beyond capitalism. Yet they focus primarily on the forms through which surplus is appropriated or capitalized. Less developed is the structural question of why the AI boom has emerged at this historical moment and absorbed such extraordinary volumes of capital despite ambiguous productivity returns.

This article, therefore, shifts the analytic focus from the morphology of digital rent to the macrodynamics of overaccumulation. Rather than treating AI as an extension of rentier income streams (Christophers), a paradigmatic instance of fictitious capitalization (Durand), a deepening of communicative enclosure (Dean), or simply evidence against techno-feudal rupture (Gane), it situates AI investment within the surplus-absorption framework of monopoly capital and reads it through Marx’s distinction between the rate and the mass of profit. The claim is that AI’s prominence cannot be explained at the level of ownership or governance alone; it must be understood as a historically specific configuration of countertendencies through which monopoly-finance capitalism manages chronic surplus and profitability strain.

The proliferation of feudal analogies, therefore, obscures the crisis tendencies internal to capitalism that render AI an attractive destination for surplus capital. What is required is not a vocabulary of rupture but a return to the problem of accumulation. Baran and Sweezy’s framework, read alongside Marx’s analysis of the tendency of the rate of profit to fall, allows AI investment to be understood as a determinate strategy through which mature capitalism manages overaccumulation.

From this vantage, the contemporary AI boom appears as a concrete instantiation of the surplus-absorption mechanisms identified in Monopoly Capital (1966: esp. chs 3, 5, 6, and 8). What they described as the growing difficulty of profitably utilizing rising economic surplus under monopoly conditions assumes a technologically specific form. (1) Demand is stabilized through elite and institutionally anchored expenditures, (2) investment proceeds less from demonstrated productive necessity than from the systemic compulsion to keep capital in motion, and (3) accumulation concentrates within a narrow technological enclave capable of absorbing extraordinary inflows. Under these conditions, AI functions simultaneously as a repository for surplus capital and as a mechanism of temporal displacement, deferring rather than resolving profitability pressures. Far from signaling transcendence, the AI boom reorganizes the contradictions of monopoly-finance capitalism without overcoming them.

Misrecognition and countertendency in monopoly-finance capitalism

From this optic, the techno-feudalist hypothesis rests on a misrecognition of the economic significance of platform rents, speculative valuations, and infrastructural monopolies. The present analysis does not simply deny that such phenomena signal the “death” of capitalism. Rather, it situates them within the historically specific dynamics of monopoly-finance capitalism, showing how they function as countertendencies that provisionally absorb surplus capital under conditions of persistent overaccumulation and declining profitability.

The central question is not whether digital platforms exercise extraordinary power but how that power is articulated within capitalist accumulation, how it mediates surplus absorption and displacement, and how it conditions the AI boom as a determinate response to structural crisis rather than a rupture beyond it.

The techno-feudalist literature is correct to identify digital platforms as key sites of enclosure. Yet the platform and AI nexus is more coherently understood as a configuration internal to capitalist monopoly rather than a revival of feudal relations. A small number of hyperscale firms control the principal data-bearing environments in which behavioral traces are captured, the cloud infrastructures required to process them, and the distribution channels through which AI applications are deployed and monetized. This concentration reflects structural dynamics whereby “network effects, combined with access to data and economies of scale and scope, have led to monopolistic trends and increased market power of the world’s largest digital platforms” (UNCTAD, 2021: 22).

The AI developers, therefore, confront not decentralized markets but privately governed infrastructures whose access costs assume the form of compute rents, licensing fees, cloud mark-ups, and platform tolls (Van der Vlist et al., 2024). This does not render AI firms feudal vassals. The relationship is structurally symbiotic within monopoly capitalism. Platform firms depend on AI innovation to enhance data capture and revenue streams, while AI firms depend on platform-controlled data, compute capacity, and distribution channels. What emerges is not extra-economic domination displacing commodity production but intensified interdependence among large capitals organized through infrastructural monopoly.

Data extraction within this nexus can be understood as a modernized primitive accumulation, appropriating social cooperation into proprietary datasets and training corpora. Yet this process does not abolish value production at the point of production or create an autonomous source of value. It restructures the terrain upon which surplus value is realized and distributed. Control over data-bearing activity secures strategic inputs for AI development, strengthens monopoly power, and enables the appropriation of surplus through rents. These rents remain derivative forms of surplus value generated elsewhere in the system and are intelligible only within the broader dynamics of capitalist valorization.

Platform rents, data enclosure, and AI development, therefore, form a differentiated unity within monopoly-finance capitalism. Platforms do not supersede capitalism; they reorganize its competitive structure by concentrating infrastructural control and shaping the channels through which surplus capital is absorbed. The AI boom unfolds not outside capitalism, but through monopolistic architectures that both condition and are reinforced by it.

Speculation as a symptom: AI investment under overaccumulation

The contemporary surge in AI investment has been widely cast by neoclassical economists as a technological revolution in utero: a general-purpose technology (GPT) ² poised to inaugurate a new regime of productivity – and, by extension, accumulation (McGeever, 2025). Indeed, across venture capital discourse, corporate strategy, and state industrial policy, AI is framed as the spearhead of a renewed growth trajectory capable of overcoming the stagnation that has marked advanced capitalist economies since the late 20th century (Andreessen, 2023). Yet the very intensity of this narrative (its insistence on imminent transformative gains despite mounting evidence of limited realized productivity) suggests that more than a straightforward technological transition is at stake (Brynjolfsson et al., 2017).

Accordingly, the AI boom should be understood neither as evidence of a “GPT driving a ‘Fourth Industrial Revolution’” (Crafts, 2021: 522; Schwab, 2024, as claimed by techno-optimists and neoclassical disciples, nor as evidence that “somebody killed capitalism” (Cadwalladr, 2023: n.p.) signaling a novel mode of production as asserted by techno-feudalist accounts. Indeed, it should be interpreted as a symptom of surplus capital’s growing difficulty in securing profitable outlets within productive accumulation.

In this respect, the present moment closely parallels earlier episodes of speculative expansion – most notably the dot-com boom of the late 1990s (Youvan, 2025) – albeit under conditions in which the spatial fixes that once absorbed surplus capital are increasingly constrained (as this document will later clarify). Rather than inaugurating a new mode of production, AI-centered investment intensifies capitalism’s promissory “accumulated claim[s], [and] legal title[s], to future production” (Marx, 1981: ch. 29) without commensurate gains in present output or surplus-value production. The result is a paradox in which unprecedented volumes of capital are mobilized in the name of productivity even as measurable productivity growth stagnates. This configuration reflects a familiar crisis dynamic: “capital never solves its crisis tendencies, it merely moves them around” (Harvey, 2010: 116), here through technological exuberance and “phantastical goals” (Desai, 2025: 7) that displace rather than resolve the contradictions of accumulation.

The stakes of this debate extend well beyond the diagnosis of a single speculative upswing. What is ultimately in question is the status of Marx’s law of the tendency of the rate of profit to fall under conditions of advanced financialization. Neoclassical critics, often invoking Okishio’s theorem, ³ contend that the persistence, or even the anticipated future expansion, of profitability in high-technology sectors calls into doubt the contemporary relevance of the law (Brynjolfsson et al., 2017; Freeman, 1998: 139–162). From this perspective, sustained or prospective profit opportunities in frontier industries such as AI appear to refute any general tendency toward declining profitability.

At the other pole, certain Marxist accounts respond by bracketing the specificity of Silicon Valley’s AI boom and the broader regime of financialization, treating these phenomena as largely epiphenomenal. In doing so, they preserve the law at a high level of abstraction but risk undertheorizing the concrete mechanisms through which it is refracted and mediated by contemporary configurations of the forces and relations of production (Dündar, 2024).

Against both positions, Marx’s distinction between the rate and the mass of profit becomes decisive. The rate of profit measures surplus value relative to the total capital advanced, whereas the mass of profit refers to the absolute quantity of surplus value realized. Crucially, as Marx demonstrates (1981: chs 13–15), the mass of profit may continue to expand even as the rate declines, particularly in periods marked by intensified accumulation, large-scale capital deepening, and speculative expansion.

Indeed, AI investment practices in Silicon Valley exemplify precisely this displacement. In line with Karl Marx’s maxim that capital dissolves “all that is solid into air” (Marx and Engels, 2023: 128), accumulation increasingly proceeds through the capitalization of future expectations rather than through proportional expansions in productive valorization. The consequence is not the suspension of the contradiction Marx theorized as the tendency of the rate of profit to fall, but its rearticulation. What appears empirically is less an immediate collapse of the mass of profitability than a systemic divergence between expanding claims on future surplus and the productive capacities required to sustain them, a divergence that increasingly manifests in “the slowing down of the overall rate of growth” (Foster, 2002, n.p.).

Although the analysis centers on AI investment, declining productivity reflects a broader structural condition of contemporary Western capitalism. Falling birth rates (shaped by rising costs of social reproduction and persistent gendered inequalities) intersect with intensified antiimmigrant sentiment to constrain labor supply and renew productivity pressures (UNFPA, 2025). These dynamics contract what Luxemburg (1972) described as the “reserve army of labour” (p. 50), even as Western states face mounting allegations of racialized labor practices from institutions that otherwise present themselves as liberal or universalist (Obokata, 2023). At the same time, deepening financialization compounds stagnation by binding real accumulation ever more tightly to fictitious accumulation, intensifying competitive pressures without restoring productive dynamism.

Within this conjuncture, AI occupies a privileged analytical position for three reasons.

First, unprecedented capital inflows are legitimated through projections of transformative productivity gains, even as high failure rates, uneven diffusion, and weak aggregate productivity growth cast doubt on both the rate and mass of realized profit. These projections function less as evidence of reorganized productive forces than as narrative and financial devices sustaining valuations and absorbing surplus capital (Ernst, 2022).

Second, as Cox (1983) cautions,

the concepts [in this case, declining rates of profit] cannot usefully be considered in abstraction from their applications [here, in an analysis of Silicon Valley’s AI investment model], for when they are so abstracted, different usages of the same concept appear to contain contradictions or ambiguities. (163)

Accordingly, secular stagnation is situated within AI’s historically specific “promissory exuberance” (Desai, 2025: 4) rather than being treated as an abstract tendency. Here, AI is understood as one determinate contradiction among others that, when mediated within a broader historical totality, helps clarify declining Western productivity (Gadamer, 1976). ⁴

Finally, and most decisively, AI offers a concrete instantiation of the three surplus-absorption mechanisms identified by Baran and Sweezy’s theory of monopoly capitalism. These interrelated processes, which “interact such that the overall impact far exceeds the sum of each part” (Whiting and Park, 2023, n.p.), may be reconstructed as: (1) class-mediated consumption, particularly elite and institutionally anchored demand, (2) investment driven by the imperatives of accumulation rather than realized productive necessity, and (3) disproportionate sectoral expansion concentrating accumulation within a narrow enclave (Baran and Sweezy, 1966) (Figure 1).

OPEN IN VIEWER

Figure 1.

Author’s illustration of the dialectical interaction between three countertendencies to the tendency of the rate of profit to fall, interpreted through the monopoly-capital framework developed by Paul Baran and Paul Sweezy.

The purpose of this analysis is therefore to demonstrate that contemporary AI investment within Silicon Valley operates through familiar capitalist mechanisms, thereby undermining the techno-feudalist claim that a novel mode of production, defined by feudal relations, has emerged. At the same time, these mechanisms (when situated alongside the tendency for the rate of profit to fall) also displace neoclassical and techno-optimist accounts that posit AI as the foundation of a new regime of productivity and, by extension, accumulation.

Overaccumulation, realization, and temporal displacement: Luxemburg to Sweezy

Marx’s tendency of the rate of profit to fall is a structural tendency rooted in the immanent logic of capitalist accumulation, not an empirical claim that profits must decline monotonically over time. In Capital, Volume III, Marx defines the rate of profit as the ratio of surplus value to total capital advanced, r = s/(c + v), where surplus value (s) is produced exclusively by living labor (variable capital, v), while constant capital (c), machinery, technology, and raw materials, transfers its preexisting value to the commodity (Marx, 1981: 317–75).

Under competitive pressure, capitalists are compelled to raise productivity through labor-saving technological innovation, thereby increasing the organic composition of capital (c/v). Because surplus value originates solely in labor, this rising capital intensity exerts downward pressure on the rate of profit even where output and the mass of profit expand. The tendency for the rate of profit to fall (TRPF) therefore names a contradiction internal to capitalism itself: capital depends upon labor as the source of value while simultaneously striving to displace it from the production process.

Crucially, Marx insists that this contradiction never manifests in a pure or mechanical form. Indeed, as Hardcastle (1960) observes, the theoretical problem is not explaining why the rate of profit falls but “of finding out why the fall is not greater and more rapid” (n.p.). Marx similarly writes that “there must be some counteracting influences at work, which thwart and annul the effects of this general law, leaving to it merely the character of a tendency” (Marx, 1981: ch. 14).

These countertendencies, including rising exploitation, the suppression of wages below the value of labor-power, the intensification of labor, the cheapening of elements of constant capital (c), accelerated turnover times, geographical expansion, and the appropriation of surplus through commercial and financial channels, do not negate the underlying tendency. Rather, they delay, displace, and reconfigure its effects across time and space, what David Harvey may later conceptualize as “spatial-temporal fixes” (2018).

This mediation underpins Marx’s insistence on distinguishing analytically between the rate and the mass of profit. Marx was fully aware that profitability could rise or fall conjuncturally and even pointed to instances of abnormally high profit rates under specific price and market conditions. The tendency of the rate of profit to fall, therefore, does not predict a continuous empirical decline in profits, nor does it imply that accumulation collapses mechanically in response to episodic downturns in profitability (Marx, 1981: ch. 14). Rather, it accounts for capitalism’s characteristic pattern of uneven expansion, speculative displacement, and recurrent crisis. The tendency thus explains why accumulation becomes increasingly dependent upon countertendencies and extraordinary measures, rather than why profits must visibly decline at every historical moment – although, as prior-noted, the mass of profits in the AI sector at the present conjuncture also remains negative; though not declining in relative terms (Challapally et al., 2025).

From this framework, Marxist political economy has long sought to specify the determinate historical forms these countertendencies assume, most notably through imperial expansion and the deepening of credit and finance. Rosa Luxemburg’s distinctive intervention was to theorize imperialism not as a contingent policy but as a structural outlet rooted in the problem of realization under conditions of expanded reproduction.

In The Accumulation of Capital, Luxemburg argues that surplus value cannot be fully realized within a “pure” or closed capitalist totality because internal sources of demand, workers’ consumption bounded by wages, and capitalists’ consumption and investment bounded by accumulation, cannot absorb an ever-expanding social product without contradiction. As she writes, the “realization of the surplus value . . . requires . . . that there should be strata of buyers outside capitalist society,” since surplus value “cannot be realized by sale either to workers or capitalists,” but only through sale to “social organizations or strata whose own mode of production is not capitalist” (Luxemburg, 1963: 351–352). Capitalism is therefore compelled, by the logic of expanded reproduction, to extend beyond itself, incorporating noncapitalist formations through colonial conquest, dispossession, market coercion, and unequal exchange.

Sweezy reformulates this problem by rejecting the claim that surplus-value realization is logically impossible within capitalism. Rather than construing realization as a formal contradiction, he conceptualizes it as a chronic instability internal to the process of capitalist reproduction. In The Theory of Capitalist Development (1942), he takes up Luxemburg’s central question, “where is the demand for the [over] accumulated surplus value?” (202), yet denies that this problem is insoluble in principle. On this basis, he concludes that realization does not depend exclusively upon spatial expansion through imperialism. Instead, realization may also proceed through temporal displacement. It can be sustained by the expansion of credit and fictitious capital, through “money that is thrown into circulation as capital without any material basis in commodities or productive activity” (Harvey, 2018: 95).

On this account, capitalism is capable of valorizing overaccumulated surplus internally through historically specific institutional arrangements, including the expansion of employment, credit creation, state expenditure, and the formation of new markets. Against Luxemburg’s conclusion that internal accumulation collapses into “a merry-go-round which revolves around itself in empty air,” Sweezy maintains that realization does in fact occur. However, it does so only contingently and at the cost of intensifying systemic instability rather than securing any durable equilibrium (Sweezy, 1942: 203).

On this account, financial expansion does not resolve the contradictions identified by Marx; rather, “temporal deferral . . . ‘fix’ the overaccumulation crises that arise from the chronic tendency of capital to accumulate over and above what can be reinvested profitably in the production and exchange of commodities” (Arrighi, 2006: 202). Crucially, Sweezy, in collaboration with Paul Baran (Baran and Sweezy, 1966), subsequently theorizes credit and finance as central mediating mechanisms of this displacement by identifying three historically contingent channels through which accumulation is sustained: (1) class-mediated consumption, (2) investment driven by the imperatives of accumulation rather than demonstrated productive necessity, and (3) disproportionate sectoral expansion concentrating accumulation within a narrow and increasingly insulated enclave of the economy (Baran and Sweezy, 1966). The AI’s Silicon Valley investment model, as argued here, exemplifies this configuration.

Waste as a stabilizer: The sales effort meets the AI adoption cycle

Sweezy and Baran’s first surplus-absorption mechanism, class-mediated consumption, is often misinterpreted as a narrow reference to elite luxury expenditure. In Monopoly Capital, however, the concept operates at a structural level. It denotes the expansion of surplus-absorbing outlays that are socially selective, institutionally organized, and only loosely connected to the production of new use values. The issue, therefore, is not simply that affluent strata consume more, but that a “qualitatively different thing . . . differing qualitatively from the preceding, the former state” becomes embedded within the accumulation process itself (Bukharin, 2013: 79–80).

Here, under monopoly capitalism, accumulation increasingly depends on expenditure streams whose systemic function is the realization of surplus value, where productive investment and mass consumption are structurally constrained. As Baran and Sweezy emphasize, the “normal modes of surplus utilisation” become persistently insufficient, such that “the question of other modes of surplus utilisation assumes crucial importance” (Baran and Sweezy, 1966: 13).

Central to this dynamic is what they term the “sales effort”: an expanded and systematized apparatus of market-making and demand engineering through which monopoly capitalism manufactures effective demand where markets do not expand organically. Conceptually, it is “identical with Marx’s expenses of circulation,” yet “in the epoch of monopoly capital, it has come to play a role, both quantitatively and qualitatively, beyond anything Marx ever dreamed of” (114). The sales effort is therefore not a peripheral appendage to accumulation but a historically specific countervailing mechanism. It “turns out to be a powerful antidote to monopoly capitalism’s tendency to sink in a state of chronic depression,” because it “absorbs, directly and indirectly, a large amount of surplus which otherwise would not have been produced” (131, 125). Its expansion marks a shift in competitive strategy away from productivity and price toward the organized production of perception, expectation, and market dependence, inseparable from capitalism’s “colossal capacity to generate private and public waste” (3).

The contemporary AI boom exemplifies this logic. Much of what is described as AI adoption does not reflect the generalized diffusion of demonstrably productivity-enhancing techniques, nor can it be reduced solely to what Desai calls “cultish hype about technology driving vast over-investment,” though it clearly exhibits that dimension and her Science-Finance-Fiction model remains analytically central (Desai, 2025: 6). Increasingly, AI adoption assumes the form of prestige expenditure aimed at signaling technological relevance and organizational modernity. Firms, universities, hospitals, and state agencies adopt AI less because it reliably raises output than because of “bandwagon effects, where firms rush to adopt AI-driven marketing solutions to remain competitive” (Ozturkcan and Bozdağ, 2025: 700). In this sense, AI functions as luxury infrastructure, a high-status organizational input whose legitimacy often exceeds its realized contribution to surplus-value production.

This sales-effort dynamic now extends well beyond conventional advertising. It increasingly assumes the form of organization-wide “digital transformation” programs, described by Sciuk et al. as “holistic and profound organizational changes,” which mobilize substantial expenditures on consultants, integration vendors, pilot projects, compliance regimes, procurement processes, “roadmaps,” and internal restructuring (2025: 5–31). Because digital transformation is a “moving target,” firms become locked into recurring cycles of “experimenting and piloting” that often fail to scale; indeed, “86% of digital transformation projects fall notably short of their objectives” (6–9).

Yet these expenditures remain economically functional. They generate contracts, billable hours, software revenues, and recurring service streams, absorbing surplus without requiring commensurate gains in labor productivity. In Baran and Sweezy’s terms, such processes constitute mechanisms of surplus utilization, organized channels for sustaining demand and economic activity when accumulation cannot advance smoothly through productive investment alone.

State-scaffolded speculation: Fictitious capital and the political guarantee

Baran and Sweezy’s second surplus-absorption mechanism, investment demand generated by accumulation itself, describes a configuration in which investment no longer responds primarily to expanding markets or realized productivity gains but becomes structurally self-referential. Under monopoly capitalism, the central contradiction lies not in the production of surplus but in its absorption. Economic surplus, defined as “the difference between what a society produces and the costs of producing it,” tends to rise as productivity advances under conditions of administered prices and restricted output (Baran and Sweezy, 1966: 9). Crisis thus emerges not from an inability to generate surplus, but from the growing difficulty of profitably reabsorbing it through productive accumulation.

Under such conditions, the “cessation of further accumulation constitutes the crisis situation, which Marx characterized as one of overaccumulation” (Mattick, 1974: n.p.). Financialization appears not as a contingent distortion but as a historically specific displacement strategy. As Foster argues, it entails a “shift in gravity of economic activity from production . . . to finance” (Foster, 2007: n.p.), functioning as a means of “kicking the can down the road” (Cutrone, 2017: n.p.) in response to this deepening crisis. Crucially, however, Foster maintains that this turn “falls short of a whole new stage of capitalism, since the basic problem of accumulation within production remains the same” (2007: n.p.). The point is not merely that narratives of technological renewal are overstated, but that they operate within a regime in which capital is “trapped in a seemingly endless cycle of stagnation and financial explosion,” displacing rather than resolving its contradictions (Foster, 2007: n.p.). The speculative valorization of frontier sectors, therefore, does not overcome crisis tendencies; it reorganizes them temporally by deferring the profitability test into an indefinite future.

The growing mass of surplus capital generates a systemic compulsion toward continual reinvestment, not to expand socially necessary production but to preserve value in motion. Investment becomes increasingly decoupled from demonstrable productive need and organized around anticipatory expectations, speculative narratives, and institutional supports that sustain accumulation without corresponding increases in surplus-value production, an accumulation process whose claims on future productivity appear “suspended in thin idealist air” (Ashley, 1984: 247). Investment demand thus becomes referential to accumulation itself, functioning as a temporary absorber of surplus under chronic stagnation rather than as a driver of material expansion (Figure 2).

OPEN IN VIEWER

Figure 2.

The graph shows that since the mid-20th century (especially after the 1990s), stock market wealth has grown dramatically and increasingly diverged from the much more gradual rise in fixed capital stock, indicating a widening gap between financial asset values and underlying productive investment. Reconstructed by the author based on data from Monthly Review Clark et al., 2021 and does not reproduce any proprietary graphical material.

The contemporary AI boom exemplifies this configuration. It corresponds to what Desai (2025: 14) describes as a condition in which “a senile capitalism is unable to muster enthusiasm about any project unless it combines vast amounts of capital, expectations for huge returns and huge state guarantees and subsidies.” The AI emerges not primarily as a response to demonstrable productive demand but as a demand-generating investment object, a vehicle through which capital mobilizes large advances on promissory horizons while deferring the profitability test. The issue is not the absence of productive applications, but the misalignment between the scale and timing of investment and realized productivity effects. Investment is mediated by surplus-absorption requirements that depend on credible narratives of transformation and institutional guarantees.

It is therefore unsurprising that accumulation here is increasingly mediated by “fictitious capital, detached from tangible assets, and thus detached from any socially necessary labour time” (Goodell Ugalde, 2024: 288). Capital becomes “twofold,” encompassing both “the ownership of real assets and also the holding of paper claims to those real assets” (Foster, 2007: n.p.). This sustains an “inverted relation” in which financial expansion no longer appears derivative of real prosperity but, much “in the way that even an accumulation of debts can appear as an accumulation of capital,” data accumulation likewise appears as capital accumulation (Foster, 2010: n.p.). The AI intensifies this inversion by converting projected productivity into present valuations and investment flows despite uncertain or temporally distant contributions to surplus-value production.

This expansion is not the spontaneous outcome of decentralized markets. It is politically scaffolded through the state’s integration into the financial architecture of accumulation. Within days of his second presidency, Donald Trump announced “a private sector investment of up to $500 billion to fund infrastructure for artificial intelligence, aiming to outpace rival nations in the [so-called] business-critical technology” (Holland, 2025: n.p.). As Foster notes, in monopoly-finance capitalism, “the role of the capitalist state was transformed,” becoming “fully incorporated into the system” as lender of last resort and guarantor of “too big to fail” policies (Foster, 2007: n.p.). Under such conditions, AI assumes the character of a quasi-public investment object, with valuations stabilized by expectations of procurement, subsidy, and strategic-industrial policy even where immediate profitability remains thin.

Further, within the monopoly-capital framework, military expenditure has long functioned as a privileged channel for surplus absorption when productive reinvestment and mass consumption prove insufficient (Baran and Sweezy, 1966: chap. 7). Postwar accumulation was “sustained chiefly through large military expenditures,” such that “only military spending could prevent economic stagnation in capitalist societies” (Vargo, 2021: n.p.). In the present conjuncture, this logic is rearticulated through the geopolitical militarization of AI. Casting AI as a national security imperative confers demand certainty upon an otherwise speculative horizon. In this configuration, “security” binds fictitious capital to state guarantees, legitimating extraordinary outlays on compute, cloud infrastructure, semiconductors, energy systems, and compliance architectures, while stabilizing revenue streams through procurement channels only weakly disciplined by commercial profitability.

The AI boom is thus increasingly tethered to a military-industrial–AI complex operating as a state-scaffolded surplus sink. The US defense cloud procurement has been institutionalized through the Joint Warfighting Cloud Capability framework, with a $9 billion ceiling through June 2028, enabling acquisition of hyperscale services “at all classification levels . . . from headquarters to the tactical edge,” concentrated among a narrow set of incumbent providers (US Department of Defense, 2022). Even where civilian diffusion remains uneven, security framing converts promissory productivity claims into contract-backed revenues. Pentagon agreements of up to $200m with major frontier labs further stabilize valuations and justify continued fixed capital expansion (Reuters, 2026). Because national security procurement privileges firms with hyperscale capacity and entrenched state relationships, it reinforces concentration within the AI enclave. Militarized demand thereby stabilizes the sector, deepens dependence on incumbents, and sustains subsidy and procurement regimes that defer the profitability test while binding speculative accumulation to geopolitical imperatives (Miller, 2022).

Bubbles as enclaves: How AI concentrates growth while stagnation persists

Baran and Sweezy’s third surplus-absorption mechanism, disproportionate sectoral expansion, names the tendency for accumulation under monopoly conditions to concentrate in an increasingly narrow enclave whose growth is out of proportion to the rest of the economy. The point is not simply that some sectors expand faster than others, which is banal. It is that, under conditions of chronic surplus and restricted outlets for profitable productive reinvestment, capital is repeatedly siphoned into a limited set of “leading” sectors that can absorb exceptional inflows through scale, market power, and institutional credibility.

Once again, the contemporary AI boom exhibits this mechanism with unusual clarity. One can see it first in the increasingly top-heavy structure of equity performance and valuation evident in that “since the beginning of 2023, the S&P 500 composite – the benchmark ‘market cap’ index increasingly dominated by the ‘Mag 7’ – has gained 67%, more than double the ‘equal-weight’ index’s 32%” (McGeever, 2025: n.p.).

This enclave structure is reproduced and deepened on the side of fixed investment. In 2025, analysts repeatedly observed that AI-related capital expenditure, operationalized as information-processing equipment and software plus the associated data center buildout, contributed an outsized share of measured GDP growth (Watts, 2025). Similarly, the St. Louis Fed, examining national accounts data, highlights how “in the first quarter of 2025, the contribution of [AI’s] information processing equipment (IPE) to real GDP growth jumped to 0.90 percentage points, which is more than two standard deviations above its long-run average” (Rubinton and Patro, 2026: n.p.).

At the same time, AI’s enclave character is revealed in the growing disjunction between the capitalization of anticipated productivity gains and the limited success of realized implementation, measured not only in the rate but also in the mass of profit. MIT-linked reporting captures this gap in the notion of a “GenAI Divide” separating widespread experimentation and piloting from sustained enterprise deployment and measurable productivity effects (Challapally et al., 2025). Whatever the limits of the underlying metrics, the substantive conclusion is clear: much AI adoption functions as a programmatic and reputational imperative, while effective integration remains uneven, contested, and largely insufficient to justify the scale of capital invested.

In monopoly-capital terms, this gap is not incidental but constitutive. It is the structural condition of enclave expansion, in which “the past decades have generally seen rising industrial concentration across many countries” alongside “shifts in the structure of production . . . away from manufacturing . . . to services, and notably industries of information technology,” developments that remain elevated precisely because the test of profitability is deferred into the future, socialized across portfolios, and buffered by institutional supports rather than disciplined by realized productive returns (Sawyer, 2023: 545–65).

This disproportionate expansion also has a determinate geography. The AI enclave is not reducible to “software” or “platforms.” It is a composite complex spanning semiconductor fabrication and supply chains, hyperscale cloud infrastructure, data center construction, grid and generation upgrades, and a surrounding ecosystem of contractors, consultants, integration vendors, and security and compliance industries. The buildout is therefore capable of absorbing surplus capital in multiple adjacent sectors simultaneously, creating what looks like broad-based “investment momentum” while remaining, in its driving logic, narrowly tethered to the credibility of AI’s promissory horizon. This is why the AI boom has the character of enclave-led accumulation rather than generalized expansion: it aggregates demand in a concentrated corridor of upstream and downstream industries while leaving the broader economy’s productivity trajectory stubbornly weak (Figure 3).

OPEN IN VIEWER

Figure 3.

Contribution of information processing equipment and software (IPE) compared with personal consumption expenditures (PCE) to real GDP growth (percentage points, two-quarter moving average, SAAR). The figure shows that, particularly from 2023 onward, IPE and software investment (used here as a proxy for AI-related capital expenditure) accounts for a disproportionately large share of GDP growth relative to consumption. This divergence illustrates the concentration of accumulation within a narrow technological enclave, where capital-intensive investment drives measured growth despite comparatively weak contributions from broad-based demand. The pattern supports the interpretation of AI investment as a surplus-absorption mechanism under conditions of overaccumulation rather than as evidence of generalized productivity-led expansion. This figure has been recreated by the author based on public data reported in U.S. Bureau of Economic Analysis (2026)

Automation, organic composition, and the recursive bubble

Neoclassical and techno-optimist interpretations of the contemporary AI boom rest on a fundamental misdiagnosis. Confronted with Solow’s paradox, ⁵ which “provided the first evidence of the paradoxical low return of technological progress to productivity” (Capello et al., 2022: 166), such accounts routinely invoke the lagged productivity effects of computerization and the diffusion of the Internet in the 1990s as evidence that apparently weak returns will eventually resolve themselves. In doing so, however, they conflate the mass of capital invested with the rate of return on that capital, mistaking the sheer scale of expenditure for productivity-enhancing accumulation. This misrecognition obscures a lesson already furnished by the dot-com boom. The Internet itself endured, yet the speculative cycle that surrounded it expressed deeper contradictions of accumulation. Accordingly, “AI will [also] survive, but the speculative bubble surrounding it is a sign of a deeper structural problem, the cost of which, when finally realized, will fall most heavily on the working class” (Goodell Ugalde, 2025a: n.p.).

Following techno-feudalist accounts, notwithstanding their oppositional rhetoric, ultimately reproduce a comparable analytical blindness. By construing the AI boom as evidence that “capitalism is dead” (Cadwalladr, 2023: n.p.) and has been supplanted by a rent-extractive regime of digital fiefdoms, such accounts obscure the contradictions internal to capitalist accumulation that generate speculative excess in the first place. They do so by adopting what “adheres to a distinctly unhistorical materialism” (Bieler et al., 2010: 1), severing contemporary forms of rent, monopoly, and enclosure from the historically determinate dynamics of capital accumulation. In treating platform rents, proprietary infrastructures, and data monopolies as indicators of a postcapitalist rupture, techno-feudalism mistakes historically specific countertendencies for systemic transcendence, reifying what its proponents themselves describe as the emergence of “economies with absolutely no limits” (Poliks and Alonso Trillo, 2025).

Against both views, the contemporary AI investment regime must be analyzed in its double determination. It is, on the one hand, a privileged outlet for surplus capital under conditions of overaccumulation. It is, on the other hand, a technology of production whose diffusion reshapes the conditions of valorization across the economy (Finio and Downie, 2024). This dialectical character is crucial. The AI simultaneously functions as an absorber of surplus capital and as a contributor to the profitability pressures that help generate renewed rounds of speculative displacement. It is precisely this dialectical role that explains why AI repeatedly emerges as a speculative sink within crisis-ridden accumulation.

Further, as a technology of production, AI is not introduced as a neutral efficiency gain but as a capital-intensive, labor-displacing innovation that systematically raises the organic composition of capital. Its deployment requires massive advances of constant capital in the form of data centers, energy infrastructure, and software systems, while compressing or eliminating living labor across administrative, cognitive, and professional domains (Robbins and Van Wynsberghe, 2022: 4829). Under these conditions, the “greater quantity of constant capital in production . . . and the relatively smaller quantity of variable capital” (Hardcastle, 1960: n.p.) intensifies the dynamic through which competitive accumulation undermines its own profitability. Even where AI generates relative surplus value at the firm level, its generalized diffusion reproduces a familiar outcome. Downward pressure on the rate of profit persists even as revenues, output, and technical capacity continue to expand.

The profitability implications of AI’s capital intensity sharpen further once Marx’s concept of moral depreciation is brought into view. Moral depreciation names the devaluation of fixed capital prior to physical exhaustion, as technical innovation lowers the reproduction cost of existing equipment or renders it competitively inferior. Marx is explicit that fixed capital is historically vulnerable in stating:

the instruments of labour are largely modified all the time by the progress of industry. Hence they are not replaced in their original, but in their modified form . . . [and] competition compels the replacement of the old instruments of labour by new ones before the expiration of their natural life. (Vol. II, ch. 8)

In the AI sector, where compute architectures, specialized chips, and training regimes evolve at extreme speed, this pressure is intensified: successive generations of hardware and model systems rapidly erode the competitive value of existing compute stacks (including newly built data center capacity) so that capital is driven into a continuous cycle of reinvestment, not simply to expand output, but to avoid technological downgrading and premature devaluation.

This matters for the present argument in two ways. First, accelerated moral depreciation intensifies the contradiction at the core of the TRPF dynamic. Ever-larger advances of constant capital are required, while the time horizon over which that capital can transfer value, and enable surplus-value realization, shortens. Second, the same process strengthens AI’s function as a speculative sink for surplus capital. The industry can absorb extraordinary inflows because its competitive structure continually demands large-scale replacement investment. The boom thus contains within itself an accelerated mechanism of capital destruction. The eventual correction need not await physical exhaustion or saturated markets. It can be triggered by a rapid revaluation of fixed capital when expectations about future returns fail to materialize.

This configuration can be summarized as two mutually reinforcing consequences. First, AI-mediated automation contributes to a secular tendency toward declining profitability across both AI-producing firms and downstream sectors that adopt AI as a labor-saving input, thereby exacerbating secular stagnation, as this article has already sought to demonstrate. Second, because AI displaces workers and disciplines wages, insofar as “there is clear evidence that recent automation, including AI has already led to job losses” (Aldred, 2024), it suppresses aggregate purchasing power and thereby intensifies the realization problem of “aggregate overproduction or lack of realization under capitalism” (Sherman, 1983: 206). Workers displaced or precarized through AI-driven restructuring do not constitute an expanding market for commodities. Rather, their constrained consumption deepens the stagnation that capital seeks to overcome through renewed rounds of technological investment (Figure 4).

OPEN IN VIEWER

Figure 4.

The dialectical interaction between AI as constant capital and as a siphon for overaccumulated capital.

This recursive dynamic assumes heightened significance under contemporary geopolitical conditions. Historically, overaccumulated capital has sought temporary relief through spatial fixes. As has been noted, “if, for example, a crisis of localized overaccumulation occurs within a particular region or territory then the export of capital and labor surpluses to some new territory to start up new production would make most sense” (Harvey, 2001: 23). Yet the Trump administration’s tariff regime, whereby “in the space of four months, January to April 2025, the average levy on imported goods rocketed from 2.5 percent to roughly 27 percent, the steepest hike (and highest level) since the 1920s” (Goodell Ugalde, 2025b: n.p.), restricts precisely these traditional outlets. Protectionist measures, export controls, and the geopolitical fragmentation of global supply chains increasingly constrain capital’s capacity to displace crisis outward. As a result, surplus capital that might previously have flowed into overseas production or foreign markets is instead resiphoned into domestic speculative circuits, further inflating AI valuations and deepening patterns of enclave accumulation.

With the spatial fix increasingly constrained, capital turns more aggressively toward temporal displacement. Here, the role of monetary policy becomes decisive. The Trump administration’s renewed pressure to cut interest rates, extending even to federal prosecutors opening a criminal investigation into Jerome Powell, underscores this shift. As Powell stated, “the threat of criminal charges is a consequence of the Federal Reserve setting interest rates, based on our best assessment of what will serve the public, rather than following the preferences of the president” (Federal Reserve, 2026). Should the Trump administration succeed in forcing rate reductions, framed as a stimulus to growth, the effect would be to make credit cheaper and more abundant, thereby encouraging investors “to take on more debt than they should” (Tranjan, 2023: 8). Cheap credit enables firms to roll over losses, sustain inflated valuations, and extend speculative investment horizons without resolving underlying profitability problems. It does not restore productive accumulation. Rather, it defers its reckoning by converting present stagnation into future obligation.

This temporal fix intensifies systemic fragility by encouraging the belief that credit “endows barren . . . and thus, money turns it into capital, without the necessity of its exposing itself to the troubles and risks inseparable from its [productive] employment in industry” (Marx, 1867: n.p.). In reality, it displaces these “troubles and risks” rather than abolishing them. Cheap credit inflates fictitious capital while binding accumulation ever more tightly to expectations of future surplus that may never materialize. The longer this deferral persists, the more disruptive the eventual correction becomes. When profitability expectations falter, the withdrawal of credit precipitates rapid devaluation, layoffs, and consolidation. As in previous crises, capital is not so much destroyed as reorganized, while livelihoods bear the brunt of adjustment.

The eventual crash of the AI bubble will therefore not signify the failure of AI as a technology, but the exhaustion of its capacity to absorb surplus capital. As I have argued elsewhere, “the destruction of capital . . . is rarely borne by those who hold it” (Goodell Ugalde, 2025a: n.p.). Workers instead confront job losses, wage compression, pension volatility, and austerity as states respond to falling revenues and financial instability. Surviving firms emerge larger, more centralized, and more capital-intensive, further raising the organic composition of capital and setting the stage for renewed overaccumulation.

In this light, the AI boom confirms rather than challenges Marx’s analysis. It demonstrates how capitalism manages declining profitability not through generalized productivity growth, but through speculative displacement, sectoral concentration, and the recursive inflation of fictitious capital. AI does not inaugurate a new regime of accumulation. It intensifies the existing one, functioning as a temporary pressure valve for surplus capital while simultaneously undermining the conditions of valorization that make its expansion possible. When this valve fails, the contradiction it has displaced will return in amplified form, reaffirming Marx’s insight that “the real barrier of capitalist production is capital itself” (Marx, 1981: ch. 15).

Conclusion: The real barrier of AI is capital itself

Ultimately, the contemporary AI boom is best understood not as evidence of capitalism’s supersession by a techno-feudal order, but as a historically specific configuration of monopoly-finance capitalism under conditions of chronic overaccumulation and profitability strain. Platform power, digital enclosure, and rent extraction do not signal a qualitative rupture in the mode of production; they operate as countertendencies through which capital provisionally manages its own contradictions.

Reconstructing Marx’s distinction between the rate and the mass of profit and situating AI investment within the monopoly-capital framework of Baran and Sweezy, this article has shown that AI functions through three interrelated surplus-absorption mechanisms: class-mediated consumption and the expanded sales effort, investment driven by the imperatives of accumulation rather than realized productive necessity, and disproportionate sectoral expansion within a narrow technological enclave. The AI thus operates simultaneously as a sink for overaccumulated capital and as a capital-intensive, labor-displacing technology that intensifies the very profitability pressures it is invoked to resolve.

If the AI boom is rooted in monopoly-finance capitalism rather than a postcapitalist rupture, its resolution will not be an emancipatory transformation but devaluation, consolidation, and intensified class power, with the costs disproportionately borne by labor. AI does not herald the end of capitalism; it reaffirms Marx’s insight that capital confronts its limits not by overcoming them, but by displacing them – until the displaced contradiction returns in amplified form.