articles
issues
support
about
In Technology's Shadow Obsolescence and the production of surplus populations
Luis Cortés jul 2026 · essay
The Pitinga mine in Amazonas, Brazil. Photo: James Martins. CC BY 3.0.

In dedication to Juan Iñigo Carrera, militant scientist and comrade, who passed away at the beginning of June this year.

In March 2026, the artificial intelligence firm Anthropic published a report that attempts to estimate the impact of AI: both on the general labour market, and on specific sectors and professions. Their report examined a number of professional tasks, associated with around eight hundred unique occupations, ranking them by how likely they were to be reproducible by artificial intelligence. Perhaps unsurprisingly, the occupations likeliest to be affected were either those involving personal communication — such as marketing and sales — or those involving computer programming. All in all, this report paints a rather bleak picture of the near future: particularly for younger workers, currently in professional training or education.1

The report does not, however, point to a particularly novel phenomenon, but rather to one that has been a mainstay of capitalism since its dawn. Thus did Marx, commenting on the Luddites and their struggles, observe that it took both “time and experience”, before the workpeople learnt to distinguish between machinery and its employment by capital”, and to “direct their attacks, not against the material instruments of production, but against the mode in which they are used.”2 The apparent novelty of these advances today lies in that this technology now appears to artificially reproduce functions that were once reserved for conscious, “intellectual” activity.

Implicit in the discussion about technological development — but often not directly addressed — is the production of obsolescence. Obsolescence is what renders previous methods and processes outdated; either because they become expensive or inefficient, incompatible with newer techniques, below the average scale of industry, or a combination of these. Machinery, equipment, organisational structures and so on are “deteriorated”, not because they cannot do what they are supposed to, but rather by sole virtue of social relations, relations that seem to operate almost as a natural process. Indeed, this notion of technological progress is something so fundamental to capitalism, that it eventually led to the development of the idea of planned obsolescence, signalling a deliberate form of technological deterioration, beyond the usual unplanned, spontaneous obsolescence.3

The question we are faced with is: what is the eventual fate of the masses of proletarians that can only work on machinery that is rendered obsolete by technological development?

Today, more than a century after Marx & Engels wrote the Communist Manifesto, “all that is solid melts into air” continues to describe the continuous movement that animates the capitalist mode of production. The process of technological change materialises in a reduction of what Marx called the socially necessary labour time — the time required with normal means of production, employing an average time and intensity of work, to produce value.4 Individual capitals gravitate around this time. If they require less time in order to produce a commodity, such as by employing a new technological innovation, they can make an extraordinary profit by selling at market prices, simply because they incur lower costs than their competitors — thus appropriating what Marx called relative surplus value.5 Scientific research and its practical application through technology are an essential part of this process, in that they increase the productivity of labour. Eventually, novel technological innovations generalise within their respective industry, thus setting a new standard for productivity.

However, as labour becomes more efficient, there is a proportional decrease in the demand for it. The increase in the quantity of use-values produced by labour develops alongside the relegation of labour-power to smaller and smaller portions of investment, relative to constant capital.6 As a consequence, capital increasingly gains more and more leverage against labour, creating the conditions for a workforce that becomes a docile instrument ready to be exploited. The only thing worse than being exploited by capital, after all, is to not be exploited by it.

This dynamic also means that capitals cannot afford to stay still. If a capital produces slower than expected by the socially necessary labour time — such as by using outdated technology that requires more work, or more raw materials, etc., — it will struggle to stay afloat, finally getting pushed out of competition, either by going under or getting bought out. This is the form in which this social relation is enforced, making this a perpetual technological race.

The Austrian school economist Joseph Schumpeter glorified this as a process of creative destruction — as an industrial mutation that “incessantly revolutionises the economic structure from within, incessantly destroying the old one, incessantly creating a new one.”7 In this light, contemporary Schumpeterian economists like Daron Acemoglu and James A. Robinson claim that the economic success of societies is associated with their openness to creative destruction. Societies that fail, they tell us, do so because they establish obstacles that protect elite structures.8 But setting aside their sycophantic praise of capitalism, Schumpeterians get at least one thing right: technological development is not a choice, but an inevitable condition of capitalist society.

Since the microelectronics revolution in the 1970s, the computerisation and robotisation of large-scale industry have dramatically transformed global production, organising it under what has been termed the new international division of labour.9 This arrangement has implied a centrifugal distribution of production, mediated by a differentiation of the working class across nation-states.10 On the one hand, a scientifically specialised working class organises and administers parts of this process; on the other, this technification also produces a progressive deskilling of sections of the working class with experience-based practical knowledge. This knowledge, once concentrated in simple and technical manual labour, is today embedded in the functioning of machinery itself. The deskilling that is as a consequence of the general mechanisation of production has, in turn, ultimately enabled the proliferation of simple, non-mechanised labour as an external appendage of technified machinery: in, for example, manual assembly warehouses or sweatshops.

Consider the post-war automobile sector as a paradigmatic example of these developments. As with other global industries, the automobile sector organises its value chains across nation-states, enabling the increasing differentiation of the working class on the basis of their so-called productive attributes and the corresponding costs for capital.11

Within this division, lead firms tend to specialise in R&D, while manufacturing activities are outsourced to their suppliers.12 As the automation of production has moved forward through competitive pressure, manual labour has been pushed to the margins of vehicle assemblage, into tasks that have thus far resisted robotisation. These include tasks associated with the manipulation of non-rigid objects, such as wires and cables.13 The latest research in robotisation today tends to focus on artificially reproducing human fine motor skills: through, for instance, robotising biological tissue, through “necrobotics”, and through reproducing haptic perception and adapting AI to control it.14 The goal here is the elimination of this manual labour to the greatest possible extent.

Meanwhile, more expensive qualified labour has been concentrated in R&D. In the case of automobiles, this has resulted in a greater reliance upon semiconductors: upon systems that can regulate engines, sensors, and other components of the vehicle. Already in the early 2000s, software development accounted for 80% of all innovative activities within the automotive industry. As of 2017, 40% of the cost of an automobile comes from the electronic components and systems within it.15

This qualified labour is not immune to competitive pressure. Indeed, massive efforts have been invested in simplifying the R&D process, or in disqualifying its more expensive workforce. Development time and labour has been reduced thanks to the implementation of integrated design languages and of graphical simulation: automobile capitals, for instance, frequently use in-house physics engines in order to test materials, simulate wind resistance and so on, dramatically reducing the time required for the design and development of new models. Reports show that the integration of AI can dramatically reduce the time and resources it takes to do this: “It takes about 15 to 18 hours to run [crash performance], depending on complexity. We’re using probabilistic methods and artificial intelligence, and we can get that down to about less than one minute”.16 With numbers like this, and given the mass layoffs of well-compensated programmers at some of the largest global tech firms, it seems difficult to argue that AI is just an iterative tool with no ability for the “opening of new spheres of value creation.”17

Technological obsolescence is the obverse of this development of the productive forces, a byproduct of the constant drive for technological innovation under capitalism.

Consider South America, a region that has consistently demonstrated lower industrial performance than the global average. If we were to examine the regional automobile industry in the past two decades, the regional giants — Brazil and Argentina — have not only been under the minimum thresholds of production scale at a company and factory level, but have also been on the lower end of labour productivity and robotisation.18 Transnational companies, in particular German automobile capitals, have profited from this obsolescence in the past. Thus, in the 70s and the 80s, Volkswagen extended their use of deprecated machinery by at least a decade by exporting it from Germany to Brazil, taking advantage of lower wages and indirect transfers from the state.19

What do these indirect transfers look like? What are, that is, the conditions that make this sustained accumulation possible, given backwards or lower-than-average methods of production? The authors of the Quantification of the Historical Development of Capital in South America, following Marxist economist Juan Iñigo Carrera’s methodology, attempt to answer this question.20 The basis for the accumulation of capital within the region, according to their shared analysis, is the appropriation of extraordinary profits arising from a monopoly over natural conditions that are not reproducible by human labour: a category that Marx called ground rent. These profits are, in turn, disputed by different social subjects within a nation-state. Subsidies, exchange rates, and special internal prices for raw materials are some of the ways in which ground rent can flow through the economy, compensating capitals that cannot reach the average productivity of social labour.21 These are, that is to say, the ways in which junk machinery — which could only be employed at a loss in a fully industrialised economy — can remain profitable in the region.

This compensation arising from the exploitation of natural resources also explains the underperformance of the progressive Bolivarian industrial development initiatives. The social production companies and cooperatives in Venezuela under Chavez, as well as the local automobile sector, were never able to reach a scale that allowed them to survive without state subsidies.22 When Venezuelan production of heavy oil was displaced from global markets by Canadian tar sands and U.S. fracking, the “ultimate horizon of revolutionary Bolivarianism” finally collapsed, together with the price of oil.23 The deindustrialisation of Venezuela during and after the Bolivarian Revolution actually made it more reliant on imports, sustained by an overvalued currency: between 2001 and 2014, imports represented 75% of Venezuela’s GDP, greater even than the 56% in neoliberal Chile during the same period.24

South American economies thus reproduce themselves against capital’s historical tendency towards the development of the productive forces, apparently freed from the need to do so. Extractivism, then, is not a curse of the region; rather, natural resources are what keep these national economies from complete breakdown. And when the extraordinary surplus profit appropriated through ground rent shrinks, monsters appear. The most dramatic example is, of course, Javier Milei, whose devaluation policies and systematic welfare cuts at the start of his presidential period raised poverty by 11% in a matter of months, bringing more than half of Argentina beneath the poverty line.25 Although these numbers have subsequently declined, estimates that account for changes in consumption patterns — including increases in rent — show that 47.6% of the country still falls below this line. In total, 81% of the Argentinian population lives either in poverty or barely above the line.26 Yet Venezuela may still be the most extreme example of the effects that shifts in commodity prices can have. Consider, in the wake of the collapse of oil rents, the impact of devaluations on living standards: the minimum legal income in Venezuela, measured in 1998 prices, fell by 96% between 2007 and 2020.27 It was this ongoing misery that Nicolás Maduro’s regime was administering when he was kidnapped by the United States in January 2026.

Critically, this is not solely a result of bad manoeuvring by establishment peronistas, nor is it an authoritarian deviation from classic Chavismo. This is, rather, a structural feature of South American politics, in which political oscillations tend to follow raw material prices. South America thus collapses during periods of low prices, as it did in the “lost decade” of the 1980s, when the region was dominated by dictatorships. It thrives during commodity highs, the material basis for the “pink tide” that brought in Hugo Chávez in Venezuela, the Kirchners in Argentina, Pepe Mujica in Uruguay, Michele Bachelet in Chile, and so on. When this underlying pattern is not acknowledged, it can lead to reactionary political endorsements. For instance, consider the characterisation of Chavismo as a “centre of gravity for political resistance in the region”, or the “revolutionary political project that Venezuela has offered to the world”, or as a “fabric of popular resistance and our anti-imperialist spirit”.28 This — which unfortunately follows the standard discourse of an important part of the Latin American left, and of the various non-Venezuelan diasporas — invokes the nostalgia of the good moments of the oscillations of this pendulum, in order to defend the reactionary politics of specific national manifestations of the same pendulum.29 These kinds of political evaluations overlook that both extremes hang from the same axis: that of socially necessary labour time. Both depend on ground rent from natural resources, either to keep them afloat or to sink. Ultimately, they are never able to escape the law of value.

What does all of this have to do with Anthropic’s report? In some sense, Anthropic’s approach to the impact of AI on labour correctly reveals the relationship between the application of science to production and its effect on employment under capitalist relations. But these are simply the echoes of conversations that began almost two centuries ago; the difference being that the object today is artificial intelligence, and not weaving machines or fertilisers.

Mainstream analysis of this process is restricted either to national economies, mostly in Europe or the U.S., or to the direct replacement of the work itself — think service workers, graphic designers, developers, etc., being replaced by AI. But for every programmer left without work, there were already thousands of workers employed in other processes elsewhere, displaced from global accumulation. The redundancy of workers has simply expanded globally, to the point where entire countries and regions are becoming reservoirs of surplus populations for capital.

Capital institutionally organises — whether privately or mediated by the state — to ensure that workers are alive, relatively healthy, and have the specific skills to participate in the work processes that operate in the country. But when the ground-rent crutches that sustain obsolete capitals in these countries fall, the working class that depends on them faces a violent deterioration in its living conditions, quickly transitioning into becoming part of global surplus populations. The dismantling of welfare infrastructure and the violent deterioration of living conditions within a country, as is currently the case in Argentina, are often signs that global or national capital will no longer spend the resources necessary to secure the reproduction of a labour-power that is no longer needed, or that commands a far lower price than before. The Argentinian case is particularly telling of the helplessness of this situation. Even the relatively well-organised Argentinian working class, with its strong (albeit bureaucratised) trade union movement, seems to have had no leverage to resist the economic austerity imposed by the Milei administration.

The collapse in South America is, unfortunately, not the worst-case scenario. Depending on the specific dynamics of different natural resource sectors, a country can rapidly devolve into civil war, spurred by whatever power might profit from it. This is the case with Sudan, where the shift from a centralised oil economy into one focused on scattered, artisanal gold mining fuels the RSF.30 The fragmented struggle between ISIS, Kurdish forces and Assadists in Syria over oil and natural gas smuggling is yet another bleak example.31 Killing and mass displacement accelerate in these contexts: no capital needs these redundant populations. At the end of 2023, the number of displaced people in the world exceeded 120 million, growing from a ratio of one displaced person for every 125 people in 2012, to one in every 69.32

The problem is that even cheap labour becomes less useful to capital as technology further develops, and as the gap that natural resource rents must cover — either subsidising industry or welfare — constantly widens. What happens when all of this fails? In 2006, Mike Davis already saw the vast stagnation of populations in megaslums with no prospects for economic employment, a consequence of structural adjustment programs implemented by international financial institutions.33 However, when containment and displacement become a risk for ongoing accumulation processes that continue to employ an active working population, the masses of surplus humanity become a direct obstacle. It is at this point that associated capitals or the state that represents them organise their systematic elimination.

The condition of possibility for the genocide in Gaza is the consolidated surplus character of its population; the viciousness of Israeli apartheid is only its political expression. Since the complete closure of Gaza in 2007, the employment of Gazans in the Israeli economy has represented a steady 0%, according to the Palestinian Central Bureau of Statistics. A little before October 2023, the Israeli state launched a program to employ around 3,000 agricultural workers, who were sent back to Gaza when the genocide escalated. Technological development today makes Gaza not even worth turning into a giant sweatshop or call centre; nor do they have mines, oil fields or vast agricultural land to subsidise this.

Palestine is South America without extractivism.

Notes

  1. Maxim Massenkoff and Peter McCrory, “Labor Market Impacts of AI: A New Measure and Early Evidence,” 5 Mar 2026. [^]

  2. Karl Marx, Capital: Volume I, 1867, pp. 554–55. [^]

  3. Vance Packard, The Waste Makers, 1960; Tim Cooper, ed. Longer Lasting Products: Alternatives to the Throwaway Society, 2016. [^]

  4. The value of a commodity is thus not the result of the time employed by an individual worker, but rather of the time set by the social conditions of production. Following this socially necessary labour time is a precondition for the appropriation of an average or normal rate of profit. [^]

  5. In this case, surplus is appropriated not by compelling workers to labour for longer, but through the use of machinery in commodity production — resulting in a reduction in the labour-time necessary to pay for the worker’s own reproduction. [^]

  6. Marx calls this the organic composition of capital. [^]

  7. Joseph A. Schumpeter and Richard Swedberg, Capitalism, Socialism and Democracy, 2013. [^]

  8. Daron Acemoglu and James A. Robinson, Why Nations Fail: The Origins of Power, Prosperity, and Poverty, 2012. [^]

  9. Juan Iñigo Carrera, El capital: razón histórica, sujeto revolucionario y conciencia, 2013. Greig Charnock and Guido Starosta, eds. The New International Division of Labour: Global Transformation and Uneven Development, 2016. [^]

  10. Nation-states organise as reservoirs of nationally-coded working class populations with specific productive attributes. This is then regulated through visa regimes, interstate agreements, etc. [^]

  11. The productive attributes of workers, according to Charnock and Starosta, include the “strictly material or technical dimension of labour-power required by the particularity and complexity of the productive functions to be performed, as well as its ‘moral’ attributes”. On this basis, the differentiation is not a simple opposition between an industrialized “core” and a dependent “periphery”, as it involves the relative cheapness of specific productive attributes necessary in a production process. An eloquent example in Charnock & Starosta’s book is that of the Irish working class, which could be integrated into computer programming since they were relatively cheap workers who were also native speakers of English. See: Tomás Friedenthal and Guido Starosta, “The New International Division of Labour in ‘High-Tech Production’: The Genesis of Ireland’s Boom in the 1990s”. In Charnock & Starosta, eds. The New International Division of Labour. [^]

  12. Guido Starosta et al., “A Critical Look at Global Value Chains Competition and Globalisation in Contemporary Capitalism: The Case of the Automotive Industry”. In Guido Starosta et al., eds. Value, Money and Capital: The Critique of Political Economy and Contemporary Capitalism, 2024. [^]

  13. See: Sotiris Makris et al., “Automated Assembly of Non-Rigid Objects”, CIRP Annals, 2023; Huong Giang Nguyen et al., “Manufacturing Automation for Automotive Wiring Harnesses”, Procedia CIRP, 2021. [^]

  14. For technical descriptions of these methods, see: Xinzhu Ren et al., “Biohybrid Hand Actuated by Multiple Human Muscle Tissues”, Science Robotics, 2025; Te Faye Yap et al., “Necrobotics: Biotic Materials as Ready‐to‐Use Actuators”, Advanced Science, 2022; Sean T. Frey et al., “Octopus-Inspired Adhesive Skins for Intelligent and Rapidly Switchable Underwater Adhesion”, Science Advances, 2022; Benj Edwards, “Google’s New Robot AI Can Fold Delicate Origami, Close Zipper Bags without Damage”, Ars Technica, 12 March 2025. [^]

  15. Alexander Ziegler and Eckhard Heidling, “The Chip Crisis in the Automotive Industry”, ISF München, 2023. [^]

  16. Jonathan M. Gitlin, “From 15 Hours to One Minute: How AI/ML Is Speeding up GM’s Development”, Ars Technica, 1 June 2026. [^]

  17. Tabea Krug, “Speculating Our Way Through Crisis”, Disjunctions, 2026. [^]

  18. Alejandro Fitzsimons and Sebastián Guevara, “Wages, Price, and Profit: Protection and Value Capture in the Mercosur Automotive Industry”, Latin American Perspectives, 2023. [^]

  19. Luis Cortés, “Eine wiederholte Geschichte?”, Lateinamerika Nachrichten, 2025. [^]

  20. Javiera Rojas Cifuentes et al., eds. La cuantificación del desarrollo histórico de la valorización del capital en América del Sur, 2023. [^]

  21. These mechanisms are explained in thorough detail in Juan Iñigo Carrera, La Renta de La Tierra: Formas, Fuentes y Apropiación, 2017. [^]

  22. Thomas Francis Purcell, “The Political Economy of Social Production Companies in Venezuela”, Latin American Perspectives, 2013. For a description of the automobile sector, see: Fernando Germán Dachevsky and Juan Kornblihtt, “Desarrollo Automotriz Y Renta De La Tierra Petrolera En Venezuela: Expansión Y Crisis Durante El Periodo 1999-2016”, Regional and Sectoral Economic Studies, 2018. [^]

  23. Quote taken from Maga Miranda, “The US Kidnapping of Nicolás Maduro”, Spectre, 27 January 2026. See also: Juan Kornblihtt and Manuel Casique Herrera, “La Crisis Venezolana Como Expresión de La Sobreproducción Mundial de Petróleo Pesado”, Cuadernos de Economía Crítica, 2021. [^]

  24. Juan Kornblihtt et al., “Las alternativas al Neoliberalismo como forma de reproducir la particularidad del capital en América del Sur”, Pensamiento al Margen, 2016. [^]

  25. “Poverty Surges by 11 Points under Milei, Hitting 53%”, Buenos Aires Herald, 26 September 2024 [^]

  26. Alfredo Serrano Mancilla and Mariana Dondo, “La pobreza real en Argentina no es la que mide el INDEC”, CELAG, 20 March 2026. [^]

  27. Juan Kornblihtt and Casique Herrera, “La Crisis Venezolana Como Expresión de La Sobreproducción Mundial de Petróleo Pesado”, Cuadernos de Economía Crítica, 2021. [^]

  28. Quotes taken from Oliver Dodd, “The War on Drugs: A Pretext for Regime Change in Venezuela”, Jacobin, 10 January 2026; Miranda, “The US Kidnapping of Nicolás Maduro”; and Bloque Latinoamericano Berlin, “Against the imperial threat—our American union and resistance”, The Left Berlin, 6 December 2025. For diasporic defences of Maduro, see: Bruno Galvão, “As War Reaches Venezuela, We Must Support the Bolivarian Revolution”, The Left Berlin, 12 January 2026. For examples of positions taken by the Latin American left, see the Chilean Communist Party’s defence of Maduro’s electoral fraud in 2024: Joaquín Barrientos, “Lautaro Carmona señala que el Partido Comunista ‘no va a declarar del estado actual de Venezuela’”, La Tercera, 4 August 2024. As a clarification: while I of course agree with opposing the U.S. military intervention in Venezuela, I object to the analysis of Chavismo as a progressive political movement. This analysis obscures that the apparent external intervention of the United States was also an internal palace coup on part of forces disputing the appropriation of oil rents within the Venezuelan state. [^]

  29. Manuel Casique, “Communes and Crisis”, Spectre, 23 March 2026. [^]

  30. Ahmed Soliman and Suliman Baldo “Gold and the War in Sudan: How Regional Solutions Can Support an End to Conflict”, Royal Institute of International Affairs, 2025. [^]

  31. Hussein Almohamad and Andreas Dittmann, “Oil in Syria between Terrorism and Dictatorship”, Social Sciences, 2016. [^]

  32. Mid-Year Trends, UNHCR, 2024. [^]

  33. Mike Davis, Planet of Slums, 2006. [^]