https://mronline.org/2026/09/21/where-do-superprofits-come-from-on-the-nature-of-digital-capitalism/
~~ recommended by tpx ~~
Today, the profits of publicly traded companies around the world are soaring. We have to go back to the 19th century and its American “robber barons” to find profits comparable to those generated by the internet giants—with their subsidiaries focused on AI—and to those forecast by firms specializing in AI.[1] They are projecting even more staggering profits, thanks to investments so colossal that they would exceed the GDP of many countries—and by no means the smaller ones. We have reached the point where some AI gurus, without calling these profits into question, are concerned about their consequences, fearing widespread social upheaval.[2]
But where do these superprofits come from? This point alone is enough to invalidate neoclassical economics, at least in its most theoretical formulation, since once the “labor factor” and the “capital factor” have been paid their fair price, profit should be zero.
More importantly still, the same point seems to definitively call into question the Marxist theory that profits are generated by labor, on the grounds that there is less and less need for labor as it is increasingly replaced by robots, including perhaps by humanoid robots, and by AI, particularly in intellectual work.[3]
As we shall see, nothing could be further from the truth. Establishing this, however, presupposes a solid understanding of Marxist theory and an analysis of the new forms of capitalism. But first, let us recall how capitalism evolved before becoming digital.
Shareholder Capitalism
Marx had already drawn attention to the revolution in capitalism brought about by joint-stock companies. Capitalism was no longer simply employer-driven; it was becoming collective. For a long time, large corporations with a broad shareholder base coexisted alongside small businesses with a more limited shareholder base. But because they were scattered, individual shareholders were more like rentiers, facing all-powerful managers. This was even true of larger shareholders during the era of “managerial capitalism” after the Second World War. When powerful investors took control of large corporations, they began demanding ever-higher returns. That is when finance seized power. And the overall structure of capital changed. On the one hand, holding companies came to control a cascade of subsidiaries. On the other hand, the relationship between these corporations and other companies changed with the development of subcontracting, which is now widespread. Let us return to profits.
In the classical form of capitalism, superprofits were simply the profits of oligopolies. These very large corporations, drawing on the sheer scale of their “constant capital” (capital invested in the means of production) and their advanced technology, not only sold their products above their market value but also, thanks to their dominant position, possessed such “market power” that they were able to make extra profit.[4] An important point to clarify here is that oligopolies are not monopolies, they are in fierce competition with one another—which, as we shall see, already challenges one of the central premises of the theory of “technofeudalism”: that their power is fundamentally rentier in character. Today’s oligopolies can increase their profits by extracting as much as possible from their most profitable subsidiaries and, above all, from their subcontractors. But it is with digital capitalism that profits really take off.
Digital Capitalism
What is new about digital capitalism is that it is platform-based capitalism, operating through computer networks. Although virtually all companies now use these networks, the largest ones make broader and more intensive use of them, thereby saving on wage labor, and this technological superiority is, as in any market economy, a source of higher profits. The most significant case, however, is that of the tech firms themselves—including, of course, those that provide AI services, or specialize in AI—and this is where the issue becomes crucial. Where do their staggering profits—whether actual or expected—come from, when they seem to use even less labor? Doesn’t all this contradict the Marxian theory of the tendency of the rate of profit to fall, as its critics argue? Not at all, as we shall see. This is because, to counteract it, there are what Marx called “counter-tendencies,” some of which are spontaneous, while others are in fact the result of deliberate policies. Let us examine what these counter-tendencies are and how they operate in the case of digital capitalism.
1. “Constant” capital (which corresponds to the means of production) tends to decline in value due to technical progress in the capital goods sector, energy supply, and raw materials as a result of the globalization of trade, with raw materials increasingly sourced from low-wage countries. (Marx had already clearly emphasized the role of international trade in the value of these goods.) The share of “variable” capital (wages) consequently becomes proportionally larger, which allows for an increase in the rate of profit, since there is more labor to exploit. Digital capitalism, however, benefits fully from these technical advances because it makes extensive use of automation, electricity from a growing number of sources, and low-cost imported raw materials, including rare metals (albeit with a geopolitical risk).
2. Another general source of profitability is the extension and intensification of labor for the same wage, and this is where the social policies specific to neoliberalism have come into play. As is well known, the push to reduce working hours, driven by pressure from the labor movement, came to a halt in the 1980s, and today, employers’ organizations and the right in general continue to call for an extension of the legal workweek and a delay in the retirement age. As for the intensification of labor—which is less visible—it is also, in a new form, the goal of “new management,” and it affects engineers and managers as well, who are themselves under pressure. Digital capitalism is no exception to the rule: the “cool” side of management (a relaxed atmosphere and various perks) masks flexible hours and constant pressure. Remuneration for skilled workers has also declined, as there is now a plethora of IT professionals.
3. Another source of profitability is unemployment, which makes it possible to create what Marx called “the industrial reserve army” and to exert a downward pressure on wages, or simply the threat of unemployment. Automation, digitization, and, today, the spread of AI threaten armies of “blue-collar” and “white-collar” workers.
4. Large corporations pressure governments to boost their profitability both by demanding lower “labor costs” (i.e., socialized wages), lower taxes on capital, and various forms of aid (subsidies, infrastructure funding, etc.).[5] They are also experts (especially the internet giants) in tax evasion.
There would be no reason to pursue and reinforce all these corporate and government policies if profits sprang spontaneously from technological progress and the globalization of trade, as liberal economists claim.
5. Financialization, by distancing shareholders ever further from the workplace, has broken the resistance of workers and their unions. Let us recall that powerful “institutional investors” (investment banks—since the end of the separation between the banks’ lending and investment activities—insurance companies, pension funds, etc.) had already, more or less discreetly, seized power on corporate boards. Now it is the unregulated “shadow banking” sector (large funds such as BlackRock or Fidelity), beyond the reach of employees and even executives, that is so often calling the shots. It is therefore hardly surprising that the share of wages in value added has continued to decline.
The fact is, this financialization has reached its apex in digital capitalism. Let us add that the executives of the very largest tech companies are so pampered by these investors that they have themselves become major shareholders, resulting in colossal fortunes that even surpass those of the CEOs of other major multinationals.[6]
One puzzling question remains: the giant firms of digital capitalism employ relatively few people, compared to the profits they generate.[7] But appearances are deceptive.
Hidden Labor in Digital Capitalism
In digital capitalism, hidden labor is primarily the labor of “click workers”—both those who, as part of their jobs, provide data to their employers for free, such as Uber drivers or Amazon delivery workers, and those who work for tech companies, labeling or “moderating” images to feed and train their models. The latter, whether working from home or employed in “click farms” by subcontractors or by the companies’ platforms—which recruit them from poor countries—receive only starvation wages. One could say they are the “stokers” of the giant ships of the tech industry, at least until they are replaced by AI.[8] In total, there are an estimated 185 million click workers worldwide.
Granted, this alone cannot explain the extraordinary profitability of these firms. To understand this, we must turn our attention to the users of their products, who simply provide them with free labor whenever they go online, potentially supplying all kinds of data (on their personal information, location, behavior, health, ideas, feelings, social circles, moods, purchases, and so on) that serve as these firms’ raw material. [9] Here, some might challenge the notion that this constitutes “labor,” since it is not compulsory; labor is defined as an activity characterized by time constraints and results, as opposed to a leisure activity, which is optional. Yet it is as though the act of logging in is, more often than not, compulsory. How so?
Within companies of any kind, as well as within government agencies and public services, all kinds of data are collected in the course of day-to-day work activities. But this data is protected by laws, the strictest of which is the European Union’s General Data Protection Regulation (GDPR), which ensures that the data remains within these institutions and cannot be exploited by external entities, including major tech firms.[10] However, when it comes to cross-border relations—which are common in transnational corporations—this protection weakens because regulations are looser in other countries. And computer hacking is very difficult to combat.
It is primarily through consumption and leisure activities that online labor gives rise to the plundering of data. It is virtually impossible to avoid accepting “cookies” in the commercial sphere (at the very least, those deemed “essential”), which provide all sorts of data about the user.[11] This data is indispensable to commercial and content platforms for fishing for customers and encouraging purchases. For AI companies, this data is just as essential for training their models. It can also simply be sold to brokers, who in turn resell it to fuel the advertising industry—which has become the third-largest economic sector on the planet, after electricity and agriculture. Finally, social media platforms—even though they offer certain restrictions (such as limiting who can see your posts)—are an inexhaustible source of data, given the number of hours spent in front of screens on platforms like Facebook, Instagram, or TikTok. Only VPNs make it possible to escape data capture.[12]
One might argue that clicking on these platforms or networks does not constitute labor. Granted, it seems to be optional. But it is, in fact, compulsory—much like, to take a very simple example, a traveler who wants to buy a train ticket and can hardly go to a station ticket window to speak with an employee anymore (and who, if they want to find the best price in a so-called “dynamic” pricing system, must spend an enormous amount of time online). The more digitization advances, the more the user’s free labor becomes the norm.[13] Generally speaking, as with any activity in the realm of leisure and consumption, as soon as a product or service is not self-produced but is a commodity—and one must go through the process of buying and selling to obtain it—one must pay the price. This is evident when a subscription is required. Simply put, when a good appears to be free, the consumer pays for it by providing data—whether through placing an order, entering content, interacting on a social network, or conversing with a chatbot. And, as we know, digital capitalism has devised strategies to drive overconsumption, much like drug dealers do (through “recommendations” and incentives to “scroll”).[14]
Of course, online labor is not entirely free. Internet users obtain a good or a service by going online, but it is free to the extent that users provide—willingly or not—data, and not just their own, but often some of their contacts’ as well. In Marxist terms, only this labor goes unpaid. But it is substantial.
Measured in working hours, online labor continues to increase as we spend more and more of our time glued to our computers—or, worse still, our smartphones. Let us add that this labor requires a certain level of skill, ranging from the most basic (the rules for using a computer or smartphone) to the most advanced (learning how to use apps, from the simplest to the most sophisticated), and that it can be quite intense, given how much attention it demands. Last but not least, everything is designed to make internet use addictive. All of this is further exacerbated by the development of AI: its use is becoming increasingly mandatory, both in traditional work settings—where it indeed lightens the load for many tasks (when it does not eliminate them altogether) but also increases the workload for the remaining tasks—and in other activities.
Like all labor under capitalism, online labor is subject to surveillance. To be sure, this surveillance is no longer carried out by human supervisors, as is the case within companies (foremen, management controllers, and the like—now rebranded as “associates,” like all employees), but rather through the use of digital systems that issue “recommendations” and other persistent prompts. Online labor is also shaped by an ideology that has extended beyond the framework of the neoliberal enterprise to become embedded in digital platforms (the cult of the commodity and the best price, novelty as a euphemism for quality, and so on). [15]
Estimates have been made about how much money tech and AI giants make from providing data.[16] For every European internet user, this amount is said to be 1,381 euros per year and, over a full digital lifetime (estimated at sixty-nine years), at 224,223 euros—the equivalent of ten years of the median full-time salary in France. In the United States, the profit generated is even higher: $5,600 per person per year, and $716,000 over a lifetime. This revenue includes advertising, AI subscriptions, corporate licenses, API access,[17] algorithmic recommendations, data brokerage, and savings enabled by AI. Yet the vast majority of the data collected is provided mostly for free by internet users. It is therefore clear where the exorbitant profits of the major companies of digital capitalism come from. The exploitation of free labor gives tech firms such a significant advantage that it allows them to escape the tendency toward the equalization of profit rates across industries—a feature that had previously characterized capitalism—on a national and even international scale through globalization.
If we were to take the analysis of digital capitalism a step further, we would see that the degree of labor exploitation is even higher, because, according to Marx, only labor that produces “real” use-values creates value, as opposed to labor that is merely formal, such as the labor involved in buying and selling (unlike, for example, the labor of transporting or preserving goods). Part of this produced value naturally goes to unproductive workers who provide useful services—such as salespeople—but another part is used to pay for entirely unproductive labor—such as that of an advertising executive who is no longer an agent of information but an agent of propaganda or influence.[18] But that would take us too far off topic.
Hypercapitalism
Some authors seek to explain this phenomenon as “technofeudalism,” arguing that the profits generated by digital capitalism are in fact a kind of rent, comparable to that extracted by feudal lords—landowners—at the expense of their serfs or tenants. This is not convincing.[19] All rent is based on a monopoly—for example, ownership of land (“absolute” rent) or of more fertile land (“differential” rent), as Marx analyzed. However, the digital space is not closed off, and, as we noted earlier, the digital world remains a competitive—even hyper-competitive—environment. As for data, it is not a tangible good, but rather pure raw material for digital use. Do major tech companies turn other enterprises into “vassals”? This is, in reality, a general trend of capitalism, as seen in the widespread use of outsourcing. Tech companies have merely amplified it. Do they produce no tangible goods at all? This is already the case for all commercial enterprises, with the exception of a few of their activities (the storage and display of products). Ultimately, speaking of “rentier capitalism” is no more relevant. Capitalism is not based on the extraction of rent, but of profit, which comes from unpaid labor, and this labor is not comparable to that of serfs, which presupposes legal subjugation of one person to another. In reality, with digital capitalism, we are dealing with a hyperbolic form of capitalism, whose internal contradictions make its future even more problematic.
Conclusion
The purpose here was not to weigh the benefits and drawbacks of digital technologies, but to examine the origins of the high—and sometimes spectacular—profits generated by digital capitalism, particularly by the internet giants. However, for the reasons we have outlined, these profits are not surprising: while there is less labor in some areas, there is much more in hidden forms. We have not switched economic systems to a rentier economic system. Digital capitalism is merely a new form of exploitation that has become widespread by spilling out of the company itself to encompass the mass of its users or customers (when subscription fees are required). In another economic system, things would be very different. The workload within the company would be reduced, both in duration and intensity, because profits would be used only for investment, rather than going into the pockets of shareholders, who would receive only limited compensation—or none at all, in a perfectly conceivable system, where there would no longer be any shareholders. As for customers and users, they would need to provide only the data needed to produce the commercial service (which would allow the collection of data for informational purposes but would rule out any advertising for propaganda or influence). When it came to a public good—rather than a private good—whether, depending on its nature, it was provided by a government agency, a public institution, or a public enterprise, and whether it was free of charge (and thus tax-funded) or fee-based, the same principle would apply: only the data needed to provide the service would be collected. This would rule out any form of mass surveillance.[20] Given their indispensable role in the economy, the major tech companies would become public industrial and commercial enterprises, with a system of governance different from that of other public enterprises. To train their models, they would use this data, rather than the indiscriminate mass of data laden with all sorts of “cognitive biases,” as is the case today. “Click labor” would no longer be outsourced but would be paid at its fair market value. Online labor would serve the common good and would not give rise to any exploitation, since public digital enterprises would no longer make profits, except those used to fund investment.[21]
Bibliography
Andréani, Tony. Historical Materialism: Revisiting The Fundamental Concepts. Translated by Sarah Erden. London: Routledge, 2026, Chapter 13 on the capitalist mode of production.
Andreani, Tony. “Un féodalisme numérique?” Blog post. March 1, 2025. https://tonyandreani.canalblog.com/2025/03/un-feodalisme-numerique.html
Andréani, Tony. “Technoféodalisme ou hypercapitalisme?” Blog post. Aug. 18, 2025. https://tonyandreani.canalblog.com/archive/2025-08/
Cohen, Daniel, Homo Numericus: The “Coming” Civilization. Translated by Steven Rendall. Cambridge, UK: Polity Press, 2024.
Durand, Cédric. Techno-féodalisme. Critique de l’économie numérique, Paris: Éditions Zones, 2020.
Mazzucato, Mariana. “Preventing Digital Feudalism,” Project Syndicate, 2019.
Varoufakis, Yanis. Technofeudalism: What Killed Capitalism? London: Bodley Head, 2023.
Zuboff, Shoshana. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. New York: PublicAffairs, 2019.
Smyrnaios, Nikos. Beyond Techno-feudalism: Toward a Critique of Digital Imperialism, HAL Open Science, 2026.
Smyrnaios, Nikos. “Au-delà du techno-féodalisme: pour une critique de l’impérialisme numérique,” Contretemps, 17 June 2026.
Zukerfeld, Mariano. Knowledge in the Age of Digital Capitalism, London: University of Westminster Press, 2017.
Notes
[1] To give you an idea, in just one quarter (the first quarter of 2026), Amazon posted a profit of $30 billion, Microsoft $31 billion, and Meta $26 billion. While OpenAI is not yet profitable, it expects to be so by 2030, with $56 billion. Anthropic is not profitable either but has announced that it will soon turn a profit. That has not stopped these two AI giants from commanding enormous valuations today ($852 billion and $900 billion, respectively), as investors expect extraordinary returns from them.
[2] This has led CEOs such as Mark Zuckerberg (Meta), Jeff Bezos (Amazon), and Sam Altman (OpenAI) to support the idea of a universal basic income, without any requirement to work in return.
[3] This is why a scholar such as Mariano Zuckerfeld argues that the digital economy renders Marxist analysis obsolete, contending that it needs to be conceived not in terms of labor, but in terms of knowledge.
[4] Their dominant position is underpinned in particular by their advertising spending, patent holdings, and agreements among themselves.
[5] Government support is particularly evident in the case of digital capitalism. As has often been pointed out, the US government—and particularly its military-industrial complex—has played, and continues to play, a major role in the rise of Big Tech in the United States.
[6] The case of Elon Musk, whose fortune was estimated at $1 trillion before declining somewhat, is emblematic.
[7] This is even more evident with companies specializing in AI: OpenAI has only 4,500 employees (mostly engineers), and Anthropic has 2,500.
[8] See Casilli, Antonio A. Waiting for Robots: The Hired Hands of Automation. Translated by Saskia Brown. Foreword by Sarah Roberts. Chicago: The University of Chicago Press, 2025. Casilli identifies a third category among “click” workers: social media users. This is also our view.
[9] Shoshana Zuboff has brilliantly demonstrated, in The Age of Surveillance Capitalism the scale and variety of the data collected, which has grown even further with the widespread use of “connected devices”.
[10] Personal data protection comes with numerous rights for individuals (including the right to access, rectify, erase, and transfer data), but these rights are difficult to exercise.
[11] The digitization of citizens’ interactions with government agencies and public services should, in principle, be exempt from data collection by the private sector, but this is far from always being the case.
[12] An application such as WhatsApp only provides partial anonymity. Similarly, the European Regulation (GDPR) merely recommends pseudonymisation, which offers limited protection.
[13] In fact, this is part of a long-standing evolution of capitalism. To take a very simple example: in the past, when a customer went to a gas station, an employee would fill the tank and even wipe down the windshield—these services were included in the price of the fuel. Then customers began filling up themselves, and today they make the payment without going through a cashier.
[14] See Yves Marry and Florent Souillot, La guerre de l’attention. Paris: Les Éditions L’échappée, 2022.
[15] See, among others, the analyses by Luc Boltanski and Eve Chiapello in The New Spirit of Capitalism. Translated by Gregory Elliott. London: Verso, 2018.
[16] See MINTED media ads tech, mnted.fr
[17] APIs are, in essence, interfaces that enable the exchange of data or functionality.
[18] On the question of productive and unproductive labor, see Chapter 5 of my book Historical Materialism: Revisiting the Fundamental Concepts.
[19] There is a considerable body of critical literature on the subject. An overview can be found in Usbek§Rica (https://usbeketrica.com: Technoféodalisme: avons-nous vraiment régressé vers Le Moyen Age ? Proponents of this theory hold various positions on its relationship to capitalism, ranging from Varoufakis, who sees it as the end of capitalism, to Durand, who views it as more of a shift in direction.
[20]. On the distinction—which I believe to be fundamental—between social goods and private goods, and on these institutions, I would like to refer the reader to my book Ten Essays for a 21st-Century Socialism. London: Routledge, 2024. Essays 4 and 5.
[21] It is in this same vein that Evgeny Morozov proposes that digital data be considered a common good.
Image via Wikimedia.
No comments:
Post a Comment