This semester, I’m teaching a seminar on Max Weber at Cornell. Max Weber (1864-1920) is a thinker who has fascinated me for a long time: a polymath who worked on law, economics, history, sociology, of course, political theory and religion, but also a complex and elusive figure who never wrote a grand work presenting his thought in a coherent system. To call oneself a “Weberian” does not refer to a precise worldview, whereas one would say that of being a “Marxist” or a “Rawlsian.” Yet Weber was an incredibly prolific researcher (his complete works number 22 volumes) and one of the sharpest analysts of modernity that ever existed.
The starting point, which is also the turning point of Weber’s career, remains his most famous work, The Protestant Ethic and the Spirit of Capitalism, a long essay published in two parts in the Archiv für Sozialwissenschaft und Sozialpolitik in 1904 and 1905. It has sparked numerous debates and criticisms from economic historians. I will set aside here some of the historical refutations and objections, legitimate as they are. They are important, but I am less interested in the historical arguments than in the theoretical implications of his argument. For The Protestant Ethic remains a text that deserves to be read for what it teaches us about the role of ideas and of action in economic life in general. A recent reading of this text has inspired me to reflect on how Weber’s argument illuminates several aspects of the current rise in investment in AI. These concerns primarily relate to the relationship between reason and capitalism.
First, Weber’s analysis suggests that, if resources devoted to the development of artificial intelligence can be justified, the scale of this commitment is irrational at the level of society as a whole, just as the Calvinist and Puritan ethic of ascetic dedication to a vocation once was.
Second, the AI sector offers neither clear profitability nor an obvious future scenario, two conditions Weber regarded as essential commercial requirements of Western capitalism. Deficit-prone and operating in an “all or nothing” mode, this sector resembles far more what Weber called the “adventure capitalism” of the early modern era: a digital conquistadors’ venture, in some sense.
Finally, according to Weber’s interpretation, the Reformation triggered a process of emancipation based on books, vectors of ideas and norms that foster human understanding, broaden literacy and stimulate rational action. This process is now being short-circuited: the diffusion of large language models (LLMs) seems to undermine literacy across the population and spread a sense of impotence across the entire Western world.
The Rational Quest for the Irrational
The central problem for Weber lay in a fascinating gap between ideas and outcomes. On one hand, Protestant doctrines were austere and not particularly joyful (I’m well positioned to know this, since my paternal family has been part of the Dutch Reformed Church for four and a half centuries). Luther and Calvin’s teachings clearly did not advocate the pursuit of happiness or any form of utilitarianism, and offered no license to enrich oneself. On the contrary, the Reformation as a whole was a vehement rejection of precisely these vices of the Catholic institution: corruption, vanity and greed. Yet, despite this explicit anti-materialism, the countries where Protestantism took root adopted capitalist behaviors very early and became deeply materialistic societies. How could this paradoxical result have been possible?
If Weber posed the problem, it was because he sought to analyze the unforeseen consequences of historical development: how a set of stern and rigorous beliefs, hostile to the maximization of utility, nonetheless gave rise to material transformations highly favorable to well-being. Why did Calvinists, in particular, attach themselves to an “intense activity in the world” as an end in itself, and not merely as a means to an end, as any sensible person would?
From a pragmatic and instrumental point of view, their fierce attachment to renunciation loaded with guilt and to strict austerity hardly makes sense. Utilitarian ethics fits poorly with ascetic Protestantism: what happiness or well-being can an ethic serve which keeps its adherents in a perpetual anxiety (the fear of having been damned by God) and drives them to seek desperately activities capable of warding off this “religious angst”? The core strength of Weber’s argument lies in this: it was the Calvinists’ adherence to a rationally extreme doctrine, the absolute creed of predestination, that pushed them to reorganize their lives in a decisively rational manner. In other words, they sought the irrational certainty of salvation (certitudo salutis) through the rational means of good works and hard work. Repeatedly in the first half of The Protestant Ethic, Weber emphasizes how the act of dedicating oneself to work as a vocation, regardless of its usefulness, is “irrational from the standpoint of a purely eudaimonist interest,” meaning from the perspective of the pursuit of happiness. That is why the theological content of Protestantism itself mattered so much.
How can we explain, today, a boom in investments based on technological hopes that have not yet translated into durable profits? Billions in operating losses accumulate, in defiance of traditional, rational, prudent corporate accounting, one of the practices Weber regarded as essential to the rise of Western capitalism. Even the most optimistic scenarios do not foresee the AI sector becoming profitable before 2030.
To explain such behavior, the only solution, it seems to me, is to apply to the current frenzy the analysis Weber used for Calvinist communities. Like these ascetic communities, hyperscalers—the tech giants building large-scale computing infrastructures—are rational devotees of the irrational. Their conviction that their cause is of paramount long-term importance overrides any short-term calculation, whether it concerns their own profit, human happiness, or the utility for society as a whole. They are also remarkably frank about the major drawbacks of what they do. The leaders of AI companies openly, for a long time now, acknowledge the possibility that what they develop could cost a significant percentage of workers their jobs, trigger a collapse of civilization, or lead to the death of humanity itself.
Yet they manage to mobilize enormous capital and capture public attention. Weber described the “unyielding Puritan merchants of the heroic age of capitalism” as “sure of themselves saints.” We find in today’s tech elite the same mission-driven mindset, humorless, and the same unwavering refusal to admit that other values deserve to be pursued. In the face of growing protests against data centers that push up prices and consume resources, and against the legitimate fear that AI will make humans less skilled and competent, they insist that the potential benefits of artificial general intelligence must take precedence over all possible objections. Do those who oppose deploying this technology fail to grasp the magnitude of the task? They speak of a God-machine to create; space to be colonized to install a superintelligence off Earth; AI will trigger a “Cambrian explosion” of progress and prosperity that will sustain itself.
Such worldviews, in many ways, stand in stark opposition to the religious attitudes of devout Protestants. There is nothing modest about the aims of hyperscalers, the hopes of their funders, and the convictions of their evangelists. They believe they can radically change the world by overturning the current order. In the eyes of early Protestants, such an attitude would have been a matter of pride and sin. For Calvinists, there was no God to create, since it is He who created us; humans existed for Him, and not the other way around. The religious vision of the 16th and 17th centuries likewise ignored the feverish obsession with rapid progress here below: the essential thing was to belong to those who would attain the better world of the beyond. Calvinists and AI accelerationists (advocates of as rapid a development as possible of this technology) indeed share one point: the intensity of their conviction that all of social life can and must be reorganized. Among AI true believers there is a distinctly eschatological vein, oriented toward the end of times. According to them, a major event will change the world; this new world will be morally good, and they seek to hasten its advent by every possible means. Anything that opposes this goal becomes, from then on, a detrimental, regressive and welfare-destroying obstacle to overcome.
Let us take a step back and ask: what do we really know about AI that would justify making its development a higher priority than all other collective priorities of humanity, and especially ahead of major civilizational challenges that require immediate solutions? Surely, it is capable of accomplishing impressive feats in computation, synthesis and analysis. Its advantages in biomedical research appear substantial. We can entrust it with a large portion of the tedious analysis and transcription tasks that modern society generates. There are good reasons to think that it can, in certain applications, contribute to improving well-being. This is precisely the more measured, purely instrumental approach that seems to dominate in China.
But in the West, notably the United States, AI acceleration has become an end in itself. It is striking, because the unchecked spread of this technology obviously imposes many drawbacks on our collective and individual well-being, on cohesion and the quality of our living environments, and on our concrete ability to act. We should especially question what the real chances are that it will solve all the planet’s problems that we already know, for which proven solutions exist but are hindered by a lack of resources. Combating catastrophic climate change is the most obvious priority for humanity; but fighting world hunger, investing massively in disease prevention and intensifying the fight against poverty in developing countries are other priorities, entirely realizable. All of these commitments are fundamentally rational, whether one reasons in utilitarian terms or according to a universal moral standard; and decades of solid scientific research show that they can be achieved if we allocate sufficient resources.
Yet, these rational and down-to-earth investment choices have been largely sidelined in public debate. Goldman Sachs projects that more than $1 trillion will be invested in AI worldwide in 2026, nearly half of the $2.4 trillion that rich countries are expected, according to the OECD, to spend each year through 2030 to maintain a chance of meeting climate stabilization goals, even the least ambitious. It is an extremely risky and irrational bet: in health, energy or the biosphere, many technologies and routine maintenance needs more tangible would deserve more of this money, for the simple reason that we know with certainty that, all else equal, their impact is positive.
What to make of the deep, broad rationality of our economic system when it ignores such evident and immediate dangers to our world and devotes such wealth to a transcendent cult of computing capable of improving itself? Weber helps us understand that with regard to a universal goal, the rise of AI is, in reality, profoundly irrational. Whatever the specific rational uses of this technology, a powerful current of general irrationality now dominates the discourse surrounding it and the enormous economic hopes placed in it. The irony is all the greater since all this occurs at the very heart of “Western rationalism.”
The Varieties of Capitalism According to Weber
The Protestant Ethic also contains another explanation of the irrational aspects of the AI boom, more focused on power. Weber’s essay does not aim to provide an exhaustive explanation of the rise of capitalism. It even asserts categorically that capitalism existed long before the Reformation, since ancient Babylon, and that Europeans never had a monopoly on it: capitalist activities appeared throughout history in various civilizations, from China to India, through ancient the Mediterranean and the Middle East. (That is why the objections most often raised against his thesis, “but the Medici, Catholics, were capitalists! Islam had merchants! Chinese civilization has had pawnbrokers for millennia!”, fall by the wayside: they challenge general statements Weber never made.)
Weber was seeking, on the contrary, above all to explain the particular form, and in his view the unique form, that capitalism has taken in the West: “bourgeois capitalism, austere with its rational organization of free labor.” But by delimiting thus the object of his inquiry, he actually left out several variants of capitalism that did not fit this narrow model. And it is among these other types that one finds better clues to understand the behaviors that today characterize AI investment.
Beyond this form, Weber’s essay explicitly mentions at least five other types of capitalism, distinct from the “bourgeois, austere” and rational capitalism founded on free labor:
1 — Political capitalism (profit conferred by political power or the proximity to it).
2 — Pirate capitalism with an “irrational and speculative” character (flibusterism, conquest, monopolization of resources, various schemes based on chance).
3 — War capitalism (mercenary activity as well as contracts and military provisioning).
4 — Financial capitalism (the ancient practice of lending money and investing capital).
5 — Commercial capitalism (mercantile and exchange-based, also very old).
Dylan Riley has recently revived the category of political capitalism to describe the political economy of the United States. This form of capitalism is characterized by exploiting profit opportunities offered by access to state power. The classic example is tax farming, which Weber knew well from studying the Roman Empire and Roman law. But political capitalism also encompasses all sorts of venal offices (as those that characterized feudal Europe, South Asia and China) and, more generally, the exploitation of monopoly rents conferred by political power. The French lords who invested their money in rights of justice, the ijaredar, tax farmers of Mughal India, and the shen-shi, Chinese literati-officials collecting fees: all counted for Weber among entrepreneurs of “politically determined capitalism.” It is evident that the commercial interests surrounding Trump display strong rent-seeking characteristics. More broadly, the valuation of many technology platforms seems to rely, wholly or in part, on various forms of monopoly rents. Political capitalism thus constitutes a viable alternative to the “rational and sober bourgeois capitalism” at the heart of The Protestant Ethic, a notion that, evidently, does not apply to today’s AI.
Nevertheless, to explain the extraordinary wave of investments triggered by AI (which now weighs more heavily, as a proportion, than those in railways in 19th-century United States), perhaps it is Weber’s second category, speculative adventure capitalism (Abenteuerkapitalismus), that is most relevant. This capitalism was the result of “speculators on the chances (Chancen) of monetary gain,” among whom were “promoters, concession hunters, large-scale speculators,” and so on.
Spanish conquistadors who ventured to the New World in search of immense treasures and enrichment through conquest and force are a good example.
Today’s AI companies are algorithmic Pizarros in pursuit of their own New World: they appropriate all usable data they can find and inject it into their large language models, in the hope of extracting a unique profitable opportunity that will eventually sustain itself. This data is their equivalent of the precious metal that, for centuries, gushed from the mines of Potosí.
Bibliocracy and Liberation
One of the intriguing sub-arguments of The Protestant Ethic is that the Reformation broadened the place of religion in individual life. For most Catholic peasants, who confessed their sins from time to time and received absolution, it was merely a balm for conscience; it became a comprehensive model of conduct in all domains, what Weber called a guide for the “conduct of life” (Lebensführung). Protestantism, and Calvinism in particular, aimed to establish a “methodical control of man as a whole.”
But Weber emphasizes that the transformation of life conduct associated with the Reformation did not stop at obscure theological debates about predestination, free will and good works. It also rested on a notable strengthening of the power of writing, which was simultaneously its consequence. Protestants, from Luther and Calvin to Baptist sects, sought to make Scripture accessible to laypeople to increase their number of followers. This concern to have texts in the vernacular led to translating the Bible from Greek and Latin into the language of the people, and to developing religious education from childhood. All of this contributed to a notable rise in popular literacy in Protestant countries, a well-established fact. It is telling that Weber described the importance of the written word for Calvinists as a “bibliocracy” or a “reign of books.”
These Protestant attempts to reshape community life on the model of the early Christians rested on a key element: they made everyone obliged to submit to the power of the written word and to organize life around it. This fostered literacy, self-control and inwardness. In the end, as individuals explored their inner selves and what drove them, this introspection became a pillar of bourgeois civilization and the foundation of modern subjectivity. It sparked advances in scientific inquiry and philosophy, artistic experimentation, the exploration of psychology, and many other areas.
For Weber, however, bibliocracy gave particular strength to the Calvinists: an absolute point of reference for ascetic conduct enabling the “liberation of individual powers.” By adopting toward the world a wholly moral orientation, humans began to regulate and discipline their actions in a new way, increasing their predictability, their calculability, their ability to make promises and to keep them, as well as their mastery of the world. All of this was profoundly liberating. (Nietzsche offered a very similar argument about the development of consciousness and self-mastery in his 1888 work, On the Genealogy of Morality.)
Here is where I want to get to: today, AI does not merely undermine the mastery of writing (the ability of individuals to read a text and to reflect on it); it also undermines the broader sense of self that is tied to it. In an era of ubiquitous AI, can we imagine that a liberating self-mastery and an emancipatory capacity for action can still develop within society? By losing mastery over the written word, do we also lose an entire set of positive behaviors that it made possible? This is one of the crucial questions raised today by The Protestant Ethic.
There are many other aspects of Weber’s thought I would like to delve into. One concerns how we can think of AI accelerationism as a theory of history. A large portion of today’s AI discourse, among both its supporters and its detractors, carries a remarkably millenarian character. The very idea of an apocalyptic event that would divide the world into winners and a permanently relegated underclass closely echoes Calvinist eschatology, especially the distinction between the elect and the damned. It remains to be seen whether this prophecy will be realized here on earth or beyond.