Our Technological Wager
The AI investment boom is a gamble of epic proportions. Yet on both sides of the wager, it seems, the house wins.
Everything is a wager these days. Gambling odds are now a ubiquitous feature of sports broadcasts, and prediction markets welcome bets on the timing of U.S. missile strikes. One popular self-help book distilled from conversations with high-flying tech CEOs urges its readers to simply “Bet on Yourself.” It’s a mantra worthy of Elon Musk, briefly the world’s first trillionaire, but surely no one follows this philosophy more than our president, who revels in trusting his own instincts against all naysayers and has defined himself against his predecessors by his voracious appetite for political risk-taking.
Before our eyes, however, a collective gamble of more epic proportions has begun to take shape. Amazon, Microsoft, Alphabet, Meta, and Oracle—five of the biggest “hyperscalers” operating massive computational and storage infrastructure—are on track to reach over $1 trillion in capital expenditure for computing power in 2025 and 2026. By one estimate, investments in computing and software accounted for 90 percent of GDP growth in the first half of 2025. Data centers are springing from the earth across the rural United States, with more than 1,700 planned projects, requiring 1 million metric tons of cement by 2028, set to increase the country’s total computing capacity nearly sevenfold. The architecture of these projects, whose hulking forms break the horizon like the scattered monoliths of an alien civilization, communicates a stark vision of a world practically empty of human beings altogether. As Jeremy Wallace argues in this issue, the Chinese state has invested in a very different kind of future defined by green technological power. The United States, however, is doubling down on fossil-fueled computer power. These competing visions of our technological future are the subject of this fall’s special issue.
The wager of this staggering investment boom is simple: Artificial intelligence will completely and rapidly reshape American society. Agentic AI will think for us, code for us, write for us, shop for us, and be our friend, our therapist, our teacher, and our doctor. Entire workforces will be automated, and new industries will spring up overnight. Half of entry-level white-collar jobs could be “displaced” within five years, Anthropic CEO Dario Amodei has claimed. Intelligence will become a fundamental utility as basic as electricity, according to OpenAI CEO Sam Altman, and customers will simply “buy it from us on a meter.” The winners of this world-historical transformation stand to make profits beyond all imagination, but there will only be a select few. With breathless predictions of the imminent arrival of “artificial general intelligence”—a hypothetical horizon in which models transcend any task-specific training and exceed human capabilities across a full range of domains—investors expect that only a handful of firms (or even just one) with the most advanced models will ultimately dominate the market. First-mover advantage is crucial, and the industry is therefore in a breakneck race to spend as much as possible, as fast as possible. They, too, are betting on themselves.
Betting markets do not stand outside the reality they purport to predict: The market odds for the removal of Venezuelan president Nicolás Maduro reshaped expectations about the geopolitical future as much as the “law enforcement operation” to abduct him did. Similarly, the wager on AI is actively seeking to reshape the conditions of its own success. The eye-popping equity valuations of firms like OpenAI, Anthropic, and SpaceX are driven by a speculative exuberance premised on a particular vision of the future. The predictions these firms make are in effect sales pitches for that speculative reality. Hence the full-court press of science-fiction scenarios of the singularity, the dead-eyed celebrity endorsements in Super Bowl commercials, the shoehorning of LLM interfaces into every conceivable consumer platform, the inscrutable ads for software startups confronting commuters in the New York subway, and the dystopian billboards in the Bay Area urging companies to “Stop Hiring Humans.” The more we collectively believe, the more the numbers go up.
Such moments of intense speculation, John Maynard Keynes once observed, are based not on “an exact calculation of benefits to come,” but something more ineffable and more psychological, rooted in “the nerves and hysteria and even the digestions” of potential investors. Undoubtedly, generative AI is a hugely powerful technology that is already reshaping major industries, software engineering foremost among them. Yet there is indeed something stomach-turning about this moment, because this wager implicates us all, and very few will escape the fallout if it all goes wrong.
If the most wide-eyed estimates of labor market disruption made by the likes of Amodei turn out to be correct—and it is precisely these sorts of predictions that undergird the forecasted demand for computational power that the hyperscalers are racing to meet—then, absent any compensatory job growth, the scale and speed of social dislocation will make the slow-motion disaster of Rust Belt deindustrialization look quaint. Conversely, in the likely event the demand turns out to be lower than predicted, it will lead to a sharp pullback in investment and financing. Given the opaque cat’s cradle of financial ties between hyperscalers, AI labs, chipmakers, and financial institutions, a recent report by the Bank for International Settlements has predicted that such a bust could cascade down the AI supply chain and strike at the foundations of the U.S. economy, inducing major recession. As if that wasn’t enough, given recent changes in Nasdaq rules, should a series of planned IPOs of AI firms go ahead, index funds in the retirement portfolios of tens of millions of U.S. households will be directly exposed to a potential crash in a historically unprecedented fashion. Either way, the shockwaves will be extreme. Yet on both sides of the wager, it seems, the house wins.
On the left, most writing about AI has fallen into three central tracks. The first engages in a kind of wishful denial about the abilities of AI to do anything useful at all. The second, its mirror image, fully accepts the worst-case scenarios of AI doomers and fixates on the possibility of a “permanent underclass,” or even of total human extinction. The third focuses on more philosophical questions of humanity, knowledge, and creativity in a world in which what Marx once called the “general intellect” of collective human knowledge has been gobbled up by a handful of corporations and regurgitated back to us as slop. This last discourse is, understandably enough, particularly important to writers, teachers, and intellectuals.
The problem is not that the potential of AI isn’t inflated by dizzying hype (it is), that extreme but highly unlikely scenarios don’t matter (they do), or that the meaning of education, the arts, and even the human person aren’t under pressure (they are), but rather that the gravitational pull of these conversations has shrunk the space available to assess our current conjuncture and the possibilities of intervening within it. By refusing to get under the proverbial hood of what will surely be one of the most significant technological breakthroughs of the last decade, we foreclose the possibility of alternative designs and uses for it, from the immediate to the utopian. Technological development never proceeds along a foreordained path but is shaped according to social values and priorities. AI is no exception; much like the bicycle or the internet, its affordances are not limited to those imagined by its designers.
By contrast, the OG Luddites of the 1810s, even as they took sledgehammers to power looms in the night, were skilled artisans who expressed a sophisticated understanding of textile machinery in their proposals to protect their livelihoods alongside it. Their revolt was not against the machines as such, but, as Kirkpatrick Sale once put it, against “what that machinery stood for: the palpable, daily evidence of their having to succumb to forces beyond their control.” A fealty to that spirit means taking the increasingly fierce local opposition to data centers seriously as a genuine social movement, while also looking beyond its short-term goals to imagine an alternative political economy, as Brian J. Chen does in this issue. It also means considering how the hard-won architecture of the U.S. welfare state—much of it constructed to withstand earlier waves of technological dislocation—can be scaled up to meet the contemporary challenge, as Marc Aidinoff proposes.
Yet the rightward turn of Silicon Valley leaders, canvassed by Paige Oamek in this issue, and their growing alliance with the U.S. security state—even in seemingly innocent domains like video games, as Jarod Facundo explores—suggests that winning any victories in this domain will require a more hardnosed strategic orientation. Even the rather mild proposals for AI regulation by Democratic primary candidate Alex Bores in New York’s Twelfth Congressional District unleashed a torrent of $24 million in outside spending by tech-backed financial groups.
In the last decade, observers on the left hailed the incipient emergence of organizing among tech workers as a sign of a potential watershed in the industry. It has not come to pass, for reasons JS Tan and Clarissa Redwine incisively analyze in this issue. As the tech elite tightens the reins over its own workforce, moreover, Veena Dubal and Katie Wells argue that the rise of algorithmic pricing and surveillance threatens what little autonomy many workers have successfully carved out. For a magazine devoted to the ideals of democratic socialism, this moment feels particularly sobering. New thinking and new strategies will be required. As the right well knows, one should never let a crisis go to waste. If the economy-wide gamble brings the whole house of cards crashing down, we must be ready for the reconstruction to follow.
Simon Torracinta is a senior editor at Dissent.