Artificial intelligence is already reshaping the US economy at a scale that dwarfs most other sectors. In the final quarter of 2025, the technology accounted for nearly 60% of the country’s economic growth, according to official data. That figure builds on a broader trend: in the first half of the year, AI-related capital expenditure added 1.1 percentage points to GDP, rising to 1.3 percentage points by the spring. Business investment in software, IT equipment, and structures tied to AI and data centres alone generated 30% of all GDP growth in the second quarter and 20% in the first.
The numbers are staggering, and they are underpinned by an extraordinary flood of money. Hyperscalers — the giant cloud computing firms — were on track to spend $342bn on capital projects in 2025, a 62% jump on the previous year. The entire AI ecosystem now draws roughly $1 trillion annually, with enterprise applications attracting $100bn in venture capital in a single year. Over the longer term, projections suggest AI could lift productivity and GDP by 1.5% by 2035, nearly 3% by 2055, and 3.7% by 2075. The buildout has added a layer of resilience to the economy, moving with some independence from interest rates and labour market swings.
Yet for all this investment, a deep unease runs alongside the boom. Since the release of ChatGPT in late 2022, more than half a million workers in the tech industry alone have lost their jobs, and warnings of far broader disruption have become routine. Jensen Huang, chief executive of chip giant Nvidia, said in 2025: “Every job will be affected, and immediately. It is unquestionable. You’re not going to lose your job to an AI, but you’re going to lose your job to someone who uses AI.” Dario Amodei, the CEO of Anthropic, went further, predicting in January that “AI isn’t a substitute for specific human jobs but rather a general labor substitute for humans.” He has previously warned that up to 50% of entry-level white-collar jobs could vanish within five years.
Those dire predictions have, however, begun to soften. In May, OpenAI chief Sam Altman retreated on his earlier claims of widespread job replacement, saying: “I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened.” He later told associates he was “delighted to be wrong” about a rapid AI-driven jobs apocalypse. Even within the tech industry, where the deepest cuts have occurred, doubts are surfacing. Reports from employees at Amazon, Meta and Block suggest that the AI productivity gains their bosses trumpet are overblown. Venture capitalist Marc Andreessen acknowledged in March that overstaffed companies are using AI as a “silver-bullet excuse” to shed workers.
Academics who study technological change urge caution about reading too much into the layoff numbers. Martin Beraja, a professor at UC Berkeley’s Haas School of Business, argues that studies linking the release of ChatGPT to a decline in entry-level software jobs are “problematic”. He points out that the tech industry built up a huge surplus of workers coming out of the pandemic, and once consumption patterns moved from online back to the real world, many of those roles were simply no longer needed. Suresh Naidu, an economics professor at Columbia University, notes that even if worst-case scenarios for tech play out, software represents only 4 to 6% of US GDP. “Is it, in fact, going to destroy all of the jobs? I’m not convinced,” he says. “Even take software. It’s a lot, but it’s not like the whole economy can be replaced by Claude Code.”
A November study from MIT using the Iceberg Index estimated that AI can technically replace 11.7% of the US labour market, with a significant portion of that disruption hidden beneath the surface — in roles such as HR, logistics and office administration. Yet overall, AI’s impact on job growth has been weak, with white-collar sectors showing stronger productivity gains rather than outright job losses. Through March 2026, there has been no meaningful change in unemployment for workers in high-AI-exposure jobs.
The hype, the surveillance, and the real threat
The conflicting signals have generated what some experts call “AI absolutism” — a way of viewing the technology as a godlike force that will either usher in a golden age of productivity or doom humanity. Naidu argues that this framing is by design. “If you want to justify this enormous valuation in your IPO, you need to point to the revenue stream that you’re going to generate in the future,” he says. “You just need to make it look like you have something that can eat all the work on the planet, so that an investor will think: ‘Oh wow, I don’t want to miss out on this thing.'” The robber barons of the age, he suggests, profit not just from enthusiasm but from terror.
Anil Dash, the former CEO of the startup Glitch who has written about technology for decades, is equally unconvinced that the AI being sold will deliver on all its promises. “Any technology that you invest like a trillion dollars into is going to be able to do a lot of things, good or bad. It’s a leap forward. I don’t think we’ve ever had a machine learning system that can do as many things as this one does,” he says. But “there’s so much noise that it’s hard to tell what the domains of applicability are.” Coding stands out as an exception, he explains, because its output can be tested objectively — it either works or it doesn’t. For most other applications, the results are far more subjective and therefore less likely to lead to immediate job replacement.
Naidu frames the current moment as a vast, uncontrolled experiment. “An experiment implies a control group of something that’s not affected. There’s no control group here,” he says. The idea that AI will replace human workers in droves, he argues, is not just a prediction but a marketing tactic. It distracts from a far more realistic — and arguably more insidious — application: using AI to surveil and micromanage employees, squeezing out ever more productivity while making workers feel grateful to have any job at all.
Gig workers have already become the guinea pigs for this kind of algorithmic management. On platforms such as Uber and DoorDash, algorithms dictate tasks, payment, performance evaluations, and future opportunities — often without any human manager involved. Research shows that algorithmic management can cut both ways: it can be associated with a greater sense of procedural justice, but it also typically reduces job autonomy. For many workers, this leads to deeper engagement even as it worsens working conditions. Now, labour experts predict the model will spread well beyond the gig economy, into offices and warehouses across the country. AI systems are increasingly being deployed for employee surveillance and micromanagement, raising concerns about invasion of privacy, misinterpretation of behaviour, and the uneven application of monitoring. Done poorly, such systems can create distrust and lower morale. Advocates for ethical AI argue that the focus should be on insights and transparency rather than direct surveillance.
Alternative futures
The version of AI being marketed does not have to be the version that society adopts, nor the story people believe. “Where I think we have to get to is, there can be alternatives,” says Dash. “What we can imagine is, rather than the ChatGPT killer, a lot of different little AIs from little responsible players.” A few such initiatives are already quietly emerging, he notes, harkening back to the more optimistic and decentralised early days of the internet.
Beraja argues that the conversation has been too narrowly focused on job replacement. Outside a few industries like tech, studies show that the most effective ways for companies and individuals to use AI are not to replace workers but to learn more — and learn faster. “Where I think we have to get to is there can be alternatives,” echoes Dash, underscoring that a diverse ecosystem of smaller, accountable developers could offer a very different future than the one currently being built by a handful of dominant players.
Meanwhile, the upheaval AI has already caused may open the door to a resurgence in worker power. White-collar employees, witnessing the instability of roles once considered safe, are beginning to see the appeal of solidarity — with colleagues in their own offices and with workers in the blue-collar world. The Industrial Revolution, another era of rapid technological transformation, served as a key catalyst for the labour movement. Its victories took time, but they reshaped the economy for generations.
The choice, in other words, is not simply between embracing AI or rejecting it outright. As journalist Shira Ovide has noted, AI is “just another technology Americans don’t like but can’t stop using” — a contradiction that mirrors the polarised debate. The more productive path, academics and technologists argue, lies in moderation: using AI to accelerate learning, supporting the growth of responsible smaller players, and ensuring that the benefits of the technology do not concentrate solely among the wealthy. “There can be alternatives,” Dash insists. And some of them are already being built.
