A potential crash in the artificial intelligence industry could put workers at risk twice over: an economic downturn could force businesses to cut costs, while falling AI prices could make replacing employees with automated systems more attractive, according to an analysis done by the National Public Radio (NPR).
The scenario, explored by NPR Planet Money writer Greg Rosalsky in a report published on 20 October, is not a prediction that the AI market will collapse. Instead, it examines how a potential market crash could trigger a recession and accelerate the replacement of workers, even as investors suffer losses.
How a Recession Could Accelerate AI Job Losses
The possibility draws on economic research suggesting that recessions can coincide with sharp declines in employment in routine occupations vulnerable to automation.
A study by economists Nir Jaimovich and Henry Siu found that 88% of the decline in the share of Americans employed in routine occupations between 1990 and 2017 occurred during recessions or in the months immediately before or after them.
The finding concerns earlier patterns of automation, not the effects of modern generative AI. It does not prove that a future AI crash would cause a similar wave of job losses, but it provides historical context for considering how economic downturns can influence employers' decisions about technology and staffing.
During a recession, businesses may face falling revenues and greater pressure to reduce operating costs. Employers could freeze recruitment, restructure departments or cut jobs to protect their finances.
If AI can perform certain tasks at a lower cost than employing people to do them, companies under financial pressure might have a stronger incentive to automate. Tasks that were previously too expensive or difficult to automate could become more attractive targets if the technology becomes cheaper and sufficiently reliable.
This creates the first part of the potential double blow: an economic downturn could put jobs at risk through conventional cost-cutting while also encouraging employers to consider replacing some workers with automated systems.
Why an AI Bubble Could Burst
The scenario begins with a possibility familiar from previous technology booms: investment can run ahead of the returns companies eventually generate. A technology can be transformative without every business built around it becoming profitable.
Artificial intelligence has attracted substantial investment in data centres, computing infrastructure and the energy needed to power increasingly capable systems. However, the scale of that spending raises questions about whether companies will generate enough revenue to justify their investments and borrowing.
According to Rosalsky, an analysis by JPMorgan published in October found that a basket of 42 AI-related stocks accounted for around two-thirds of the S&P 500's price gains since January 2024.
The same JPMorgan analysis attributed 67% of the index's price returns, 58% of its earnings growth and 81% of its capital spending and research and development growth since January 2024 to that basket of 42 AI stocks. These are separate measures, and the figures do not establish that the market is in a bubble.
Meanwhile, AI-related companies are borrowing hundreds of billions of dollars to help finance data centres and the energy infrastructure needed to run them, according to the NPR report.
If expected revenues fail to justify that spending, some companies could struggle to repay their debts or sustain their investments. Falling asset prices and financial losses could then put pressure on lenders and businesses, potentially worsening an economic downturn.
However, a crash would not necessarily trigger a systemic financial crisis. Rosalsky notes that the bursting of the dot-com bubble was followed by a relatively mild recession, with debt playing a smaller role in the boom than it did in the housing-market collapse around 2008.
The comparison illustrates why an AI bust would not automatically produce a crisis on the same scale. Nevertheless, even without a systemic financial crisis, a downturn could leave businesses under pressure to cut costs.
Could Cheaper AI Make Replacing Workers More Attractive?
The second part of the scenario concerns the potential effect of competition on AI prices.
Technology analyst Benedict Evans has argued that competition between companies offering similarly capable AI models could undermine their pricing power, leaving them with limited profits despite the substantial costs of developing and operating the systems. His analysis of token pricing also highlights the uncertainty surrounding supply, demand, efficiency and the long-term profitability of AI models.
If providers cut prices to attract customers or retain market share, businesses outside the AI industry could gain access to cheaper tools. That would not necessarily mean AI companies were profitable: providers could face financial difficulties even as customers benefit from lower prices.
For employers, that combination could change the economics of automation. If AI becomes cheaper while a recession puts pressure on operating budgets, replacing some tasks previously performed by employees could become a more attractive way to reduce costs.
The potential irony is that an AI market crash could hurt investors and technology companies while making automation more affordable for other businesses. As Rosalsky puts it, 'Investors could lose their money, and workers could still lose their jobs.'
However, lower AI prices would not automatically lead to widespread job replacement. Businesses would still need to consider implementation costs, reliability, the nature of the work and whether automated systems could perform tasks to the required standard.
A downturn could also leave companies with less money to invest in new technology, limiting adoption even if prices fall. The result would depend on how these competing pressures affect individual businesses and industries.
Why the Outcome Remains Uncertain
Rosalsky stresses that several parts of the scenario remain unresolved: whether AI companies can generate enough revenue to justify their spending and borrowing, whether a market crash would trigger a painful recession, and whether the automation patterns observed during earlier downturns would recur with AI.
The historical research provides a reason to examine the possibility, not proof that an AI crash would cause a new wave of job losses.
AI could also prove more useful for helping employees become more productive than for replacing them outright. Businesses might use the technology to improve existing roles, expand their capabilities or support workers rather than reduce their headcount.
The distinction matters because the economic consequences of AI will depend not only on how capable the technology becomes, but also on how employers choose to use it. Falling prices could make automation more attractive, but they would not make every job replaceable or every automation project worthwhile.
Rosalsky also points to possible policy responses, including unemployment insurance and proposals to tax AI and robots. Such measures could help support workers who lose their jobs, although their effectiveness would depend on how they were designed and implemented.
Ultimately, an AI market crash would not necessarily end the advance of automation. If a downturn pushes businesses to reduce costs while competition makes AI cheaper, some workers could face a greater risk of displacement even as AI companies struggle financially. Whether that double blow materialises remains an open question, not an established forecast.
