Will it be boom, bust or backlash?
Nov 10th 2025|By Rachana Shanbhogue, Business affairs editor, The Economist
FOR THREE years the world has marvelled at ChatGPT’s cleverness, and more recently, been mesmerised by the creativity of Sora, the chatbot’s video-generating sibling. In the expectation that artificial intelligence will be transformative, large American tech firms poured over $400bn into data centres and other necessary infrastructure during 2025; by one estimate, a staggering $7trn will be spent by the end of the decade. Yet revenues from AI so far amount to a measly $50bn a year, about an eighth of Apple or Alphabet’s entire annual revenues. As the world gets used to the technological feats of AI, the focus is shifting. In 2026 expect its economic, financial and social consequences to grab attention.
Consider economics first. Some 800m people worldwide use ChatGPT; many employees admit in surveys to using AI at work. But when it comes to formal adoption by businesses, the numbers are still modest. According to America’s Census Bureau just over 10% of businesses with more than 250 employees say they have embedded AI into their production processes. A survey by the Massachusetts Institute of Technology released in July found that 95% of businesses’ AI pilots failed to generate a return at all.A market correction in AI would have knock-on consequences for America’s economy
As a consequence, venture capitalists, entrepreneurs and tech giants in Silicon Valley are all focused on one thing: how to speed up adoption by companies. There is a potential fortune to be made if businesses can integrate AI into their operations and boost productivity. Startups have cropped up to help businesses in particular industries, or with particular processes: Harvey AI, for instance, works with lawyers to parse reams of contracts, while Sierra helps companies use AI in customer service. Even OpenAI and Anthropic, the world’s leading AI labs, are tailoring some of their services in order to help financiers or life-science researchers. A key indicator to watch in 2026 will therefore be the rate of formal adoption, and the extent to which these efforts are succeeding.
This matters not just for the impact of AI on productivity and economic growth, but also for the huge financial boom that is predicated on the technology’s success. According to the Bank of England, the stocks of companies that depend on AI made up 44% of the market capitalisation of America’s S&P 500 in early October. The ratio of prices to forward earnings for this basket—a measure of stock valuations—is an eye-watering 31, compared with 19 for the overall index. If adoption picks up, then investors will continue to believe that their extravagance and patience will ultimately result in handsome returns. But any sign that the gains from AI will be slow to arrive, or will not come at all, could cause valuations to deflate.
History shows that even useful technologies, such as railways and the internet, have been accompanied by financial exuberance. A market correction in AI would have knock-on consequences for America’s economy. Investment in data centres and the wealth effects from a roaring stockmarket have helped mask the impact of tariffs, lower migration and uncertainty. But if the AI boom falters, trillions of dollars in American households’ wealth could be wiped away.
Faster uptake of the technology by companies would reassure investors, but risks raising a different concern: about AI’s impact on jobs. To spur adoption, AI firms are pushing virtual “agents” that are capable of performing a series of tasks semi-autonomously, from start to finish—like human employees, but round the clock, and at lower cost. Packaging AI in this way, as a drop-in replacement for software programmers or customer-service agents, may be easier for managers to grasp. Artisan, a startup based in San Francisco, for instance, has run an ad campaign that calls on companies to “Stop Hiring Humans”. But it is hard to think of a more effective way to fuel a backlash against AI as a job-stealing, worker-displacing menace.
Already, some commentators are blaming AI for high rates of graduate unemployment in America. The evidence for this is slim. Graduate unemployment may simply reflect employers’ shifting demand for skills, or could be the result of trends that predate the launch of ChatGPT, such as overhiring by tech and professional-services firms during the pandemic. Some studies, including one by the Yale Budget Lab, conclude that AI has not had disruptive effects on the labour market, and find no evidence that AI-intensive industries have been laying off more workers than others.
Past technology booms suggest that worries about mass unemployment tend not to come to pass. Firms that adopt technology faster benefit from higher demand, leading them to hire more workers; as some jobs become obsolete, new ones crop up. But it is only natural that people worry about change. The hype and the hopes around AI have been like nothing the world has seen before, and the true nature of its impact is still unclear. Will it bring an economic revival, a financial bust or a social backlash, or some combination of the three? The world will start to find out in 2026. ■