History has seen this pattern before. From railway tracks in Victorian Britain to dot-com bubbles at the turn of the millennium, transformative technologies invite euphoric investment. Capital rushes in, dreams soar higher than fundamentals, and the eventual correction leaves a trail of bankruptcies and cheap infrastructure for the next generation to exploit.
Artificial intelligence is now following that familiar script. By 2028, global spending on AI-ready data centres alone is projected to exceed three trillion dollars. America’s largest technology firms are already pouring hundreds of billions each year into server farms, chips and power-hungry networks. Start-ups with unproven revenue models raise capital in billion-dollar tranches, cheered on by investors convinced that machine intelligence will redefine everything from health care to warfare. Some of these bets will pay off spectacularly, but many will not.
The danger is not merely that investors will lose money. An AI bust could ripple through labour markets, energy grids and municipal finances. Vast data-centre construction projects distort local economies by bidding up land and electricity. When demand collapses, empty shells and stranded power contracts could burden taxpayers. Pension funds that chased AI-themed exchange-traded funds may find themselves underfunded. Governments tempted to subsidise the “next industrial revolution” could be left with white elephants ~ and political backlash. Moreover, the scale of energy consumption required to sustain these vast computing hubs carries environmental and geopolitical implications. Securing steady electricity supplies may intensify competition for renewable resources and strain power grids already battling climate stress.
Regions that welcome AI megaprojects for jobs and tax revenue could later face community opposition when soaring electricity demand drives up local prices or worsens emissions. An investment bust would not only destroy capital but also leave behind ecological debts and contested infrastructure. Yet history offers a paradoxical comfort. Britain’s speculative railway boom left investors destitute, but the tracks they financed still carry passengers today. The dot-com crash wiped out paper fortunes, but also laid the fibre-optic backbone of the modern internet. Even a messy AI correction would leave behind compute capacity, algorithms and trained talent that society can repurpose at bargain prices. The long-term productivity gains may outlast the financial carnage.
The lesson is twofold. Policymakers should resist the urge to pick winners or shield investors from losses; public money should flow to basic research, not speculative valuations. Investors, meanwhile, must distinguish between the enduring utility of AI infrastructure and the frothy multiples of companies racing to dominate it. Capitalism’s genius lies in funding bold experiments and absorbing failure, not in guaranteeing perpetual returns. A trillion-dollar bust would be painful. But if history is any guide, the rails for tomorrow’s economy will be built on the wreckage of today’s exuberance. The real question is whether we can manage the fallout wisely enough to ensure that society, not just speculators, reaps the eventual reward.