The Global Economy Is Leaning Dangerously Hard on the AI Boom

Every so often an economy comes to rest on a single wager, and for the moment the rich world's wager is artificial intelligence. The spending on it has grown so vast, and so central to the numbers that define prosperity, that growth, share prices and even the arithmetic of national budgets now lean on the same assumption. That assumption is that the technology will soon deliver a productivity windfall large enough to justify the enormous sums being poured into it. It might. But a great deal is riding on a bet that remains, for now, unproven.

The dependence is easy to miss because it hides inside otherwise cheerful headlines. In the first quarter of this year, investment tied to artificial intelligence accounted for something like three quarters of all American economic growth. The building of data centres, the buying of chips, the wiring and the power to run them did the work that consumers usually do, and the consumer, long the engine of the American economy, nearly vanished from the story. Strip out the AI spending and the picture of a healthy expansion begins to look thin.

A boom measured in the trillions

The raw figures are without recent precedent. Capital spending on AI infrastructure is on course to add about 2.5 percent to American output this year and more than 3 percent next, and such investment now amounts to roughly 5 percent of the whole economy. The last time capital formation reached that height was the telecoms and dot com frenzy of the late 1990s, when money flooded into fibre optic cable and networking gear in much the same giddy spirit. That episode is not remembered as a happy precedent.

The concentration is as striking as the scale. A handful of giants, among them Amazon, Microsoft, Alphabet, Meta and Oracle, have between them committed well over 440 billion dollars to this build out in a single year, and the wider group of cloud providers is spending closer to 830 billion. A similar concentration shows up in the stockmarket, where a slim band of AI linked companies has driven the bulk of recent gains. The prosperity is real, but it rests on the decisions of remarkably few firms.

The state joins the wager

What turns a corporate spending spree into a public gamble is the way governments have quietly folded the same optimism into their own plans. Faced with swelling deficits and aging populations, finance ministries have begun to lean on the hope that an AI driven surge in productivity will lift growth, widen the tax base and ease the strain on the public finances. It is a seductive idea, because it promises to solve a fiscal problem without the pain of higher taxes or lower spending.

The awkward truth is that the sums do not obviously add up. Careful analysis suggests that even a genuine AI productivity boom would do surprisingly little for a country's budget under existing tax rules, because much of the gain would flow to capital and to a narrow set of firms rather than into broadly taxed wages. The windfall that ministers are counting on may be smaller and later than they assume, while the bills for energy, infrastructure and interest on the debt are arriving now. Betting the public finances on the boom looks less like prudence than like wishful thinking dressed up as a forecast.

What happens if it stumbles

The danger is not that artificial intelligence is a hoax, for it plainly is not. The danger is that the economy has come to need the boom to continue at full speed simply to stand still. If the pace of investment merely slows, because the returns disappoint or the financing tightens, the growth that AI has been generating would fade at the same time, and there is no obvious second engine waiting to take over. A technology that was supposed to be pure upside has quietly become a source of fragility.

History offers an uncomfortable rhyme. The late 1990s taught that a real and transformative technology can still be paired with wildly excessive investment, and that the correction, when it comes, punishes the overbuilt and the overborrowed regardless of how genuine the underlying revolution turns out to be. The internet changed the world exactly as promised, and the companies that financed the change on the assumption that it would pay off immediately were still wiped out in the crash.

A more honest reckoning

None of this argues for turning away from artificial intelligence, which would be its own kind of folly. It argues for honesty about what is being staked. Investors should remember that a wonderful technology and a wonderful investment are not the same thing. Governments should stop pencilling in an AI miracle to paper over deficits they lack the courage to address directly, and should build their budgets on revenue they can actually count on.

The boom may well vindicate the believers, delivering the productivity leap that makes today's spending look modest in hindsight. But an economy that can only grow while a handful of companies keep pouring money into one technology is not as strong as its headline figures suggest. The wise course is to enjoy the boom while planning for the possibility that it disappoints, rather than treating a bet as though it were a certainty. Right now, far too many are doing the opposite.