Governments are making a dangerous bet on the AI boom

The culprits are clear. Inflation has not quite been beaten, deterring central banks from cutting interest rates. Even Japan is, haltingly, leaving behind its loose-money policies (though not enough to strengthen the yen, which both Japan and America have propped up with official purchases). Gaping rich-world budget deficits show little sign of narrowing meaningfully, raising the possibility that governments eventually force central banks to inflate their debts away. Geopolitical ructions, like the Iran war and tariffs, are lifting the compensation for risk that investors demand. It does not help that Kevin Warsh, the new Federal Reserve chair, fumbled his first impression with markets.

High bond yields are a problem for indebted governments. Every percentage-point rise in America’s bond yields, for instance, costs it extra interest payments worth 1.3% of annual GDP within a decade. Yet it is getting harder to hope that governments will tighten their belts to offset the squeeze. In America each political party seems to be hoping the hot potato of a debt crisis lands on the other. France’s presidential election next year could devolve into a far-left v far-right fight. Japan is, bizarrely, undertaking a fiscal stimulus. The longer governments wait to adjust, the bigger the adjustments become.

Governments look increasingly as if they are betting on economic growth to pay the bills—which today means a bet on ai. The trouble is that faster growth, by boosting investment, usually brings higher interest rates and higher bond yields. Today higher market rates have arrived before faster growth. The sheer scale of data-centre investment, pegged at $1trn this year by Goldman Sachs, a bank, is making capital scarcer and helping raise yields. However, AI has yet to boost productivity measurably, even in China where diffusion of the technology is arguably the furthest along.

It could be a while before the boost comes. Electricity and computing took decades to raise productivity growth perceptibly. Not all economies will fare equally well, either. Europe’s fiscal woes could worsen if AI brings more growth for America, pulling up global interest rates, but gets jammed up in the old world by rigid labour markets and high energy costs. That would mean costlier debt without any more growth.

There is another problem. If the productivity boost from AI is large enough to change any country’s fiscal arithmetic profoundly, vast job disruption is probably coming too, meaning more spending on unemployed workers. Income may also shift from labour to capital, which is more lightly taxed. Should AI prompt an arms race, countries will also need more defence spending. Even the exciting prospect of longer lifespans, fuelled by AI-powered scientific advances, raises state spending on pensions. Economists at the Brookings Institution, a think-tank, reckon these factors plus higher interest rates could more than halve the positive impact of AI-induced growth on American deficits.

The last time much of the rich world, including America and Britain, ran budget surpluses was the turn of the millennium. To do that required fast productivity growth, which back then came from computers and the early internet. But deficits had to be cut too. Without fiscal prudence, bonds are a punt on AI that looks just as risky as betting on wobbly tech stocks.