Davos, January 21st 2026. At 08:30 the World Economic Forum staged a familiar spectacle: today’s troubles explained through yesterday’s decade. In a livestreamed session developed with The New York Times, Andrew R. Sorkin, author of “1929”, asked whether the 2020s are becoming the new 1920s. With him were Adam Tooze; Christine Lagarde, president of the European Central Bank; and financiers Laurence D. Fink of BlackRock and Ken Griffin of Citadel.
Mr Sorkin’s opening caution was sound. Parallels are not prophecies. Yet the similarities are tempting. The 1920s had electrification, assembly lines and motor cars. The 2020s have artificial intelligence and ubiquitous computing. Both decades have buoyant markets and the sense that technology can outrun politics. That is where the danger lies.
Mr Tooze supplied the historical sting. The 1920s, he argued, were an early experiment in a world centred on America. After the first world war, liberal powers failed to build robust political arrangements and leaned on a substitute: money and machines. The new technologies were largely American, as was the finance. The dollar’s rising dominance, channelled through the gold standard, was meant to stabilise a fragile system, until politics and legitimacy caught up with it. The lesson is not that booms must end in catastrophe, but that a monetary order cannot be more coherent than the politics beneath it.
Ms Lagarde updated the analogy with a colder focus on trade. In the 1920s, she noted, global integration slipped even as technology advanced and stock markets rose. Today trade has not collapsed, but fragmentation is building through tariffs, export controls and a growing thicket of restrictions. The twist is that AI is more dependent on cross border scale than its predecessors. Training frontier models requires vast capital, torrents of energy and, above all, access to data. If the world splinters into incompatible standards and fenced off datasets, the promised productivity gains may come late, arrive unevenly, or fail to show up at all. And even if the economics worked, she added, the politics might not: energy intensity, climate consequences and labour disruption can turn a productivity story into a social crisis.
That is also why, in Mr Fink’s telling, geopolitics now sits inside the technology stack. If western countries cannot cooperate on scale, he warned, China will not have the same problem. A huge domestic market and different privacy constraints give it an advantage in data accumulation and fast deployment. On the fashionable question of an AI bubble, he was cautious but not alarmist: there will be failures, yet he doubts the whole boom is froth. The real test is diffusion. If AI remains concentrated among a small club of hyperscalers, the hoped for surge in productivity will disappoint and the political backlash will be fierce.
Diffusion has a political edge. Mr Fink described an economy that increasingly looks K shaped: a handful of winners pulling away, and many firms falling behind. Scale operators have cash, data and expertise, so they adopt AI first and widen their lead. Smaller firms, the usual nursery for competition and social mobility, risk being squeezed. That, more than any stock market chart, is the seed of backlash.
Mr Griffin, cast as the sceptic, shifted the recklessness charge from private markets to public balance sheets. The great excess of the moment, he argued, is government spending beyond means. Policymakers are quietly hoping that AI will deliver a productivity miracle big enough to square the fiscal arithmetic. But hope is not a forecast. He punctured the romance of the investment boom with a hardware question: how long do advanced chips last? If obsolescence comes in a year, today’s data centre build out will look like a monument to haste. If it lasts several years and can be repurposed, it will look like prudent overbuilding.
Debt hovered over the discussion. Mr Sorkin noted that the United States entered the 1920s with a very different fiscal position from today’s. Modern crisis management often involves vast fiscal cheques and central bank support. What happens if bond markets decide the playbook has limits? Ms Lagarde would not name a bright red line. She drew another one instead. Debt used for productive investment, including security, is easier to finance. Debt that does not sustain growth becomes politically and financially brittle. Central banks, she implied, cannot be treated as permanent substitutes for fiscal responsibility.
Tariffs provided the final rhyme. Mr Sorkin recalled Hoover’s 1930 bet that helped push global trade into collapse. Mr Tooze argued the catastrophe required more than tariffs: currency disorder and quotas after the gold standard cracked. Today’s world, with fiat money and more tools, is far from that abyss. Even so, Ms Lagarde and Mr Griffin stressed that tariffs behave like regressive taxes, often borne by consumers and importers, and they invite cronyism.
The panel did not forecast a repeat of 1929. Its warning was subtler and more useful. The 1920s ended badly not because technology was false, but because cooperation failed and politics could not absorb the shocks that prosperity created. If the 2020s are to rhyme rather than repeat, they will need what Ms Lagarde called “minimal cooperation” on trade, standards, data and the social bargain. In Davos, that sounded less like idealism than like basic risk management.