On One Leg

Last month the International Monetary Fund and the World Bank held their annual meetings in Bangkok. Kristalina Georgieva, the Fund’s managing director, described a world economy standing on one leg, the AI leg, and asked what happens if it wobbles.

Anyone who has tried to stand on one leg knows the answer depends on two things. The first is whether the leg is strong. The second is what you are holding on to when it starts to give. A strong leg can carry you a long way. A weak one, with nothing to grab, puts you on the floor. The IMF’s warning is really about both.

Start with the leg. The technology giants are pouring money into chips, data centres and power, much of it borrowed against revenues that have yet to arrive. The question is not whether the buildings are impressive, but whether anyone will pay to use them. In the Fund’s downside scenario, the answer disappoints. Productivity gains fall short, investment drops and asset prices follow. Global output ends up 1.1 per cent below its expected level by 2028, and American output 1.8 per cent lower. That is a shortfall in the level of output, not in annual growth, and the Fund’s economists said even this case was unlikely to tip the world into recession. The headline is not a prophecy of collapse. It is a bill that arrives slowly.

Now the second question: what the economy is holding on to. Here the real worry is how the damage would travel. Jason Wu of the Fund’s markets division warned that if investors doubt that AI can make money, shares will lose value, debts will become harder to service and promised revenues may never materialise. That last point matters most. The leading firms have bound themselves to one another through equity stakes, supply contracts and purchase commitments stretching years ahead. One company’s spending is another’s revenue, and that revenue justifies the next round of spending. While the circle turns, everyone looks rich. When it stalls, trouble at one firm becomes trouble for every partner and lender beside it. Each is holding on to the others.

Trade shows how much rests on that circle. Silvana Tenreyro, the IMF’s chief economist, noted that trade in AI goods grew 29 per cent between 2023 and 2025. That is more than three times the pace of world trade. A reversal would fall hardest on Asia, where South Korea and Taiwan have prospered on semiconductors and Southeast Asia on data centres.

Israel belongs on that list, and Israelis should be asking both questions of themselves: how strong is the leg, and what are we holding on to?

Few economies lean on the AI leg so heavily. The Israel Innovation Authority reports that high tech reached a record 18.3 per cent of GDP in 2025, supplied 58 per cent of exports and produced roughly half of the year’s growth. Nearly four fifths of the sector’s output is sold abroad. When the world’s appetite for AI shifts, Israel feels it early.

Nvidia employs more than 6,000 people across five research centres in Israel. It has chosen Kiryat Tivon for a campus to house up to 10,000 staff, with building from 2027 and first occupants due in 2031. Jensen Huang calls Israel Nvidia’s second home. It is a vote of confidence worth celebrating, and a wager placed on the very leg the IMF is watching.

Then there is the state’s ledger. Israel entered the war in October 2023 with public debt at about 60 per cent of GDP; the Bank of Israel expects 68 per cent by the end of this year. The Fund urged governments to rein in borrowing as bond yields rise, and warned that cyberattacks on payments and markets could shake confidence. A country that must keep room in its budget for the next war has less to spare for a shock from California. The cyber warning lands differently here, because Israel’s cyber industry belongs to the same high tech sector now riding the AI wave. The shield and the target stand on the same leg.

None of this is an argument against the bet. Israeli strength in chips, networking and cyber defense is real, and the companies built on it would outlast a correction even if their share prices did not. It is an argument for asking of every new campus and valuation what it would be worth if the circle stopped turning, and who would be pulled down with it.

The world is paying heavily for buildings it hopes AI will fill. In Kiryat Tivon the land is bought and the drawings are published. For now it is a hillside, and the room is empty. Nvidia is betting that its first staff will move in by 2031, and it may well be right. But whether the room fills will be decided less on that hillside than on the one leg the whole world is standing on.