The International Monetary Fund’s forecast for the world economy is increasingly a forecast for AI.

Global growth is tracking at about 3% this year, exactly where it was expected to be 18 months ago, despite two wars, $100 oil, simmering trade battles, and higher borrowing costs. IMF chief Kristalina Georgieva’s explanation: The Middle East conflict is a supply shock dragging growth down, while the AI investment boom is a demand shock pushing it up.

She worries about what happens when one side gives. Stockpiles cushioning the oil shock will run out eventually, and the AI boom is only as durable as investor enthusiasm. Georgieva, who has run the IMF since 2019, points to Amara’s Law — that we overestimate new technology in the short run and underestimate it in the long run — as a warning that the risks are front-loaded, as investment piles up before widespread adoption brings economic gains to justify it. The bill is also landing on countries that did nothing to run it up. Emerging economies that spent years tightening their budgets, working their way back into bondholders’ good graces, are watching rising Treasury yields erase those gains — “the punishment for somebody else’s sins,” she said.

Liz Hoffman: There are two major wars, oil is at $100, and everyone is worried about asset bubbles. Yet the IMF’s growth forecast for 2026 is 3%, exactly where it was 18 months ago, and the outlook for 2027 has actually improved. How?

Kristalina Georgieva: The negative supply shock from the war in the Middle East pulls the economy down. The positive demand shock from AI is pushing it up. So far, these forces have balanced each other.

The second reason is we once again have demonstrated that there is a lot of creativity in how to deal with problems. You hear a lot of talk about, ‘oh, cooperation is dead.’ But actually, we cooperate because of enlightened self-interest. The United States, Norway, a couple of countries in Africa, have increased oil production very rapidly. Saudi Arabia and the Emirates found ways to bring oil to market through different routes. We have seen the incredible restraint of a large economy like China in going to market to buy oil, using its reserves [instead]. As an economist, I say bravo: We react to market signals. Prices go up, we consume less.

Now, can I give you some bad news? We cannot take this resilience for granted. This energy shock is not over, and nobody can tell you when exactly it will end. What I know is it will get worse before it gets better. Do you like Game of Thrones? Winter is coming. In the Northern Hemisphere, we will use more.

Is that a literal winter, or a metaphorical winter that you’re warning of?

Actually, I’m warning you. We will see pressure building up. We might see prices of refined products going even further up. That pushes inflation up, and that forces central banks to [raise interest rates]. In normal times, that would not be a giant problem. But these are not normal times, because we have debt levels in advanced economies at historic highs. And when interest rates go up, interest payments go up, and that suffocates the government’s ability to do anything, including to help people with the high cost of living. So we should prepare for people being more unhappy in many places, maybe on the street. Now enters AI. The reason there is so much enthusiasm is because AI companies are profitable. And AI is no [longer] an exciting story only for the United States. You look at the AI supply chain, and you see countries in Asia plugged into it, like Malaysia, Thailand, Singapore.

Imagine what would happen if our enthusiasm cools off and money pulls out from the massive investments that have been made. Then we can be in real trouble. So be careful. Keep your eyes open and listen to the IMF.

Is the flip side of that AI supply chain that’s been benefiting countries in Asia that we’ve just exported the risk and globalized what might otherwise just be a Silicon Valley bubble?

Yes. It also stabilizes it, because then you have more production that may not be so bubbly. But yes, it makes it a global phenomenon. I think in the long run, we will be amazed at what has been done as a result of this technology. But if we don’t come to our senses to have some ethical foundation and rules, we might not be able to see these benefits coming. I was going to say we may not be able to live long enough, but that’s a little too extreme.

I think in the long run, we will be amazed at what has been done as a result of this technology. But if we don’t come to our senses to have some ethical foundation and rules, we might not be able to see these benefits coming. I was going to say we may not be able to live long enough, but that’s a little too extreme.

AI is often pitched as an equalizer for developing nations. But it requires a lot of limited resources — power, compute — and there are a lot of languages in the world that the current models don’t serve well. How do you think about that? There is, of course, a tremendous opportunity for leapfrogging for developing countries, and also for plugging themselves into the value chain. There is also tremendous risk that the accordion of inequality would open even wider. In developing countries, prerequisites like access to electricity, access to the internet, are just not there. And you can’t really take advantage of AI without them.

Do you worry that at the company level, at the economy level, people are saying, “with AI we can grow our way out of this hole” and that it’s allowing a little bit of financial indiscipline to come in?

Please don’t do that [laughs]. Yes, AI is likely to deliver significant productivity gains. But it’s not going to happen overnight and in the meanwhile, the story of the last years is: shock, governments borrow, spend, then they do nothing to reduce [spending]. So we are on this staircase, not to heaven. I think we are not really seriously assessing the implications of this problem in a world of interest rates going up. Please don’t spend money you don’t have.

If investors can get US government debt at almost 5% today, is that crowding out the ability of emerging economies to borrow?

Many emerging market economies have done miracles to reduce their debt levels, to put in place fiscal discipline, and to shrink spreads for their borrowing. And now as interest rates go up, it’s washing away hard-earned gains in emerging markets. So it is not competition for money per se. But it is the punishment for somebody else’s sins that emerging markets have to bear.

Do you worry that the AI companies are crowding out the ability of the Treasury to borrow? There’s a lot of very juicy investment-grade corporate yields out there.

Do you worry that the AI companies are crowding out the ability of the Treasury to borrow? There’s a lot of very juicy investment-grade corporate yields out there.

Not yet. We don’t see it at this point. They go to different pots of money. The US government, in a quarter, takes $750-plus billion. The whole borrowing from the companies is way below that. But is it possible that this becomes a problem? Yes. We have to watch it.

It’s been more than a decade since an advanced economy came to the IMF for a bailout. Is there some team inside the IMF that is actively preparing for an advanced industrialized economy to have trouble paying its bills?

The honest answer to this question is: I’m not going to tell you. I take my job very seriously. That’s why I’m not going to tell you. I do too, which is why I had to ask.