NextFin News - The IMF is drawing a more specific line through the AI boom: the bigger worry may not be that technology shares have become expensive, but that the growth of artificial intelligence is increasingly being funded with debt. Tobias Adrian, who leads the fund’s Monetary and Capital Markets Department, told the European Central Bank’s annual forum in Portugal on June 30 that recent market behavior does not necessarily amount to a bubble, even as he questioned how companies are borrowing to finance the expansion.
That distinction matters because the usual valuation debate can miss the way financial stress actually builds. A stock can trade on a rich multiple for a long time if earnings keep improving or if investors stay willing to pay up for growth. Borrowing, by contrast, creates a harder constraint. When companies finance a capital-intensive race with debt, the issue is not only how much the market is willing to pay today, but how resilient those balance sheets will be if the expected returns on AI spending arrive later than planned.
Adrian’s message does not amount to a declaration that AI is in a bubble. It is more precise than that. He is saying the market may be focusing on the wrong side of the AI trade. Valuations tell investors what enthusiasm looks like. Leverage tells them where the pressure will show up first if the cycle slows, cash generation disappoints, or funding conditions tighten.
That is a useful distinction for a sector where the headline narrative still centers on software, models, and user adoption, while the underlying economics increasingly depend on chips, data centers, power contracts, and other forms of physical infrastructure. The AI story has moved well beyond a simple software re-rating. The buildout requires large amounts of capital, and once borrowing becomes a bigger part of the funding mix, the risks move from sentiment to solvency.
In other words, Adrian is not asking whether AI is exciting enough to justify higher equity prices. He is asking whether the financing structure behind the boom can absorb disappointment. That is a different, and often more consequential, question.
Why Debt Changes The AI Debate
Debt matters because it changes the timing of pain. A pricey stock can fall gradually as expectations cool. A leveraged company can run into a more abrupt problem if its cash flow fails to match the pace of its obligations. That is why the IMF official’s warning is best read as a financial-stability concern rather than a valuation call.
The AI economy is expensive to build. It requires infrastructure, specialized equipment, energy, and a constant flow of upgrades. Those spending needs are not inherently dangerous, but they become more sensitive when they are financed with money that expects a near-term return. If the payoff from AI adoption arrives slowly, or if competition makes returns harder to capture, debt can turn a strategic investment cycle into a funding problem.
That is also why leverage is a more revealing lens than a simple multiple check. Valuations can stay elevated if the market remains optimistic. Borrowing, however, forces a company to meet fixed claims. If the business model underperforms, lenders do not reprice the narrative; they reprice the credit. That difference is what makes leverage more worrying in a sector that is still spending aggressively to secure future scale.
“Recent market behaviour doesn’t necessarily signal a bubble,” Tobias Adrian said at the ECB forum, adding that he wonders “about how companies are borrowing.”
The quote is telling because it leaves room for the possibility that AI remains a valid growth story while still warning that the financing path deserves scrutiny. In IMF terms, that is a classic stability concern: not whether one asset class has become fashionable, but whether the capital structure behind the boom has become fragile.
There is also a broader macro point embedded in Adrian’s remark. When money is cheap and growth is abundant, leverage can look harmless. When rates are higher, credit is tighter, or earnings visibility is lower, the same borrowing can become much more consequential. AI is being developed in an environment where the cost of capital matters again, which means the financing model cannot be ignored for long.
Why Valuations May Be The Lesser Problem
Valuations are the market’s most visible anxiety, but they are often the least dangerous one in the early stages of a new technology cycle. Investors can argue about multiples, compare them with history, and decide whether enthusiasm has gone too far. What they cannot easily ignore is the balance-sheet strain that appears when a spending race becomes dependent on external funding.
That is especially true in AI because the return profile is uncertain. The industry has made rapid progress, but the monetization path is still uneven. Some firms are already generating meaningful revenue from AI products and services. Others are spending to avoid being left behind. The more that competitive pressure forces companies to keep pace with rivals, the more borrowing becomes part of the competitive toolkit.
In that sense, Adrian’s concern is less about whether the largest AI winners deserve premium valuations and more about whether the broader ecosystem is piling up obligations to sustain the race. A highly valued company with ample internal cash can absorb volatility. A less mature firm leaning on debt has less margin for error. If the cycle changes direction, the weakest balance sheets will feel it first.
This helps explain why the IMF warning lands differently from the standard bubble talk. Bubble calls usually imply a broad market top driven by exuberance. A leverage warning implies something more targeted and, in some ways, more dangerous: a sector can keep looking healthy in the stock market while its financing base quietly deteriorates.
That is the point at which the story stops being about sentiment and starts being about system risk. If debt-funded AI investment becomes large enough, trouble in one part of the sector can spill into lenders, suppliers, and the wider market for corporate credit. The market does not need a crash in the leading stocks for that to matter.
What Investors Should Watch Next
The next phase of the AI trade is likely to be judged less by headline enthusiasm and more by the terms on which the buildout is financed. That means paying attention to debt issuance, refinancing activity, and the size of capital commitments relative to current cash generation. If companies keep expanding while funding costs remain manageable, the leverage concern may stay theoretical. If spending rises faster than cash flow, the warning becomes more immediate.
It also means watching whether the market starts to separate the strongest AI franchises from the rest of the ecosystem. The biggest technology companies can often fund expansion from cash flow, but smaller names, suppliers, and newer entrants may rely more heavily on external financing. In a late-cycle stress scenario, that gap can widen quickly.
For now, Adrian’s comments suggest the IMF is not trying to puncture the AI story. It is trying to identify where the fragile point might be if the story changes. That is an important distinction. The question is not whether AI can keep reshaping the economy. It is whether the financing of that transformation is becoming more vulnerable than the market realizes.
That is why the warning lands as a balance-sheet story, not just a valuation story. Prices can correct. Debt has to be repaid.
The sharper takeaway is that the AI boom may be less exposed to the problem of looking expensive than to the problem of being funded too aggressively. If that is right, the market’s next anxiety will not be about how much investors are paying for the theme. It will be about who is carrying the bill.
Explore more exclusive insights at nextfin.ai.
