For more than a year, the promise of artificial intelligence has functioned as a kind of secular faith in financial markets — a belief that transformative returns lay just ahead, justifying almost any price. Now, with US 10-year Treasury yields reaching 5.16 percent, their highest point since the eve of the 2008 financial crisis, that faith is being asked to compete with arithmetic. When risk-free bonds pay more than 5 percent, the patient optimism required to hold expensive, future-dependent stocks becomes harder to sustain — and the debt-funded machinery powering the AI expansion becomes cos
Rising US bond yields pose existential threat to AI-driven market boom
The math that has held the market aloft is suddenly looking fragile.
Mark
So the bond yield spike is the threat here—but why does that matter so much to AI stocks specifically?
Mimi
Because AI companies are borrowing enormous sums to build data centers and buy chips. When borrowing gets more expensive, those projects become harder to justify financially. The whole sector is betting on future profits that will be discounted at lower rates. Higher rates mean those future profits are worth less in today's dollars.
Luke
But the market hasn't actually sold off yet. How much of this is real risk versus financial media alarm?
Mimi
That's fair. The stock market has absorbed a lot of bad news this year without breaking. But the bond market structure has changed—it's now dominated by leveraged traders rather than stable foreign buyers. That makes it more prone to sudden unwinding.
Mark
What does that unwinding actually look like?
Mimi
If traders get spooked and start closing positions simultaneously, you get a feedback loop. Selling begets more selling. The MOVE index—the bond market's volatility measure—jumped from 78 to 96 in a week, suggesting some of that is already happening.
Luke
Is there any scenario where this doesn't become a problem?
Mimi
Yes. If AI actually delivers the productivity gains companies are betting on, earnings growth justifies the valuations even at higher discount rates. But that's a big if, and it requires those gains to materialize before debt service becomes crushing.
Mark
And the government debt situation—is that connected?
Mimi
Absolutely. The US is running a 6 percent deficit on $40 trillion of debt. As old cheap debt matures and gets refinanced at higher rates, the government's interest costs climb. That crowds out other spending and could slow the economy, which would hurt corporate earnings.
Luke
So we're really asking whether the AI boom can outrun the cost of capital before the whole structure becomes unsustainable?
Mimi
Exactly. There's a window. How long it stays open is the multitrillion-dollar question.
O Pulso
- US Treasury yields have surged to a 19-year high of 5.16%, making bonds a genuine rival to equities for the first time in a generation and threatening the valuation logic underpinning AI stocks.
- The AI sector is burning through more than $700 billion a year in debt-financed investment — data centers, chips, energy infrastructure — and every basis point rise in borrowing costs makes those projects harder to justify.
- The bond market's traditional stabilisers — foreign central banks — have retreated, replaced by leveraged hedge funds whose hair-trigger strategies sent volatility indicators surging nearly 25% in a single week.
- Beneath strong corporate earnings lies a circular economy where AI companies inflate each other's valuations through cross-holdings, masking the absence of broad external revenue.
- With US government debt exceeding $40 trillion, deficits near 6% of GDP, and equity valuations at dot-com-era extremes, the margin for error is narrowing — and credit markets are already beginning to price in the risk.
For more than a year, the promise of artificial intelligence has functioned as a kind of secular faith in financial markets — a belief that transformative returns lay just ahead, justifying almost any price. Now, with US 10-year Treasury yields reaching 5.16 percent, their highest point since the eve of the 2008 financial crisis, that faith is being asked to compete with arithmetic. When risk-free bonds pay more than 5 percent, the patient optimism required to hold expensive, future-dependent stocks becomes harder to sustain — and the debt-funded machinery powering the AI expansion becomes costlier to run.
The calculation that has kept stock markets near record highs is showing cracks. In a single week, yields on US 10-year Treasury bonds climbed to 5.16 percent — a level not seen since 2007, just before the financial crisis. The significance is structural: when government bonds pay more than 5 percent with virtually no risk, a stock market yielding barely 1 percent in dividends starts to look difficult to defend. The sector most exposed is the one that has driven most of the market's gains — artificial intelligence.
The AI industry has absorbed more than $700 billion in investment across the United States in the past year alone, with projections pointing toward $1 trillion next year. That capital is flowing into data centers, semiconductor plants, and the energy infrastructure to power them — and most of it is borrowed. Even the largest technology companies, the hyperscalers like Amazon and Alphabet, have been tapping debt markets at scale, a sign that their cash flows can no longer keep pace with their ambitions. Rising bond yields make that borrowing more expensive and the economics of long-horizon capital projects harder to sustain.
The bond market itself has become less stable. Where foreign central banks once provided steady, predictable demand for US Treasuries, trade tensions and sanctions have made them more cautious. Leveraged hedge funds have filled the gap, running strategies that can unwind rapidly when sentiment shifts. The MOVE index — a measure of bond market volatility — jumped from around 78 to 96 in a week, suggesting some of those players were already repositioning.
The stock market has so far absorbed the yield rise without a serious sell-off, supported by strong corporate earnings and confidence in AI's transformative potential. But that earnings growth contains a quiet fragility: much of it reflects AI companies marking up the value of their stakes in one another, a circular flow of paper gains rather than revenue earned from the broader economy.
The wider backdrop is growing harder to ignore. The US government carries more than $40 trillion in debt while running a deficit near 6 percent of GDP. The Federal Reserve has resumed raising rates, and markets expect further increases. Equity valuations, measured by cyclically adjusted price-to-earnings ratios, sit at levels last seen in 1999. The yield curve is flattening. The cost of insuring against corporate defaults has risen sharply, with some hyperscalers now costing more than $900,000 to insure per $100 million of debt. Whether the AI sector can deliver on its promises before the cost of capital becomes prohibitive is the question on which a multitrillion-dollar bet now turns.
The math that has held the stock market aloft for the past year is suddenly looking fragile. In the span of a single week, yields on US 10-year Treasury bonds climbed from 4.95 percent to 5.16 percent—the highest level since 2007, just before the financial crisis. The jump matters because it reshapes the entire calculation investors use to decide where to put their money. When bonds start paying more than 5 percent with virtually no risk, a stock market trading near record highs and yielding only about 1 percent in dividends begins to look expensive by comparison. That shift in perception could unwind the boom in artificial intelligence stocks that has driven much of the market's gains.
The timing is particularly precarious for the AI sector. Over the past year, the industry has absorbed more than $700 billion in investment across the United States alone, with projections that spending could exceed $1 trillion next year. This money is flowing into data centers, semiconductor fabrication, and the energy and water infrastructure needed to support them. Most of it comes from debt markets. When borrowing costs rise—as they do when bond yields climb—the economics of these massive capital projects become harder to justify. The companies leading this expansion, the so-called hyperscalers like Amazon and Alphabet, have already begun tapping debt markets at scale, a sign that even their enormous cash flows can no longer cover the appetite for AI investment. Higher interest rates make that debt more expensive and harder to service.
What makes this moment particularly unstable is the structure of the bond market itself. A decade ago, foreign central banks and institutions were the dominant buyers of US Treasury bonds, providing a steady, predictable source of demand. That has changed. Trade wars and sanctions have made foreign buyers more cautious, diversifying their holdings into gold and other assets. In their place have come hedge funds and other traders running leveraged strategies—borrowing cheaply in one market to profit from price differences in another. This shift has made the bond market far more volatile and prone to sudden reversals. When traders become spooked, they can unwind their positions quickly, amplifying price swings. The MOVE index, which measures bond market volatility, spiked from its typical level of around 78 to 96 in a single week, suggesting that some of these leveraged players were already getting nervous.
The stock market has absorbed the rise in yields without a major sell-off so far, buoyed by conviction that AI will deliver transformative productivity gains and by a strong US economy. Growth has been supported by government spending—the federal deficit is running at nearly 6 percent of GDP—and by continued spending from wealthier households, whose finances have been cushioned by stock market gains and tax cuts. Corporate earnings have grown at roughly 25 percent year-over-year in recent quarters. But those numbers contain a hidden fragility. Much of the profit growth in the AI sector comes from companies marking up the value of their holdings in each other, creating what analysts call a circular economy where paper gains circulate within the industry rather than reflecting genuine revenue from outside customers.
The larger economic backdrop is becoming unsustainable. The US government is carrying more than $40 trillion in debt while running deficits that cannot be financed indefinitely at rising interest rates. The Federal Reserve has just begun raising rates for the first time in more than three years, and markets are pricing in at least three more increases over the next year. As older, cheaper debt issued after the 2008 crisis matures and must be refinanced at higher rates, the cost of servicing government and corporate debt will only climb. The stock market is trading at a cyclically adjusted price-to-earnings ratio last seen in 1999, just before the dot-com bubble burst. The yield curve has flattened, with the gap between two-year and 10-year bond yields shrinking from 85 basis points at the start of the year to 30 basis points, a pattern that historically precedes economic slowdown.
The cost of insuring against corporate defaults has already begun to rise. At the start of the year, it cost $510,000 to buy insurance against a $100 million default on investment-grade corporate debt. That price has climbed to $580,000. For some of the hyperscalers, the cost exceeds $900,000 per $100 million—a signal that markets are beginning to price in real risk. The question now is whether the stock market can continue to ignore these warnings, as it has shrugged off trade wars, Middle Eastern conflict, and oil prices above $100 a barrel. The answer depends on whether the AI sector's promises of transformative returns can be realized before the cost of capital becomes prohibitive. If not, the multitrillion-dollar bet on artificial intelligence may be about to face its first serious test.
Citações Notáveis
The resort to debt by even the hyperscalers is an indication of the stresses even the biggest and most profitable companies are experiencing in funding their ever-escalating AI investments.— Stephen Bartholomeusz, The Age
Quer a matéria completa?
Leia o original em The Age ↗