Storm clouds gather as AI bubble risks, Iran war and soaring bond yields threaten market crash

Three months ago, the financial world felt almost buoyant. The artificial intelligence boom had lifted American stocks to record heights, and investors were betting that the trillions flowing into AI infrastructure would more than compensate for the economic damage of escalating conflict in the Middle East. That confidence has evaporated. Now, as fighting in Iran intensifies with no visible path to resolution, markets are convulsing under the weight of three converging pressures: the Middle East war is driving oil above $100 a barrel and stoking inflation fears; the AI investment spree that was supposed to save the economy shows signs of slowing; and government debt markets are flashing distress signals that have not appeared since 2007.

The most immediate alarm is sounding in the bond market. This week, the US government's borrowing costs climbed to their highest level in nearly two decades, with the yield on 10-year Treasury bonds crossing above 5 percent—a psychological threshold that some analysts believe could trigger broader financial instability. The ripple effects are already visible. Higher government borrowing costs make it more expensive for households to take mortgages, for businesses to finance expansion, and for other nations to service their own debts. Oil prices climbing in response to Middle East tensions are adding fuel to inflation fears, which in turn is forcing central banks worldwide to raise interest rates even as their economies show signs of strain. The Federal Reserve, defying pressure from President Trump, raised rates this week for the first time since 2023. The Bank of England is expected to raise rates four times before the end of next year. The European Central Bank and Bank of Japan have already moved. The logic is sound—higher borrowing costs can prevent temporary price spikes from becoming entrenched inflation. But the human cost is immediate and severe: households already struggling with energy bills will face higher mortgage payments; businesses will cut hiring; job losses will likely rise.

The second threat is the AI bubble itself. The US stock market's valuation has reached levels not seen since the dotcom crash of 2000. The CAPE ratio, a standard measure of whether stocks are overpriced relative to corporate earnings, now stands at nearly 41 points—more than double its long-term average of 17 and approaching the record of 44 points set in December 1999, just before the internet bubble burst. The seven largest tech companies—Nvidia, Apple, Google, Microsoft, Meta, Amazon, and Tesla—have a combined market value exceeding $20 trillion. For this valuation to be justified, research by Fathom Consulting shows that AI-related sales would need to grow by between $600 and $800 billion within two years. The consultancy estimates a 30 percent chance the AI bubble pops next year, given that such growth appears unlikely. One economist at the firm put it plainly: the economics of the current spending spree do not work. Yes, AI could unlock enormous productivity gains, but the infrastructure is being built far ahead of any proven demand.

The third pressure comes from credit markets showing signs of stress. In South Korea, retail investors have been buying shares in AI-linked chip makers using borrowed money, doubling the value of the country's main stock index. When markets began falling, these investors faced margin calls—demands to deposit more cash to cover their loans. According to Goldman Sachs, 1.2 million South Korean retail investors were hit with such calls, equivalent to one in every 30 adults in the country receiving a sudden demand for more money from their broker. Many were forced to sell shares at losses. Oracle, the database software company, saw its stock halve after announcing ambitious AI infrastructure plans, as investors worried the company was borrowing too much. The Bank of England reported that the gap between yields on safe and risky debt has widened since the Iran war began, a sign that investors are becoming more cautious about lending to companies in distress.

There are historical echoes in all of this. The dotcom crash of 2000 followed a period of euphoria about transformative technology and massive overinvestment in infrastructure that took a decade or more for demand to catch up with. The Great Depression of 1929 was preceded by widespread buying of stocks on margin—borrowed money—that evaporated when the market turned. Some analysts see parallels with the British canal and railway booms, where investors financed far more infrastructure than the economy could absorb, and many never recovered their capital.

Not everyone is convinced a crash is inevitable. Some analysts argue that markets are overestimating the inflation risk from the Middle East conflict, and that geopolitical shocks historically have less economic impact than feared. There are also genuine signs that AI is beginning to drive productivity growth—the US economy has shown rising productivity, and the UK has grown faster than other major economies this year, partly due to AI. If productivity gains accelerate, the sky-high valuations of tech companies might eventually be justified. Andy Haldane, the former chief economist of the Bank of England, said this week that he does not expect an outright dotcom-style collapse, but rather a slow release of air that could slow the global economy without destroying it.

What happens next depends on forces largely beyond investors' control. If the Middle East conflict escalates further, oil prices could spike higher, forcing central banks to raise rates even more aggressively and tipping economies into recession. If AI companies fail to generate the sales growth their valuations demand, the bubble could deflate quickly. If Treasury yields climb above 5.25 percent—the level some analysts identify as a true inflection point—stocks and bonds could begin reinforcing each other's losses in a downward spiral. The market is now a taut wire, and the question is not whether there are risks, but whether the world's policymakers and investors can navigate them without a crash.