Bubble, Shock, or Slowdown? Global Markets Confront the AI Era’s Reckoning

At the height of summer, optimism reigned across the world’s financial capitals. Fueled by the explosive promise of artificial intelligence, US stock markets rallied to record highs, with investors betting that a multitrillion-dollar technology boom would insulate growth from the economic fallout of the war between the US and Israel against Iran. Today, that confidence is vanishing. The conflict drags on without resolution, oil has surged above $100 a barrel, government bond markets are in turmoil, and doubts are deepening over whether AI can deliver enough revenue to justify the staggering sums being invested in it. The warning lights are flashing red, and the question on every trader’s mind is whether this is a correction, a slowdown, or the beginning of something far more severe.

Bond Markets Lead the Retreat

The most immediate alarm is sounding in the market for government debt. In the past week, US 10-year Treasury yields climbed above 5%, a threshold not crossed since 2007, with ripple effects reaching households, businesses, and governments across every continent. The reasons are multiple: persistent inflation driven by energy costs, massive US borrowing needs that have pushed national debt above $40 trillion, and the realisation that interest rates will remain higher for longer than anyone anticipated. As bond yields rise, they pull capital away from equities. Why accept the risk of stocks when safe government debt offers returns not seen in nearly two decades?

Central banks are responding with action that compounds the pressure. The Federal Reserve has raised interest rates for the first time since 2023, acting to head off inflation fuelled by the Iran conflict’s impact on energy prices. The European Central Bank has followed suit, the Bank of England is expected to hike repeatedly, and even the Bank of Japan lifted its policy rate to a 31-year high. Each increase makes debt more expensive to service for governments, for companies building data centres, and for families buying homes. Albert Edwards, senior strategist at Société Générale, captured the mood: “These are febrile times. The key worry is whether the oil price shock will necessitate recession-inducing interest rates.”

The AI Question: Promise vs. Profit

Underlying every market concern is the most fundamental question of all: can AI actually pay for itself? The “Magnificent Seven” tech giants Nvidia, Apple, Google, Microsoft, Meta, Amazon, and Tesla have added more than $20 trillion in combined value since the boom began. Yet the gap between expectation and reality is widening. The cyclically adjusted price-to-earnings (CAPE) ratio for the S&P 500 has climbed to nearly 41 more than double its long-term average of 17 and approaching the 44.2 record set just before the dotcom crash of 2000.

Fathom Consulting estimates that for current investment levels to be justified, AI-related revenue across the sector would need to rise by $600 billion to $800 billion within two years, growth that appears increasingly unlikely. The consultancy assigns a 30% probability to the AI bubble deflating in 2027. “The economics behind the capital expenditure boom do not work,” warns Brian Davidson of Fathom. “AI could unlock huge productivity gains, but sales need to increase by hundreds of billions annually to justify the spend.”

Even within the industry, voices of caution are growing louder. Sam Altman of OpenAI, Elon Musk, and Anthropic’s Dario Amodei have all called for a pause in the race toward ever-more-powerful models a development that has only deepened investor anxiety. More than 1,000 clients registered for a Jefferies analyst call titled “AI Extinction Warnings,” a measure of how quickly sentiment has shifted.

Parallels with History

The echoes of earlier eras are impossible to ignore. The dotcom boom and bust of 2000 showed that even revolutionary technologies can destroy fortunes when capital is deployed too fast. “It took a decade or more for demand to catch up with infrastructure laid down in canal, railway, and fibre booms and many investors never recovered,” notes Deutsche Bank’s Adrian Cox. The comparison extends further: in South Korea, millions of retail traders bought AI-linked stocks with borrowed money, mirroring the “margin trading” frenzy that preceded the 1929 crash. When prices reversed, a wave of margin calls forced mass selling. 1.2 million accounts hit with demands for additional funds, equivalent to one in 30 adults.

Individual companies illustrate the strain. Oracle, which surged after announcing a cloud partnership with OpenAI, has seen its stock halve as investors fret over the debt burden of its data centre expansion. Across the sector, the cost of capital is rising just as spending plans peak a precarious combination.

Collapse or Cooling?

Not all forecasts are catastrophic. Optimists point to early signs of real economic gain: US productivity is accelerating, and the UK also driven in part by tech investment is growing faster than any other G7 nation. Andy Haldane, former Bank of England chief economist, suggests a “slow release of air” rather than a catastrophic pop: “I don’t see outright collapse, but a cooling that slows things down.”

The distinction matters. A gradual recalibration would redirect capital toward profitable applications and away from speculative hype, painful for some, but healthy for the long term. A rapid crash, driven by spiralling bond yields and forced deleveraging, would spread damage across the economy. The inflection point, according to Bloomberg strategist Simon White, lies at a 5.25% yield on US 10-year bonds beyond which stocks and bonds could reinforce each other’s decline.

For now, the world holds its breath. The AI revolution is real, but its timeline is longer than markets have priced in. Geopolitical conflict is real, and its inflationary impact is still unfolding. The reckoning is here, not necessarily the end of an era, but the end of the easy part.