US AI-themed stocks suffered a sharp selloff in the week of 23 June 2026, with the so-called Magnificent Seven — Apple, Microsoft, Alphabet, Amazon, Meta, Nvidia, and Tesla — sliding into correction territory as investors questioned whether the spending boom supporting them could last, according to The Guardian. The losses rippled into Asian markets overnight, extending the decline across regions.

The immediate trigger was a shift in how investors viewed AI infrastructure spending. For roughly two years, these seven companies traded as a single coherent group: large US tech names with direct or leveraged exposure to AI buildout. That cohesion is now breaking apart. Most of the seven are underperforming the broader S&P 500 on a trailing basis — a structural change that matters beyond the headline moves.

What the Sell-Off Reveals

The question of whether AI spending can sustain itself is not new. But the market has now shifted from taking management at face value to pricing in genuine doubt. The hyperscalers — Microsoft Azure, Google Cloud, Amazon AWS, and Meta's infrastructure arm — are spending hundreds of billions annually on AI. That spending was supposed to justify Nvidia's valuation jump and lift semiconductor and data-centre companies alongside it. The bullish case always rested on two linked bets: that AI-driven revenue would grow fast enough to pay back the investment, and that competition would not crush profit margins before that payback arrived.

Neither of those bets has been cleanly validated. AI products have generated some revenue growth, but analysts genuinely disagree on whether the extra revenue from AI is growing as fast as the extra spending on it. When that gap widens or stalls, a trade priced for perfection reprices sharply.

The spread into Asia is straightforward mechanics. Taiwan's TSMC and South Korea's memory chip makers depend heavily on orders from US cloud giants. When doubt about those orders emerges in New York, it reaches Hsinchu and Seoul within hours.

Why the Seven Are Splintering

The fact that the Magnificent Seven are now lagging the market as a group may be more important than the overall decline. Their momentum was self-reinforcing: passive funds tracking broad US indices automatically bought more of the same big names, amplifying moves. When that correlation snaps — when some members hold while others sink — the trade unwinds in unexpected ways. Portfolio managers who built overweight positions on the assumption that the group would move together now face a different problem: picking individual winners in companies that were valued on a single macro thesis, not on their separate fundamentals.

Nvidia shows this dynamic most clearly. Its valuation absorbed the entire AI spending supercycle as settled fact. Any shrinkage in that cycle's scale or length hits Nvidia proportionally harder than it hits Microsoft, whose revenue comes from many sources across enterprise software and cloud.

The divergence also reflects real differences in execution. Meta's AI investments are showing up in advertising returns. Apple's AI features remain mostly story rather than a measurable revenue driver. Alphabet is squeezed between defending search revenue from AI alternatives and spending to stay competitive in model development. Treating all seven as identical components was always a shortcut; the market is correcting it.

Whether this is a normal pullback within a still-bullish trend, or the start of a longer repricing, hinges on data that has not yet arrived: Q2 earnings and guidance from the hyperscalers, signals from Nvidia's order book, and — crucially — whether businesses are turning AI pilots into real, profitable ongoing workloads. Until that data comes, the repricing is a shift in mood, not in measurable fact. That gap between mood and reality is where the biggest mispricing risk sits, in either direction.