For two years, the market's answer to every valuation concern about artificial intelligence was the same: just wait. The infrastructure buildout would justify itself. The hyperscaler capex would translate into revenue. The picks-and-shovels plays would mint money forever. On Monday, patience ran out.

Technology shares resumed their slide, with the sector dragging the S&P 500 and Nasdaq lower as investors rotated out of the AI complex that had powered markets since late 2022. The move wasn't triggered by any single catalyst—no earnings miss, no regulatory bombshell—but rather by the accumulating weight of a question that can no longer be deferred: where, exactly, is the return on all this spending?

The capex problem

The numbers are staggering. Microsoft, Google, Amazon, and Meta have collectively committed north of $200 billion to AI infrastructure over the past eighteen months. Data centers are sprouting across the American Southwest. Nvidia's market capitalization briefly exceeded $3 trillion. Yet for all this activity, the revenue story remains stubbornly incremental. Enterprise adoption is slower than the hype suggested. Consumer applications remain largely experimental. The killer app that justifies trillion-dollar valuations has not materialized.

Investors tolerated this during the zero-interest-rate hangover, when growth at any price still commanded a premium. But with the Federal Reserve signaling that rate cuts may not arrive this year—and inflation proving stickier than hoped—the calculus has shifted. Capital has a cost again, and that cost demands accountability.

Rotation, not capitulation

It would be premature to call this a tech crash. The selloff is orderly, the declines measured. What's happening is better described as a rotation: money flowing out of high-multiple AI plays and into sectors that offer more immediate cash flow visibility. Energy stocks, defensive names, and dividend payers are catching bids. The trade that defined the post-pandemic market is being unwound, not abandoned.

The distinction matters. A rotation implies that capital isn't leaving the market—it's simply demanding different terms. Investors aren't declaring AI a fraud; they're declaring it overpriced relative to near-term fundamentals. That's a healthier correction than a panic, but it's still a correction.

Our take

The AI selloff is less about artificial intelligence than about the market's relationship with narrative. For years, the story was enough. Now the spreadsheet has to match. This is how markets are supposed to work: eventually, the future has to become the present, and the present has to show up in earnings. The companies that survive this reckoning will be the ones that can demonstrate real revenue from real customers doing real things with AI—not just selling shovels to other shovel salesmen. The rest will learn what every bubble eventually teaches: gravity is patient, but it is also inevitable.