The great lesson of the last 20 years of Silicon Valley is this: Meta, Amazon, and Google—and now their younger, hoodie-wearing cousins in the AI labs like OpenAI—have thoroughly reshaped the world, made piles of money so large they require their own weather systems, and done it all while mostly shrugging at the mess left behind. They’ve pursued growth and scale with the single-minded focus of a Roomba that’s found a power outlet, and to be fair, they’ve largely succeeded. The data-center boom is the final boss of this strategy: scale, for the sake of more scale, forever and ever. For everyone else, the reward appears to be a front-row seat to very real, very uncomfortable disruption—and no option to click “skip ad.”

Even as tech stocks have blasted off since 2022, the companies’ share of S&P 500 profits has stubbornly refused to join the ride. Job openings are shrinking in the middle of a stock-market victory lap, 22 states are either in or flirting awkwardly with recession, and while data centers are heroically bench-pressing the construction industry, U.S. manufacturing is quietly sneaking out the back door. It’s an economy that looks fantastic on a chart—and increasingly confusing everywhere else.

Here’s How the AI Crash Happens