For three years we were told that the artificial intelligence buildout was destiny: too big to question, too important to regulate, too promising to tax. Then investors blinked, and in a week the destiny looked like a trade.
The details are revealing. Chip, cloud and data center stocks sold off. Money moved into drugmakers, food companies and consumer goods, the things people buy when they are nervous. Microsoft was rewarded for spending less. Amazon, meanwhile, told shareholders to expect big returns from its chip investments in the second half of 2027. That is a long time to ask anyone to take it on faith.
The market does not know whether this spending is wise. It knows only whether the spending is currently applauded. Yesterday the applause went to whoever spent the most. This week it goes to whoever spent the least. A technology that supposedly will remake civilization is being priced on the mood of portfolio managers.
That would be Wall Street's problem if Wall Street were the only one exposed. It isn't. Fed Chair Kevin Warsh noted that capital spending has spread well beyond the tech sector. That is the part that should concern everyone. When the buildout reaches construction, power generation, real estate and the credit markets behind them, a pullback is no longer a tech story. It is a story about whether the economy has been leaning on a handful of companies' spending plans to look healthy.
BlackRock's Jeffrey Rosenberg offered a phrase worth pausing on: a slowdown in AI investment could be "disinflationary." Translate it. Less hiring, less building, less demand, and therefore less price pressure. In the bond market, that is a feature. For an electrician in a town that was promised a campus, or a county that granted a decade of tax abatements for a facility that now may not be finished, it is a layoff notice with a technical name.
The costs of this boom were never evenly shared. Communities have been asked to host enormous, power-hungry buildings, often with few permanent jobs. In several regions, soaring electricity demand has fed worries about higher bills for ordinary households. States of both parties competed to hand out tax breaks, because no governor wants to be the one who lost the future. The upside was supposed to be broad prosperity. So far it has been concentrated in shareholders and a few very large companies, while the downside is being socialized in advance.
Democrats should not be smug, because many of us were also cheering. Too many in my own party treated AI investment as a growth story to be courted rather than a power question to be answered. We rolled out incentives without clawbacks, welcomed the megaprojects without demanding enforceable local hiring, and accepted corporate promises about efficiency without asking who would be made redundant by it. If the bubble deflates, we will have no standing to say we were not warned.
The lesson is not that AI is fake. Some of it is plainly useful. The lesson is that a technology this consequential should not be governed by whichever executive most recently impressed an analyst. If public land, public power grids and public tax breaks are underwriting the buildout, the public is an investor, and should be treated as one. That means binding job commitments, ratepayer protections that keep residential customers from paying for corporate demand, disclosure of energy and water use, and clawbacks when promised projects vanish.
The companies want the public to take the risk and wait until 2027 for the reward. The public is entitled to ask for terms first.
How it may affect me
If you own index funds or a 401(k), a large share is likely tied to the same AI-linked megacaps that just sold off, so check your concentration. Homeowners near planned data centers should watch local utility rate filings and tax-abatement votes, since ratepayers and county budgets can be left exposed if projects stall.
