THE MACHINE BREAKS. NOW WE FIND OUT WHAT COMES NEXT.

Two days ago, the market decided that all the money poured into AI infrastructure maybe wasn't a good idea.

Not because anyone actually knows the final bill. Not because the math suddenly doesn't work. But because real institutions with real capital started asking the question that should have been asked six months ago: If we're spending $300 billion a year on chips and servers, what exactly are the economics of the companies supposed to look like when all this pays off?

The Nasdaq lost 2% on Thursday. Amazon is on its longest losing streak in nearly 20 years. Cisco—which has been on the NYSE since 1990—finally admitted in guidance that the kingdom they've been selling shovels to might be hitting capacity, and it dropped 12%. CBRE, the commercial real estate colossus, cratered 8.8%. C.H. Robinson, a logistics giant with exposure to everything that moves, fell 14.5%.

Here's what matters: This isn't rotation out of speculative nonsense into boring defensives. This is contagion. This is the market beginning to worry about the negative side of the artificial intelligence buildout, which threatens to disrupt entire business models and raise unemployment. Software firms already got hammered. Financial stocks got hit when Altruist launched a new AI-powered tax planning tool—not because it's a good tool, but because the existence of it means Morgan Stanley and LPL Financial have a shelf life. Now it's spreading to infrastructure, logistics, real estate. The dominoes are too closely packed.

Friday brought a small reprieve. The consumer price index rose 0.2% in January, reflecting a 2.4% annualized gain, lighter than the expected 0.3% monthly increase and 2.5% year-over-year. This wasn't supposed to be good news—it was supposed to be deflationary panic. Instead, the market took it as confirmation that the Fed will cut rates multiple times this year, maybe starting as early as June. Treasury two-year yields dropped to their lowest level since 2022. Bond markets cheered.

But equities barely moved. The S&P 500 ended the week at 6,836, up just 0.05%. The Nasdaq dipped 0.22%.

And here's the structural insight everyone's dancing around but won't say plainly: The rotation isn't broken. It's accelerating.

The Russell 2000 is up 7.5% year-to-date while the Nasdaq is down 1%. International equities are up 7% while American mega-caps are getting picked apart. A gauge of megacaps lost 1.1%. This isn't a market correction. This is a repricing of the entire leverage structure that held together for the last 18 months.

The megacaps—Nvidia, Apple, Alphabet, Amazon, Meta—were the only game in town. When earnings growth broadened across the S&P 500 and 79% of S&P 500 names beat estimates so far this earnings season with an average upside surprise of 8.2%, the market could have celebrated. Instead, it started asking which of those winners would still be standing in three years when every business process they touch has been touched by a language model.

Cisco's error wasn't that they gave weak guidance. It's that they admitted it. They told the market: the spending cycle has limits. The buildout isn't infinite. The game of perpetual capex expansion has physics.

What happens next depends on whether this is a forced liquidation of crowded positions or the start of a legitimate structural repricing. The answer is: both. And both take time.

The Fed is pricing in rate cuts. Banks are underwater on their bond portfolios. Real estate is empty. The labor market is still strong enough that unemployment hasn't spiked—non-farm payrolls increased by 130k in January and the unemployment rate eased 0.1pp to 4.3%—but what happens when software layoffs ripple through the data centers? When the $300 billion spend finally generates actual revenue, not theoretical revenue, and it turns out the return on that capital is closer to 5% than 50%?

The machine doesn't break because of one Cisco call. It breaks because the entire system was priced for a single outcome—infinite growth in AI capex—and now market participants are doing what they always do when assumptions crack: they're rotating into whatever is most different from what they owned yesterday.

That's what Friday's rally actually was. Not relief. Permission to start hedging.

The VIX closed at 20.82. That's not panic. But it's not complacency either. It's the sound of people in a big room all deciding to move toward the exits at exactly the same time.