Here’s a plot twist nobody saw coming: the companies throwing billions at AI are getting absolutely wrecked, while the companies *selling them the tools* are printing money. JPMorgan just noticed something that should make you nervous—this exact dynamic happened right before the dot-com bubble exploded in 2000.
Let’s break down what’s happening. Chip stocks are having the year of their lives. The Philadelphia Semiconductor Index is up 87% in 2026 alone, and it just posted its best quarter ever. Memory stocks? Even wilder—the Roundhill Memory ETF is up 141% since April. These are the companies making the hardware that powers AI. They’re basically selling shovels during a gold rush, and business is *booming*.
Meanwhile, the Magnificent Seven—Meta, Microsoft, Google, Amazon, and friends—are getting absolutely demolished. These are the companies spending $725 billion this year on AI infrastructure. Meta and Microsoft, two of the biggest spenders, are down 5% and 18% year-to-date respectively. Microsoft just had its worst month since 2000. Ouch.
The irony is delicious: the companies betting the farm on AI are losing money, while the companies selling them the equipment are crushing it. It’s like watching someone take out a massive loan to buy a Ferrari, while the Ferrari dealership is laughing all the way to the bank.
JPMorgan’s Jason Hunter noticed this exact pattern happened in 1999. Back then, companies making communications equipment saw a “parabolic rise” while the companies actually *using* that equipment tanked. Then the whole thing exploded in early 2000. Hunter’s basically saying: “Hey, we’ve seen this movie before, and it doesn’t end well.”
The setup is eerily similar. You’ve got hardware suppliers soaring while the hyperscalers—the companies actually building out AI infrastructure—are getting punished. Investors are clearly asking the uncomfortable question: “Wait, is all this AI spending actually going to make money?” And right now, the market’s answer seems to be “lol, probably not.”
Here’s the thing that keeps JPMorgan up at night: if the Magnificent Seven stocks don’t find some footing this summer, we could be looking at a serious sentiment shift heading into fall. That’s banker-speak for “this could get ugly.”
The real issue? Nobody actually knows if AI will deliver returns that justify the insane amount of capital being poured into it. Goldman Sachs projects that by the end of the decade, AI capex from just four companies could exceed Japan’s entire GDP. That’s not a business plan—that’s a fever dream.
So what does this mean for you? The market is basically saying: “We love the *idea* of AI, but we’re not convinced anyone’s actually going to make money from it.” The hardware guys are winning because they’re selling the picks and shovels. The big spenders are losing because nobody knows if their AI bets will ever pay off.
It’s a classic case of “show me the money,” and right now, the only people making money are the ones selling the equipment. History suggests that’s usually a sign something’s about to break.
