The AI trade is splitting in two, and JPMorgan thinks the next few weeks will determine whether that’s a healthy rotation or the early rumblings of something uglier.
JPMorgan strategist Jason Hunter flagged a growing divergence between AI chip and infrastructure stocks and the hyperscalers pouring hundreds of billions into AI buildout. The Philadelphia Semiconductor Index has surged 87% year-to-date through early July 2026. Meanwhile, Meta is down 5% and Microsoft is down 18% over the same period, with Microsoft posting its worst monthly loss since the year 2000 in June.
The dot-com echo nobody wants to hear
Hunter’s note draws a pointed comparison to the late 1990s, when hardware and infrastructure companies ran hot while the companies spending on that infrastructure started wobbling.
Today’s version features four hyperscalers, Meta, Microsoft, Amazon, and Alphabet, projected to spend a combined $725 billion on AI-related capital expenditures in 2026. JPMorgan separately estimated that five major hyperscalers would collectively spend around $697 billion.
