JPMorgan’s flagged divergence puts fresh focus on hyperscaler share price behaviour into the summer, with framing of individual stock charts for Meta, Microsoft, Amazon and Alphabet as the key signal to watch for a potential sentiment driven setback into autumn. Chip and memory names have carried the AI trade higher through 2026, but a sustained gap between hardware performance and heavy capex spenders raises the risk of a broader rotation or pullback if investors continue questioning AI monetisation. The parallel drawn to 1999, when communications equipment makers rallied while heavy capital spenders fell before the 2000 crash, adds a historical risk marker without JPMorgan explicitly calling a bubble. Positioning and sentiment around the hyperscaler names are likely to remain a swing factor for broader index performance given their outsized weighting.
—
JPMorgan warns a growing split between AI hardware stocks and heavy AI capex spenders echoes market dynamics from 1999, just before the dot-com crash.
Summary:
- JPMorgan strategists flagged a growing divergence between AI hardware stocks and major AI capex spenders in a client note
- The Philadelphia Semiconductor Index has risen 87% in 2026 and just posted its best ever quarter
- The Roundhill Memory ETF has climbed 141% since its April launch as memory stocks rallied
- The Roundhill Magnificent Seven ETF is down 7% from its peak, with Meta and Microsoft down 5% and 18% year to date respectively
- Microsoft posted its worst monthly loss since 2000 in June
- JPM compared the current split to 1999, when communications equipment makers surged while heavy capital spenders fell before the dot-com crash in early 2000
- Combined AI capex from Meta, Microsoft, Amazon and Alphabet is on track to reach $725 billion this year
JPMorgan has identified a pattern in markets that echoes conditions seen shortly before the dot-com crash, according to a client note from strategists. The signal centres on a widening split between AI hardware stocks and the companies spending most heavily on AI infrastructure, a divergence analysts says mirrors what unfolded in the final months of 1999.
Chip and memory stocks have led markets for much of 2026. The Philadelphia Semiconductor Index has surged 87% this year and just delivered its strongest quarter on record, reflecting intense investor focus on the physical AI buildout. Memory stocks have performed similarly well, with the Roundhill Memory ETF up 141% since it launched in April.
That strength stands in contrast to the fortunes of the biggest AI spenders. The Roundhill Magnificent Seven ETF has fallen 7% from its recent peak, as investors increasingly question whether massive AI investment will generate adequate returns. Meta and Microsoft, both among the heaviest AI capital spenders, have been hit particularly hard since the start of the year, down 5% and 18% respectively, with Microsoft posting its worst monthly performance since 2000 in June.
JPM draws a direct line to 1999, when communications equipment suppliers experienced a sharp rally even as companies making large capital investments in that buildout began falling from their highs. That divergence preceded the bursting of the dot-com bubble in early 2000, within roughly a year of the split first becoming visible. Hunter described the current setup as reminiscent of that dynamic, pointing to the negative price performance among hyperscalers alongside the widening gap with hardware names. He said the pattern keeps his focus trained on individual hyperscaler charts through the summer, watching for signs of stabilisation that could reduce the risk of a sentiment and positioning driven setback later in the year.
The divergence comes against a backdrop of surging AI investment. Combined capital expenditure from Meta, Microsoft, Amazon and Alphabet is on track to reach $725 billion this year, based on company disclosures.
