Markets are currently exhibiting trends reminiscent of the dot-com bubble, as noted by analysts from JPMorgan. Jason Hunter, a prominent analyst at the bank, highlighted a significant divergence in the artificial intelligence (AI) trade. This notable split involves hardware stocks, which have been performing exceptionally well, in contrast to some of the largest AI spenders facing scrutiny from investors.
In 2026, chip stocks have led the market surge, with the Philadelphia Semiconductor Index reporting an impressive 87% increase this year, marking its best quarter ever. This remarkable rise reflects the intensifying focus on the development of AI technology. Companies associated with AI hardware are also witnessing considerable success. The Roundhill Memory ETF, which focuses on memory stocks, has seen its value skyrocket by 141% since its launch in April.
However, the situation is starkly different for those companies heavily investing in AI. The so-called “Magnificent Seven” stocks, which enjoyed considerable market enthusiasm related to AI in 2025, have recently faced downturns. The Roundhill Magnificent Seven ETF has fallen 7% from its peak this year. Notably, major spenders like Meta and Microsoft have struggled significantly, experiencing respective year-to-date declines of 5% and 18%. Microsoft, in particular, recorded its worst monthly loss since 2000 at the end of June.
Hunter pointed out that this market divergence draws parallels to trends observed in 1999, when suppliers of communications equipment experienced rapid increases while firms making substantial capital investments in the sector plummeted. He cautioned that the current dynamic, characterized by the disparity between hardware providers and AI capital investors, bears similarities to the pre-burst conditions of the dot-com bubble.
Addressing the growing concerns regarding massive investments in AI, Hunter referenced the projected capital expenditure (capex) from major players such as Meta, Microsoft, Amazon, and Alphabet. This amount is projected to reach $725 billion in 2023, which, if trends continue, could surpass the gross domestic product of major economies, like Japan, by the end of the decade.
Hunter concluded his analysis by emphasizing the need to closely observe the performance of individual hyperscaler stocks through the summer, as any stabilization could mitigate the risk of a market downturn driven by shifting sentiments and positions in the fall.
