Memory chip index plunges 8% as AI profit doubts spark tech rout

A broad sell-off swept US technology stocks Thursday, with the memory chip and hardware supply chain index plunging 8.08% to 190.86.

The memory chip and hardware supply chain index tumbled 8.08% to 190.86 on Thursday, after Apollo Global Management published analysis showing AI profits have failed to materialize outside the technology sector.

"The entire premise for the huge amount of R&D being poured into AI infrastructure is the expectation that it will reduce delays and errors across industries, but outside tech, those gains have yet to show meaningful movement," said Torsten Slok, chief economist at Apollo Global Management.

The sell-off was broad-based across tech. The Information Technology index fell 5.16% to 378.19, while the AI winners index dropped 5.24% to 379.58. Among hyperscalers, Oracle led declines with a 6.25% loss and Alphabet fell 4.44%, dragging the hyperscaler index down 2.73% to 93.47. The Magnificent 7 index slipped 2.71% to 220.42, and the mega-cap tech index lost 2.02% to 437.51. The AI software pioneers index was the sole relative outperformer, closing down just 0.14% at 110.71 after recovering from earlier intraday losses.

The rout shows a growing disconnect between AI infrastructure spending and commercial returns. Apollo's analysis found that implementation costs for B2B clients run up to 300% more than replacing human workers, pushing the timeline for tangible AI profit gains from months to as long as five years. With Amazon, Microsoft, Meta, and Alphabet committing nearly $750 billion in combined 2026 AI capital expenditures, any delay in enterprise adoption could trigger further selling in AI-heavy exchange-traded funds.

The memory chip index's gap-down open and sustained decline through the session, with selling accelerating after 22:10 Beijing time and stabilizing only after 01:30, pointed to institutional rotation out of semiconductor and AI-exposed names. The move erased gains from earlier in the week, when the S&P 500 had closed at 7,575 and the Nasdaq at 26,281.

Apollo's analysis identified several obstacles to AI adoption outside tech. Profit margins in industries such as banking, healthcare, and manufacturing have shown virtually no increase from AI, according to Slok, while integration with existing network systems and regulatory compliance requirements have proved more costly than expected. The analysis also flagged a market timing divergence: many valuations were calculated based on five-month to one-year implementation timelines, but the window may be closer to five years.

The Roundhill Magnificent Seven ETF, which holds only the seven mega-cap tech stocks, faces particular exposure if AI profit timelines stretch, according to Apollo. The iShares Expanded Tech-Software Sector ETF, which also holds Palantir and Oracle alongside cybersecurity names, could see divergent performance as a slowdown in AI enterprise use may benefit cybersecurity companies by allowing protocols to catch up with new network developments.

This article is for informational purposes only and does not constitute investment advice.