This is reporting and analysis, not investment advice. Price moves are attributed to exchange data; analyst views are scenarios, not predictions.

The AI trade powered the market to records. Its reversal is now the market's biggest story.

What's happening

A cluster of large technology and AI-linked stocks have fallen 20% or more from their recent highs — the conventional definition of a bear market for a single stock. Per exchange data compiled by Yahoo Finance / 24/7 Wall St., the drawdowns are steep: Oracle down roughly 57% from its 52-week high, ServiceNow about 55%, Palantir ~47%, Salesforce ~44%, Microsoft ~34%, Meta ~31%, and chip names Arm and Broadcom each around 26%. Nvidia, near $193 against a $236 high, is down about 18% — just shy of the threshold, though it has touched it intraday. (Some individual 52-week highs are sourced to 24/7 Wall St.; treat the exact percentages as snapshot figures.)

Crucially, the broad indexes are not in a bear market: the damage is concentrated in the most richly valued, AI-exposed names rather than the whole market — a sector reassessment, not (yet) a systemic one.

Why it's happening

Several threads we've tracked converged:

  • The OpenAI IPO-delay report. A report that OpenAI may push its listing to 2027 — rather than go public below a $1 trillion valuation — removed a near-term catalyst that had underpinned AI optimism, and hit AI-linked names broadly.
  • Spending without clear payback. The market increasingly wants proof that enormous AI capex will earn a return. Oracle's stock has slid even as it touted large AI-infrastructure contracts — robust demand hasn't reassured investors worried the spending delays free cash flow for years.
  • Stretched valuations. Names that had run up severalfold left little room for disappointment; when sentiment turned, the percentage falls were correspondingly large.
  • A hawkish macro backdrop. Hot PCE inflation and a Fed holding higher-for-longer lift the discount rate on future earnings — and nothing is more sensitive to that than high-growth tech. The broad risk-off mood and equity-fund outflows have amplified the moves.

Two readings

Analysts split on what it means. The bull case: AI infrastructure demand is structurally real — Micron's record quarter (revenue up ~346%) shows end-market traction — so this is a valuation correction of an overheated trade, not a repudiation of the technology. The bear case: the sheer scale of the capex cycle creates execution and balance-sheet risk the market hadn't priced, and even after these falls, multiples remain elevated, leaving room for more de-rating. Neither side can claim proof; the swing factor is the same as for the rest of the market — the Fed, inflation, and whether AI revenue catches up to the spending.

The hardware/software divide

One useful distinction: the hardware/infrastructure layer (Nvidia, Broadcom, Arm, Micron) has had its demand verified by earnings, so the question there is whether buyers earn adequate returns. The software/platform layer (Palantir, ServiceNow, Salesforce) faces a sharper valuation question — whether AI-narrative revenue justifies the prices. The selloff hit both, for partly different reasons.

What it means

The episode is a reminder that a powerful technology trend and richly priced technology stocks are not the same thing. Declines of 25–57% from highs are real capital destruction for anyone who bought near the top, and the path back depends on some mix of recovering earnings expectations, a Fed pivot, and a return of risk appetite. We're reporting the moves and attributing the analysis — not calling a bottom. For what these terms mean, see our companion explainer on corrections, bear markets and crashes.