Warning: The Stock Market Is Doing Something Last Seen Before the Dot-Com Crash -- and Investors Should Pay Attention

JPMorgan warns of a “growing divergence” between AI hardware stocks and AI infrastructure spenders after the pattern was last seen before the dot-com crash; Micron is up ~250% YTD while Amazon and Microsoft are down on the year in 2026. Citi’s Bear Market Checklist shows 10 global bear-market warning flags and 11.5 U.S. flags—highest since the 2008 crash, with valuation metrics like the S&P 500 Shiller CAPE at its highest since early 2000 and the Buffett indicator at 233.8%. Despite the caution, both firms say investors shouldn’t panic, but Citi notes conditions can deteriorate quickly once flags reach double digits.

Analysis

This is less a crash setup than a regime-risk setup: the market is rewarding the upstream beneficiaries of AI capex faster than the cash-generative platforms that fund it. That usually persists until management teams start trimming build plans or investors demand higher free-cash-flow yields, so the first pain point is more likely relative multiple compression in AMZN/MSFT than an instant collapse in semis. Second-order, the real winners are the picks-and-shovels names tied to memory, networking, and datacenter buildout; the losers are the hyperscalers whose depreciation and opex burden rises before revenue monetization catches up.

The warning matters most if it is confirmed by credit: valuation alone can stay stretched for months, but widening high-yield spreads and weaker bank lending standards would turn this from a sentiment issue into a funding issue. That is the key falsifier for the bullish hardware trade: if spreads stay calm and cloud capex guidance re-accelerates, the crowding can persist and the "bubble" call remains premature. Time horizon matters: days = headline volatility, 1-3 months = factor rotation, 6-18 months = whether AI returns on invested capital are actually high enough to justify the buildout.

Contrarian view: the consensus is treating high CAPE as a timing tool, when it is often just a low-quality signal for broad-market returns. In the near term, the better expression is relative value, not outright shorting the index. If the market is wrong, it will likely be because earnings revisions and liquidity keep overpowering valuation warnings longer than bears expect; if the market is right, the first confirmation will show up in capex discipline, not in a single scary checklist reading.

Key Decisions for Investors

• Long MU / short AMZN or MSFT on strength over the next 2-6 weeks; use the pair to isolate the AI capex spend-vs-supplier spread. Target 15-20% relative outperformance; exit if hyperscaler capex guidance re-accelerates or MU loses relative strength for two straight weeks.
• Avoid outright shorting QQQ/SPY solely on valuation warnings; wait for credit confirmation. Add a bearish index hedge only if HY spreads widen meaningfully and breadth deteriorates, because valuation-only squeezes can persist for months.
• If already long mega-cap AI spenders, trim 25-33% into rallies and rotate into semis/hardware exposure. The risk-reward improves when you own the earnings visibility instead of the capex burden.
• Watch JPM/C research signal as a market-health indicator, not a trade catalyst. Falsify the cautionary thesis if the 10Y yield falls and credit spreads remain range-bound while AI-related guidance stays firm.
• For tactical hedging, consider short-dated QQQ puts only on an index breakout failure after a strong run, not immediately on the headline. The edge is in timing the first liquidity crack, not in preemptively paying up for protection.