Morningstar Warns Of 30% Crash In AI Stocks, Is It Time To Sell Micron?

Morningstar warns AI/memory stocks may face a “reckoning,” with large parts of the group potentially correcting 20%-30% before becoming buyable again. The call targets memory-name upside driven by an AI-capex boom—while Morningstar expects supply to catch up with demand as Samsung and SK hynix add capacity and memory pricing cools, and capex growth peaks in 2026 with slower increases through 2029. Despite Micron’s strong recent fundamentals (Q3 FY26 revenue $41.46B, +345.7% YoY; Q4 revenue guidance $50.0B), the article notes Micron’s equity has already surged (+304.6% YTD, +838.8% YoY) and stresses investors should monitor HBM4 pricing and hyperscaler capex into 2027.

Analysis

This looks like a classic late-cycle memory trade: the market is extrapolating peak pricing, peak margin, and peak capex duration at the same time. The most vulnerable names are the ones with the cleanest beta to memory ASPs and the least buffer from recurring software-like revenue; that puts MU, SNDK, WDC and the equipment stack LRCX in the crosshairs. The second-order effect is that a slowdown in one part of the AI supply chain can quickly become a multiple problem for the whole "AI infrastructure" complex as investors stop paying growth premiums for capital-intensive cyclicals.

The timing matters. A 20% to 30% drawdown can happen in days to weeks on any evidence that HBM pricing is softening, but the more durable earnings reset likely unfolds over 1 to 3 quarters as new capacity comes online and hyperscaler ordering gets normalized. The contrarian miss in consensus is that forward P/Es are being treated as if current earnings power is sticky; in memory, that is usually when the denominator rolls over hardest. The clearest falsifier is continued tightness in HBM4 and another round of upward capex revisions from the big cloud customers.

TSM is the cleaner relative winner because it captures AI spend without the same direct exposure to commodity-like pricing. If the market shifts from "AI is scarce" to "AI is oversupplied," capital should rotate toward the highest-quality manufacturers and away from the most crowded end of the boom. NVDA is less vulnerable on fundamentals, but it is still exposed to sentiment spillover if the market decides the entire AI capex trade needs a reset.

Key Decisions for Investors

• Short MU via 1-2 month put spreads into strength; thesis is a 20%+ re-rating lower if memory pricing and guidance both normalize. Cover if HBM pricing stays firm or if management raises FY margin assumptions again.
• Pair trade: long TSM / short LRCX for 1-3 months. This isolates the likely winner of AI capex quality while expressing downside to capex-sensitive tooling if spending tapers after 2026.
• Use SOXX or SMH puts rather than single-name shorts if liquidity is preferred. Best risk/reward is a 5-10% index pullback hedge against a sector-wide de-rating rather than trying to pick the exact weakest memory name.
• Avoid chasing SNDK and WDC after the vertical move; they are the highest beta expressions of the same cycle and would likely underperform first if the group unwinds.
• Set a watch alert on Samsung/SK Hynix supply adds, HBM4 pricing, and hyperscaler capex guides for 2027. If any of those stay hot, the short thesis becomes a timing trade only, not a structural one.