Got $1,000? 2 Magnificent Artificial Intelligence (AI) Stocks Down Over 15% to Buy Hand Over Fist

Micron (MU) is framed as a “buy on the dip” as memory tightness driven by AI data-center demand is expected to persist beyond 2027, with FY2027 estimates calling for ~81% revenue growth and EPS near ~$150 (vs. ~$73.32 for FY2026). Broadcom (AVGO) is pitched as a major AI semiconductor growth winner, targeting >$100B revenue in its AI semiconductor business in 2027 (up from $10.8B in the most recent quarter) and projecting ~67% revenue growth this year and 62% next year. Overall, the article is bullish on both names’ tailwinds through end-2027, implying limited near-term upside risk if the current tightness/growth backdrop remains intact.

Analysis

This is less a fresh fundamental inflection than a reminder that the AI capex cycle is still translating into component scarcity. The cleaner expression is in memory: if DRAM/NAND stays tight, MU should continue to get operating leverage from pricing rather than unit growth, which usually sustains EPS revision momentum for 2-4 quarters after the market starts to fade the story. The risk is that semiconductor cycles tend to break when everyone starts to model shortage durations into out-years; once supply response arrives, the multiple can compress before the earnings peak is visible.

For AVGO, the key mechanism is not generic AI exposure but substitution: custom accelerators can displace some GPU demand at hyperscalers, but only where workloads are stable enough to justify bespoke silicon. That should support mix and gross margin if customer concentration remains manageable, yet it also makes the stock dependent on a small set of cloud budgets and design wins. The market likely underprices how lumpy these programs can be if one or two hyperscalers pause capex for a quarter.

Contrarianly, the consensus is treating both names as duration assets rather than cyclical semis with a strong narrative overlay. The missing data are order-book conversion, memory ASP elasticity, and whether hyperscaler capex is shifting from expansion to efficiency. If those metrics soften, the move can reverse quickly over days for sentiment, but the real risk is over 6-12 months when supply catches up and margin assumptions reset.

Key Decisions for Investors

• Long MU on any 5-8% pullback; target 3-6 month hold into the next guide cycle. Favor size only if memory ASPs remain firm and inventory days stay below historical cycle peaks; falsify if management guides to easing tightness or customer inventories rebuild.
• Buy AVGO versus short SMH for a 2-4 month relative-value trade. Thesis: custom AI silicon should hold margins better than the broader semi basket if hyperscaler spend stays intact; exit if AVGO underperforms SMH by ~7-10% or if AI revenue commentary turns less linear.
• Use MU call spreads instead of outright equity if you want exposure to the shortage thesis but want to cap cycle risk. The cleanest structure is 6-9 months out, financed by selling upside where the market starts to price a perfect 2027; cut if memory pricing rolls over two consecutive monthly checks.
• Avoid chasing NVDA/AMD here on the basis of this note alone. The article actually strengthens the case that a slice of AI value creation is migrating toward custom silicon and memory, not necessarily toward the most crowded GPU beta.
• Watch for any hyperscaler capex pause or Broadcom customer commentary in the next earnings season; that is the most likely near-term falsifier for the entire AI semiconductor complex.