Micron Is Hot Again: Here's Why The Rally Has Legs

Micron’s Q3 FY26 revenue surged to $41.5B (from $11.3B in Q4 FY25) alongside a sharp margin reset, with non-GAAP gross margin rising to 84.9% (up from the prior year’s ~40% level) driven by HBM pricing. Management guided Q4 FY26 to $50B revenue and $31 non-GAAP EPS, keeping the stock around ~6x forward earnings, while 16 Strategic Customer Agreements cover ~$100B of minimum committed revenue with ~$22B in cash/LC backing. Offsetting risks include heavy insider selling (e.g., CEO 122 sell transactions) and potential competitive supply from SK Hynix’s planned $38B fab expansion; Citi also cut its price target to $1,150 from $1,400.

Analysis

The main implication is not that memory is suddenly “software-like,” but that the margin pool in AI is shifting upstream. If HBM stays allocation-constrained, the economic rent sits with memory suppliers first, then gets partially passed through to accelerator vendors and cloud buyers via higher server BOMs and delayed shipment volumes. That makes MU the cleanest expression of the AI buildout, while NVDA/AMD are only indirect beneficiaries unless they can fully pass through memory cost inflation.

The second-order risk is supply response. Once a segment earns monopoly-like margins, capital follows fast; the real bear case is not today’s demand, but 6-18 months of new capacity normalizing pricing and compressing the multiple before the market expects it. Insider distribution matters here because it suggests management sees the current slope of the cycle as unusually steep, even if the floor has structurally improved versus prior downcycles.

Near term, the stock is still a momentum name with a rich catalyst path, but the setup is less attractive to chase after a parabolic move. The best falsifier is not a vague “AI slowdown,” but evidence that HBM pricing and qualification economics have peaked: softer gross margin guide, stalled strategic customer expansion, or signs that competitor capacity is arriving faster than demand. If those metrics hold, the valuation can remain unusually high; if they roll, the stock can de-rate quickly despite strong absolute earnings.

Key Decisions for Investors

• Long MU on 5-10% post-rally pullbacks, with a 3-6 month horizon; keep a tight stop if gross margin guidance falls below the low-80s or if strategic customer coverage stops expanding.
• Use MU as the preferred AI-memory expression versus a broad semi basket: long MU / short SOXX or SMH to isolate HBM scarcity from general semiconductor beta; target 2:1 risk/reward if memory remains the tightest node.
• Do not chase NVDA or AMD here on the memory-scarcity thesis alone; treat them as latency-sensitive beneficiaries, but only add if they prove pass-through power in upcoming margin commentary.
• Set an alert on SKHYV capacity announcements and HBM4/HBM4E qualification milestones; if competitor capacity ramps ahead of demand, reduce MU exposure because the cycle can normalize faster than consensus expects.
• If MU gaps higher on earnings, consider trimming into strength rather than adding at once; the move is already extended, and the risk/reward improves only if guidance re-accelerates rather than simply reiterates consensus.