Micron Just Did Something Even Nvidia Hasn't Done. Is the AI Stock a Buy?

Micron reported revenue up 196% to $23.9B (vs $19.2B consensus) and adjusted EPS $12.20 (vs $8.65 consensus), with gross margin surging to 74.4% and net income $13.8B ($12.07/sh). Operating margin reached 67.6% (above Nvidia's 65%), and management guided fiscal Q3 revenue ~$33.5B (+260%), gross margin ~81%, and adj EPS $18.75–$19.55 (vs $1.91 a year ago). Company plans >$25B capex including a Taiwan facility acquisition and expects DRAM/NAND tightness beyond 2026 with ~20% bit shipment growth. Despite blockbuster results and aggressive guidance, the stock fell ~3% after hours, reflecting investor skepticism about sustainability in a cyclical memory market.

Analysis

Micron’s outperformance is reshaping the memory value chain: suppliers of advanced lithography, etch and wafer services and Taiwan-based contractors gain pricing power as high-margin AI-specific DRAM/NAND skews revenue toward fewer, more specialized fabs. Hyperscalers and AI OEMs become de facto demand anchors — their procurement cadence and inventory strategies will now set price floors and volatility bands for the sector rather than consumer PC refresh cycles. Expect increased bargaining leverage for memory sellers when product mix tilts to HBM and specialized DDR variants used in accelerators.

Primary risks are structural (multi-quarter capacity expansion from incumbents), cyclical (inventory-led price corrections), and geopolitical (concentration of advanced packaging/foundry services in Taiwan). The most realistic timing for a meaningful oversupply shock is 12–36 months if competitors accelerate capex or if cloud orders normalize after a front-loaded data-center refresh. Near-term catalysts to monitor are equipment order flow, wafer-starts metrics, and hyperscaler multi-quarter purchase commitments — each will precede price movement by 1–2 quarters.

The consensus sell-off signals fear of margin mean reversion, but that’s a blunt read. If AI-optimized memory becomes a structurally distinct product category with higher ASP sustainability (driven by performance/latency rather than commodity bits), current multiples underweight recurring margin premium. That said, position sizing must assume a non-trivial tail where margins revert toward historical cycle midpoints; hedging around 12–24 month horizons is prudent.

Key Decisions for Investors

• Core position: Buy MU 6–18 month equity exposure (size 3–5% of tech allocation). Hedge with a 9–12 month MU put spread (buy 1 OTM put / sell a lower OTM put) to cap downside to ~30–40% while retaining >2x upside if AI-driven ASPs hold; exit or re-evaluate on sustained weakness in hyperscaler order flow over two consecutive quarters.
• Relative-value pair: Long MU / Short INTC equal notional for a 3–12 month trade. Rationale: MU captures memory-led revenue re-rating while INTC remains exposed to CPU/accelerator margin pressure; target asymmetric payoff of 20–50% if AI capex stays elevated. Cut losses if MU falls >25% on confirmed broad-based memory price declines or if INTC announces a major new product win with hyperscaler purchase commitments.
• Insurance sleeve: Buy 3–6 month NVDA puts sized at 20–30% of MU position notional as tail-protection against a rapid AI sentiment unwind. Cost is an insurance drag but preserves ability to hold MU through a transient derating; unwind puts if NVDA implied vol compresses >40% from post-earnings levels or if NVDA issues multi-quarter guidance that materially increases addressable AI spend.