Data center capex surge bolsters Micron's AI-memory narrative, but peak-cycle valuation and structural risks keep us on the sidelines.

What happened

A new Motley Fool article highlights accelerating data center capital expenditure as a catalyst for Micron Technology, reinforcing the bullish AI-memory scarcity narrative that has driven the stock up over 800% from its 2025 low. This narrative aligns with Micron's own fiscal second quarter results, which showed record revenue of $23.9 billion and a 74% gross margin, buoyed by tight HBM and conventional memory pricing. However, our analysis of Micron's 10-K and 10-Q reveals that the company's substantially short-duration customer contracts leave it exposed to rapid pricing reversals, and management explicitly warns that weakening HBM demand could trigger a shift back to conventional DRAM, creating oversupply and margin compression. Moreover, with the stock trading at 42x trailing earnings and 55x EV/EBITDA, the market already prices a multi-year continuation of peak conditions that history suggests are unsustainable in memory cycles. The article's positive catalyst is not new information; it is embedded in consensus expectations, and we would reiterate a WAIT rating until either a better entry price emerges or evidence of durable pricing beyond allocation language becomes clearer.

Implication

The reported acceleration in data center spending supports the near-term demand thesis for HBM and server DRAM, which is already reflected in Micron's record margins and elevated stock price. However, the key risk is that this demand is not locked in through long-term contracts; Micron's 'substantially all' customer agreements are short-term with periodic renegotiations, meaning pricing can reset quickly if supply dynamics shift. The company's own risk factors detail a direct pathway to oversupply: if HBM demand normalizes, suppliers reallocate capacity to conventional DRAM, flooding the market. With FY2026 capex guided above $25 billion and FY2027 capex expected to step up further, free cash flow could deteriorate rapidly if pricing rolls over, leading to multiple compression. We would wait for a pullback toward our attractive entry of $780 or for evidence that pricing benchmarks remain robust beyond the next two quarters before considering an entry.

Thesis delta

The article reinforces the existing bullish narrative around data center capex and memory tightness, but does not change our assessment that current valuations already discount peak-cycle conditions. The key risks remain: short-duration contracts, rising capex, and the potential for HBM demand normalization to trigger a conventional DRAM oversupply. Therefore, we maintain our WAIT rating and see no compelling reason to adjust our stance.