Micron Undervalued at 7x Earnings as AI Memory Demand Stays Constrained: MSFT, META, GOOGL

In consecutive earnings calls over a two-day period last month, the leadership teams of five mega-cap tech giants (Microsoft, Meta, Google, Amazon, and Apple) delivered the same core message: memory is constrained and that constraint is not ending soon. Each CEO underscored the ongoing crunch in AI infrastructure capacity, a bottleneck that has shaped capital allocation across the sector for over a year. Yet despite this uniform and forceful signaling from trillion-dollar operators, the market continues to price memory-chip leader Micron at 7x forward earnings, a valuation that traders are flagging as potentially disconnected from fundamentals.

The thesis is straightforward: if the biggest spenders on AI infrastructure are saying memory constraints will persist, then demand for Micron's DRAM and NAND products should remain robust. Investors are interpreting this as either a mispricing or a slow market recognition of Micron's structural advantage. The earnings commentary from Nvidia, Broadcom, and other infrastructure names has reinforced the narrative that capex cycles are not peaking soon, but rather accelerating into inference workloads and longer-durationBond price sensitivity to interest rate changes. model training. Memory is the linchpin of this infrastructure build, and the multiple mismatch suggests either the market is underweighting the durability of demand or is hedging against competitive supply additions.

This narrative crosses multiple asset classes: semiconductor stocks (particularly Micron, but also benefiting Nvidia, AMD, and Broadcom on the input-supply side), big-tech equities that consume memory (Microsoft, Meta, Google, Amazon, Apple), and broader AI infrastructure indices like the Nasdaq and semiconductor sector trackers. If Micron's valuation re-rates higher, it could accelerate a rotation from AI software plays (which face margin pressure from rising infrastructure costs) to hardware and semiconductors. Conversely, if memory supply normalizes faster than expected, the thesis breaks.