AI Memory Shortage Drives Memory Chip Demand, MU Trades at 7x Earnings

The memory constraint narrative has crystallized across the industry. Within two days last month, the CEOs of Microsoft, Meta, Alphabet, Amazon, and Apple all communicated the same critical message on earnings calls: memory is constrained and this bottleneck will not resolve quickly. Yet despite this widespread industry acknowledgment and persistent demand signals, Micron trades at just 7x earnings, suggesting the market has yet to price in the structural undersupply driving capex cycles across data centers and AI infrastructure.

This creates a curious disconnect between narrative and valuation. If memory constraints are real, durable, and acknowledged by every major consumer of semiconductor products, traditional valuation multiples should reflect scarcity premiums. Instead, Micron and other memory players appear overlooked relative to the optical intensity of the AI capex cycle. Broadcom, which supplies advanced packaging and substrates critical to AI chips, and other semiconductor suppliers similarly benefit from this elongated bottleneck cycle, yet many trade with muted enthusiasm.

The implications ripple across multiple layers. Data center operators face margin pressure from elevated memory costs. GPU makers like Nvidia gain pricing power when memory cannot be economically substituted. Chip equipment suppliers benefit from elevated capex intensity needed to unlock new memory nodes. Conversely, memory constraints risk prolonging the timeline for AI deployment in price-sensitive verticals like consumer and automotive applications.

Sceptics argue that constraints often prove transitory; once capex cycles ramp, supply typically overshoots, leading to price compression and margin collapse. The semiconductor industry's history of boom-bust cycles raises questions about whether current multiples will expand in time, or whether the market is pricing in a near-term correction.