AI Memory Shortage Sustains Capex Cycle; Chip Stocks Trade at Discount

The past month has crystallized a structural constraint that the market is only beginning to price in: artificial intelligence deployment is hitting memory bottlenecks faster than supply can respond. On their earnings calls, the CEOs of Microsoft, Meta, Google, Amazon, and Apple all articulated the same theme with striking consistency: memory availability is the binding constraint, and it will not resolve in the near term. This is not a temporary inventory glitch; it reflects deepening capex intensity across the entire AI stack.

The narrative matters because it directly contradicts the "capex peak" thesis that has periodically circulated among skeptics. If five of the world's largest technology firms are unified in disclosing memory pressure on the same issue, the market should be bracing for sustained, elevated infrastructure spending. Goldman Sachs and other analysts have pivoted their forecasts accordingly, extending expected capex cycles into 2026 and beyond. This keeps data-center equipment makers, semiconductor suppliers, and chipmakers in a structural demand environment.

Yet pricing does not reflect this reality. Memory suppliers, particularly Micron (MU), remain valued at depressed multiples despite disclosed demand backlogs. Institutions have rotated into mega-cap equities on earnings relief, pushing SPY and QQQ higher, but the semiconductor sector and AI infrastructure supply chain remain underweight in many institutional allocations. The disconnect suggests either the market discounts the durationBond price sensitivity to interest rate changes. of memory constraints or has priced in competitive oversupply.

Skeptics counter that capex cycles always appear infinite until they abruptly terminate. Management's public disclosure of capacity constraints could simply reflect negotiating posture with suppliers or a need to set conservative guidanceCompany-issued forecasts of future financial performance.. Additionally, capacity expansion from rivals may accelerate faster than current disclosures suggest, flooding the market and causing multiples to compress even if volumes remain elevated.