Micron Technology Inc (MU.NASDAQ) AI OPINIONS & ADVISOR ANALYSIS

The bull case activates if the 2028 wafer wave is absorbed rather than dumped. Inference demand and edge-device memory content grow into the new capacity while five-year rolling supply agreements with advance payments keep utilisation contracted [16][17]; margins plateau in the sixties instead of collapsing. A $60-100 billion buyback then compounds a shrinking share count against stable earnings, and investors re-rate memory from commodity to contracted industrial. That combination supports a price materially above $1,600 by 2030.

The bear case activates when contract prices roll over and capex depreciation arrives in the same fiscal year. Four suppliers adding wafers simultaneously in 2028 [15], with CXMT unconstrained by Western contract discipline, would push ASPs down while Idaho fab-2 depreciation inflates cost per bit. Free cash flow would evaporate within two quarters, as fiscal 2025's $1.67 billion demonstrated, and a peak-cycle $1.16 trillion capitalisation would halve. Buybacks executed near the top would compound the error.

The settled belief is that strategic customer agreements have abolished memory's cyclicality. Forty-nine analysts rate MU Strong Buy with a $1,513 average target [24], and the bull argument reduces to one anchor: eight times forward earnings is cheap. That anchor is the bias. Target dispersion from $361 to $2,200 [24] reveals a crowd that agrees on tightness and disagrees entirely on which regime applies.

The crowd overestimates what these shares are durably worth, despite the optically cheap multiple. The decisive overlooked evidence is the second derivative: contract-price increases decayed from roughly +92% to +60% to +15-25% QoQ across three quarters [7][8][9], and the revenue explosion is overwhelmingly price, not bits [3]. Memory equities de-rate when acceleration ends, not when earnings fall. A $1.16 trillion capitalisation therefore capitalises a run-rate that the industry's own 2028 wafer schedule is designed to end [15].

The trigger is the first flat or negative QoQ server DRAM contract print, plausible in late 2027 per TrendForce's moderating path [9], arriving alongside 2028 wafer-start disclosures. The first observable sign of repricing is the stock falling on a record earnings beat.