Analysis

The market is repricing MU less as a cyclical memory supplier and more as a constrained AI infrastructure toll road. The important second-order effect is that long-dated customer commitments reduce near-term earnings volatility and effectively de-risk capex, which should compress the equity risk premium even if the business remains capital intensive. That benefits not just MU, but also NVDA indirectly: a more secure HBM supply chain lowers the odds of AI accelerator shipments being bottlenecked by memory availability rather than GPU demand. The catch is that the current profit mix is unusually price-led, which means the stock’s multiple can stay elevated only as long as supply discipline survives. The 2027-2028 capacity wave is the real regime-change date, not the next quarter; that is when this trades from scarcity economics back toward classic memory economics. In other words, the equity is discounting several years of tightness, while the industry is already spending to break it. Consensus is probably underestimating how much of the upside has already been pulled forward by analyst target inflation and momentum flows. A move of this magnitude tends to create a self-fulfilling de-risking of the bear case in the short term, but it also sets up a high bar for any moderation in margin expansion. The best contrarian read is not that MU is expensive today, but that the market may be overconfident about the duration of pricing power into the next supply cycle.