Micron Predicts DRAM Crisis Until At Least 2028

Micron said memory and storage supply shortages will take considerable time to improve, with industry supply expected to recover only gradually in 2028 and no clear line of sight to demand catch-up. The company is also expanding DRAM capacity via a new New York megafab due by early 2029 and a planned $1.8 billion PSMC fab acquisition that could add meaningful output by late 2027. The article implies continued elevated DRAM pricing, which could pressure consumer hardware, including upcoming consoles and other PC components.

Analysis

The market is still underestimating how sticky memory inflation becomes once hyperscaler demand captures allocation. This is not a one-quarter pricing spike; it is a multi-year capex re-rating where supply additions themselves become lagged by node transitions, tooling, and fab ramp risk, so margin expansion at the large memory vendors can persist even if unit growth slows. The key second-order effect is that buyers will increasingly substitute toward system design changes, not volume buys, which pushes pricing power from component vendors into whoever controls platform architectures and procurement scale.

The biggest losers are downstream OEMs and consumer hardware brands with fixed launch calendars and low ability to re-spec mid-cycle. Gaming consoles, PCs, and edge devices face a double hit: higher BOMs and weaker promo elasticity, which can delay refresh cycles and compress attach rates for peripherals and services. That creates a hidden beneficiary set in memory-adjacent equipment and materials, because every announced capacity project implies a longer tail of lithography, packaging, and substrate demand well before meaningful DRAM output arrives.

Consensus seems too linear on “shortage eventually normalizes.” The more plausible path is a staircase: brief easing windows get absorbed by AI buyers, while consumer demand remains price-sensitive and strategically deferred. If AI capex sentiment rolls over, memory is the first place where sentiment can reverse violently because inventories are thin and expectations are crowded long; but absent a sharp AI demand shock, the upside asymmetry remains with suppliers through at least 2027-2028.

Key Decisions for Investors

• Long MU shares vs short a consumer-exposed PC/OEM basket (HPQ, DELL, LEN) for 6-12 months: best expression of sustained DRAM pricing power and downstream margin compression.
• Buy MU Jan-2028 call spreads on pullbacks: favorable if shortages persist into the next supply ramp, with limited downside versus outright equity given cyclical headline risk.
• Pair long AMD or NVDA against short consumer hardware assemblers if you want AI capex exposure without directly paying peak memory costs; the hyperscaler channel can pass through more BOM inflation than retail device makers.
• Take profit discipline on memory suppliers only if spot pricing starts rolling over for multiple quarters; until then, the risk/reward remains skewed to the upside because supply response is too slow to matter on a 2-3 quarter horizon.
• Watch for any sign of AI capex deceleration or hyperscaler inventory draws; that is the main catalyst that could break the trade and would likely hit MU first, then spill into the broader semiconductor complex.