Phison Electronics Sees Worst NAND Shortage in 2027

This is not a clean “buy semis” signal; the more interesting read-through is storage mix and substitution. If NAND tightens, the obvious winners are WDC and MU via ASP leverage, but the more durable relative beneficiary is STX because hyperscalers and enterprise buyers can substitute toward HDD for cold/nearline data, especially as AI datasets accumulate. That substitution matters more than the headline shortage: it can reprice the storage stack even if total server demand is unchanged.

The first tradable catalyst is pricing data over the next 1-3 months: enterprise SSD lead times, spot NAND quotes, and channel inventory. If those do not tighten into the next earnings season, this is likely just cycle-talking from a supplier with incentive to talk up scarcity. Over 6-18 months, the structural effect is higher non-volatile storage content per rack, but the benefit may accrue to vendors with the best mix and pass-through power rather than to flash names alone.

Contrarian view: consensus will likely focus on flash suppliers and miss the margin squeeze on low-ASP PC and consumer OEMs, where storage inflation is harder to pass through. That makes HPQ the cleaner loser than the large AI compute names, because the BOM hit is a larger share of the end product and can delay entry-level refreshes. Falsifiers are flat-to-down NAND contract pricing, rising channel inventories, or any evidence that PC demand rolls over before shortage economics show up.