A two-speed market: AI-grade ultra-flat 300mm for 2/3nm is tightening even as legacy 200mm softens — so headline 'wafer demand' masks a real squeeze on exactly the wafers AI needs, sitting on a five-firm oligopoly with multi-quarter qualification moats.
Tightness
+3 this quarter — Aggregate wafer supply looks adequate — even soft at legacy diameters — but the AI-grade 300mm segment is tightening and concentrated (top-5 ~80%). High qualification barriers and capacity being repointed toward 2/3nm keep the relevant slice tight without being acutely sold out.
Wafers are the literal substrate of every chip. The constraint isn't total volume — it's the ultra-flat, defect-controlled AI-grade 300mm wafers that 2nm and 3nm logic and advanced memory require. That segment is tightening while legacy 200mm and smaller go soft, and qualifying a new wafer source at a leading fab takes multiple quarters, so supply can't pivot quickly to the wafers in demand.
Legacy 300mm capacity being repurposed to AI-grade lines, new greenfield 300mm coming online (e.g., GlobalWafers Texas), or a slower foundry/memory ramp easing demand for ultra-flat wafers.
Get the squeeze before the headlines.
A weekly read on the AI buildout's tightest constraints. Free, sourced.