Kioxia Killing Off Old NAND Chippery

Why It Matters

The move accelerates the industry’s consolidation around high‑layer NAND, tightening supply for older formats while driving down cost‑per‑bit for data‑center and consumer storage markets.

Key Takeaways

• Kioxia ends 64‑layer NAND sales by Sep 2026.
• Shipments of discontinued chips stop end 2028.
• Focus shifts to 162‑218‑layer BiCS NAND.
• Competitors Micron, SK Hynix, Samsung already phased out planar NAND.
• Higher‑layer wafers boost cost‑per‑bit efficiency.

Pulse Analysis

The NAND flash market is entering its third decade of rapid scaling, and Kioxia’s recent end‑of‑life announcement underscores how manufacturers are pruning legacy product families to free up silicon real estate. By terminating sales of 64‑layer and older planar designs, Kioxia aligns its roadmap with the industry’s push toward denser 3D architectures. This shift not only simplifies its product portfolio but also reduces the complexity of maintaining multiple process nodes, a strategic move that mirrors actions taken by Micron, SK Hynix and Samsung over the past two years.

From a financial perspective, higher‑layer wafers deliver dramatically lower cost‑per‑bit. A single 218‑layer BiCS wafer can store millions more bits than a 64‑layer counterpart, translating into higher margins and better utilization of expensive fab equipment. The reallocation of capacity also allows Kioxia to respond more flexibly to the surging demand for TLC and QLC storage in cloud and AI workloads, where density and price efficiency are paramount. As older designs fade, the company can invest in process improvements that further shrink cell dimensions and improve endurance, reinforcing its competitive position.

For OEMs and system integrators, the phased exit means revisiting supply contracts and potentially redesigning products that still rely on legacy NAND. However, the extended runway—sales ending in 2026 and shipments in 2028—provides a buffer to qualify newer high‑layer parts. In the longer term, the market will likely see a consolidation of inventory around 162‑layer and above chips, with QLC becoming the default for cost‑sensitive applications while SLC and MLC migrate to niche, performance‑critical segments. Kioxia’s timing, though later than peers, signals that the transition to ultra‑dense NAND is now an industry imperative.