Kioxia Ends Planar NAND: What the September 30 Order Deadline and 2028 Final Shipments Mean

Kioxia notified customers on March 31, 2026 that it would discontinue a range of older NAND products. Multiple reports reproducing the notice put the last-time order and final purchase-forecast deadline at September 30, 2026, while final shipments can continue through December 31, 2028. The approaching September deadline therefore closes the normal ordering window; it does not mean every affected part stops shipping that day.

That long gap matters for lifecycle planning. A component can be commercially end-of-life while manufacturers continue fulfilling committed orders for years. For embedded, industrial, automotive and other long-lived designs, the practical problem is often qualification and continuity: replacing a flash device can require electrical, firmware, package and validation work even when newer NAND technology already exists.

The affected legacy portfolio includes floating-gate NAND built on 32 nm, 24 nm and 15 nm processes as well as third-generation BiCS FLASH products. Reported product forms span wafers and packaged products including BGA, TSOP, eMMC, UFS and standard SD products, so this is broader than the retirement of one consumer SSD or one package type.

The important architectural boundary is that planar, or 2D, NAND and BiCS FLASH generation 3 are not the same technology. Planar NAND lays the memory-cell structure out in the older two-dimensional approach; BiCS is Kioxia’s vertically stacked 3D NAND family. Including BiCS generation 3 in the same lifecycle notice means Kioxia is retiring both legacy planar products and an early generation of its 3D portfolio.

The notice spans products using SLC, MLC and TLC storage. Those labels describe how many bits are stored in a memory cell; they do not by themselves identify whether the physical NAND architecture is planar or three-dimensional. That distinction prevents a misleading conclusion from the phase-out: Kioxia is not announcing that every SLC, MLC or TLC NAND product everywhere is ending.

The specific older process and generation combinations are what matter. A lifecycle decision for 32 nm SLC floating-gate NAND, for example, does not establish an industry-wide end date for SLC as a cell type. Likewise, TLC remains widely used in modern 3D NAND even though some 15 nm planar TLC products are part of this legacy transition.

Consumer storage moves quickly toward denser NAND generations, but component lifecycles do not move at one speed. Long-lived systems may value a qualified part, package, controller interface and known firmware behavior more than the density advantages of a newer flash generation. That creates demand for mature components after they have stopped being technologically current for mainstream SSDs.

This is why a last-time-buy period and a multi-year shipment runway are meaningful. Customers can secure requirements and manage redesigns while the supplier winds down the older platform. The notice does not disclose Kioxia’s customer mix, shipment volumes or the amount of manufacturing capacity involved, so those quantities should not be inferred from the length of the transition.

The legacy exit is happening while Kioxia advances newer 3D NAND. In July 2026 the company announced sample shipments of ninth-generation BiCS FLASH 1 Tb TLC devices, and separately began sampling tenth-generation 1 Tb TLC devices planned for production at the Kitakami Plant Fab2. Those announcements show the portfolio moving toward newer 3D NAND while older floating-gate and early BiCS products wind down.

Kioxia publishes performance, interface-speed, capacity and efficiency claims for those newer generations. Those are vendor product claims, not independent measurements and not evidence that a particular legacy customer can substitute a newer die without redesign. A newer NAND generation can differ in interface, package, controller requirements, qualification and intended market even when both products are called NAND flash.

A supplier retiring old NAND can affect customers that still depend on those exact parts, but this notice alone does not establish a shortage of mainstream SSD NAND. It also does not quantify how much wafer capacity will be freed, where that capacity will go, or whether consumer SSD prices will rise or fall. Those questions depend on broader production, inventory, demand and contract conditions that are outside the lifecycle notice.

For the same reason, attributing the phase-out directly to AI demand would go beyond the confirmed facts. Kioxia is publicly investing in newer NAND for data-center and AI-related storage, but the March lifecycle action is sufficient on its own to explain the concrete event: older product families have a defined final-order window and a long final-shipment runway.

Confirmed by the reported Kioxia customer notice are the retirement of specified 32 nm, 24 nm and 15 nm floating-gate products and BiCS FLASH generation 3 products, a September 30, 2026 last-time-order deadline, and final shipments extending to December 31, 2028. An April 2026 ATP product-change notice independently cites Kioxia’s September 30 deadline for affected 15 nm MLC and 24/32 nm SLC NAND and says Kioxia will terminate shipments after 2028.

Not established by that evidence are a universal end date for all planar NAND made by every supplier, the end of SLC or MLC as cell types, a consumer SSD shortage, a particular SSD price direction, or a measured reliability/performance advantage from the transition. The durable takeaway is narrower: Kioxia’s legacy NAND lifecycle is entering its final ordering stage while fulfillment of already planned demand can continue for more than two years afterward.