Intel and Micron to end NAND flash partnership after the introduction of 3rd gen 3D NAND in late 2018/early 2019, will continue 3D XPoint memory collaboration
LAS VEGAS, NV — The long-running partnership between Intel and Micron for the development and manufacturing of NAND flash memory will soon come to an end.
Context & Ripple Effects
The Intel–Micron memory alliance has been the vehicle for both companies' biggest storage bets: the 3D NAND program announced in 2015 to triple SSD capacity and the 3D XPoint chip unveiled mid-2015 as a flash alternative already in production. This announcement splits those two tracks — joint NAND development and manufacturing ends once 3rd-generation 3D NAND ships in late 2018/early 2019, while the XPoint collaboration continues.
The split lands just as the partners are still shipping jointly built product: their QLC NAND, launched in May 2018, trades endurance for lower cost per bit. What follows in the corpus makes the split look like the first domino — both companies eventually abandoned the very technology they chose to keep collaborating on.
First-order effects
- After 3rd-gen 3D NAND, Intel loses its co-development and co-manufacturing stake in NAND and must source or build its flash roadmap separately, while Micron gains sole control of the shared NAND development and manufacturing line.
- Both companies keep 3D XPoint as a going joint effort, leaving Lehi-era shared R&D as the surviving piece of the alliance.
Second-order effects
- With separate NAND roadmaps, Intel and Micron shift from partners to direct competitors in the same flash market they jointly supplied, competing on cost-per-bit curves like the QLC trade-off they co-launched.
- The retained XPoint collaboration now carries the full strategic weight of the relationship — and when it underdelivers commercially, each company has an independent exit path rather than a binding joint commitment.
Third-order effects
- That exit path gets used: Micron ceases all 3D XPoint R&D in 2021 and lists the Lehi fab for sale, and Intel winds down Optane in 2022 with a $559M inventory impairment — the sixth non-core business CEO Pat Gelsinger has shed — showing the split enabled each side to unwind the specialty-memory bet on its own schedule.
- The pattern points to memory alliances as transitional vehicles: partners pool capital to cross a process transition like 3D NAND, then decouple once each can fund its own roadmap — and successor collaborations inherit none of the original's durability.
The trend: Semiconductor memory partnerships formed to share the cost of a process transition tend to dissolve at the transition's end, leaving each partner free to pursue — or abandon — the follow-on technologies independently.