Last summer, Samsung kicked off mass production of three-dimensional NAND that layers memory cells vertically. It's not the only one working on stacked NAND structures, though. Toshiba and SanDisk are collaborating on 3D tech of their own, and the duo is building an entirely new facility to handle fabrication.
The new building will actually replace an existing one. Toshiba's Fab 2 facility in Mie prefecture, Japan, is being torn down to make way for a plant geared toward 3D NAND production. Demolition of the old fab will begin this month, and construction of the replacement is scheduled to begin in September. Toshiba expects construction to be complete by the summer of 2015, but it looks like the facility may not start producing 3D NAND until 2016. Here's the relevant snippet from the press release:
The clean room within the new fab will be built in phases to align the clean room investment with the timing of conversion of 2D NAND capacity to 3D NAND. Construction of the initial cleanroom will be complete in time for 2016 output. Decisions on capacity conversion ramp and equipment investment, the start of production, and production levels in the new fab will reflect market trends.
Although it's unclear exactly when Toshiba will start producing 3D NAND, the groundwork is being laid for what looks like an inevitable transition. SanDisk has committed to help fund the new facility as part of a "non-binding memorandum of understanding." The two firms collaborate closely on 2D NAND production, and that partnership appears to be secure for the next generation of non-volatile memory. Interestingly, SanDisk told us a couple of weeks ago that it takes roughly 50% of the NAND produced by its joint venture with Toshiba. That figure applied to 2D NAND, but I wouldn't be surprised to see a similar split with the 3D stuff.
Flash makers typically increase storage densities by using smaller process geometries. Flash cells become more difficult to produce and manage the smaller they get, though. Stacking these cells vertically is seen as a good way to keep ramping densities without transitioning to finer fabrication technologies.