SK Hynix to Mass-Produce 375-Layer NAND by Year-End, Introducing Molybdenum Material for the First Time
Alina Collins
SK Hynix plans to begin mass production of 375-layer 3D NAND by year-end, partially replacing tungsten with molybdenum for the first time; this marks NAND flash's entry into a new era where material science, not stacking height alone, determines the next leap.
Why did 400 layers become 375?
The product was internally codenamed "400-layer-class NAND," but manufacturing challenges — especially channel-hole etching (drilling an ultra-deep, ultra-narrow vertical hole through hundreds of thin-film layers) — forced the target down to 375 layers.
This means → layer count is not simply "stack as high as you want." At these heights, fabrication precision pushes back and resets the design target.
The roadmap already extends to 480 and 604 layers, confirming that 375 is a compromise with today's process capability, not a ceiling.
No new fab — how does mass production work?
SK Hynix will not build a new wafer fab. Instead, it is retooling existing lines at its Cheongju M15 plant — currently producing 176-, 238-, and 321-layer NAND — to switch to 375 layers.
In plain terms = same factory, same floor space, new process. Total capacity stays flat, but each wafer yields more storage bits at lower cost.
The 375-layer product has passed production verification and is being transferred to the mass-production line.
Why change materials — what went wrong with tungsten?
As layers increase, word lines (the signal channel in each storage layer) keep getting narrower. Tungsten's resistivity rises sharply at smaller dimensions, slowing signal transmission.
Molybdenum (Mo) offers lower resistance in fine-pitch structures, directly improving read, write, and erase speeds.
There is a hidden bonus: tungsten deposition requires an extra barrier liner on every layer, each consuming precious thickness. Molybdenum deposits directly with no liner, freeing physical space for denser stacking.
This reflects a shift in the NAND race — from "who stacks highest" to "whose materials can sustain the height." Material science is overtaking pure structural engineering as the decisive battleground.
Equipment choice: why did SK Hynix pick TEL over Lam?
Samsung led the way in April 2024, introducing molybdenum in its 9th-gen 286-layer NAND using Lam Research single-wafer deposition tools.
SK Hynix evaluated both Lam and Tokyo Electron (TEL), then chose TEL's batch-furnace deposition system — capable of processing roughly 100 wafers at once.
This means → the furnace approach is more efficient on equipment cost, floor space, and molybdenum consumption — better suited to high-volume production. The two giants have now split onto different equipment paths for the same material.
Molybdenum supply chain: who supplies, and where are the gaps?
Expected molybdenum suppliers to SK Hynix include Air Liquide (France), Entegris, and Merck KGaA.
Korean domestic supplier SK Specialty is also in talks, but currently lacks its own storage and delivery infrastructure. It is reportedly discussing use of Air Liquide's distribution network to fill the gap.
In plain terms = Korea wants to localize this supply, but the "last mile" — tank storage and transport — still depends on an overseas partner.
How fast will molybdenum demand grow?
Industry estimates for Samsung's molybdenum purchases: ~4 tonnes in 2024 → ~10 tonnes in 2025 → 25 tonnes in 2027 → 80 tonnes by 2030 — a 20× increase in six years.
SK Hynix begins formal molybdenum use next year, with initial annual consumption estimated at roughly 4 tonnes.
This means → with Samsung and SK Hynix ramping simultaneously, the molybdenum supply chain is entering a rapid expansion cycle. Upstream suppliers' capacity planning becomes the next bottleneck to watch.
Content is for reference only, not financial advice.