“Tungsten Retreats, Molybdenum Advances”: SK Hynix Completes Validation of 375-Layer NAND; Three U.S. Stocks Stand to Benefit
TechFlow Selected TechFlow Selected
“Tungsten Retreats, Molybdenum Advances”: SK Hynix Completes Validation of 375-Layer NAND; Three U.S. Stocks Stand to Benefit
The real winners of this material substitution are not the memory manufacturers, but rather the upstream suppliers that sell equipment and consumables.
Navigating Web3 tides with focused insightsAuthor: Ada, TechFlow
According to TrendForce, a semiconductor industry research firm, SK Hynix has completed design verification for its 375-layer NAND flash memory, with mass production scheduled to begin by the end of 2026. The company plans to convert existing capacity to produce this new generation of chips. The successful verification of SK Hynix’s 375-layer NAND brings into focus a long-anticipated inflection point in the industry: tungsten—a material used in chips for nearly a quarter-century—is being replaced by molybdenum. Yet the true winners of this materials transition are not the memory manufacturers themselves, but upstream suppliers of equipment and consumables.
Tungsten Held the Line for Nearly 25 Years—Scaling Is Now Pushing It Beyond Its Physical Limits
Notably, Samsung—not SK Hynix—was the first to introduce molybdenum into metal interconnects. Samsung adopted molybdenum in its 286-layer, ninth-generation NAND, which entered mass production in April 2024, and is now expanding molybdenum usage across additional process steps. SK Hynix’s adoption marks its first use of molybdenum on its own product line—an effort to catch up, rather than pioneer.
This 375-layer product itself has undergone a revision downward. According to TheElec, SK Hynix originally targeted a 400-layer architecture, but scaled back to 375 layers due to manufacturing complexity at ultra-high stack heights. Even so, it represents a pivotal leap in SK Hynix’s NAND roadmap; future generations—480-layer and 604-layer products—are expected to rely even more heavily on molybdenum.
Although molybdenum’s replacement of tungsten is not exclusive to memory, it triggers an industry-wide inflection point. Tungsten has served as the interconnect metal in NAND, DRAM, and logic/foundry mid-end processes for nearly 25 years—but scaling demands are now breaching tungsten’s physical limits, making molybdenum the most promising alternative.
Molybdenum’s advantages extend beyond lower resistivity. Unlike tungsten and copper, molybdenum does not require a diffusion barrier layer, simplifying process flow and improving yield. Its high melting point and oxidation resistance support direct deposition—making it better suited for high-aspect-ratio structures like those found in 3D NAND and gate-all-around (GAA) logic devices. In short: the higher the stack, the smaller the node—the more strained tungsten becomes, and the greater molybdenum’s penetration potential grows. This forms the foundational logic of the “pick-and-shovel” narrative: once adoption scales, the real beneficiaries are the providers of tools and materials.
Lam Research: The Sole Supplier of Mass-Produced ALD Molybdenum Tools
The most direct and compelling player in this chain is Lam Research (LRCX). Its ALTUS Halo system—launched in February 2025—is billed by the company as the industry’s first atomic layer deposition (ALD) tool for molybdenum in mass production, delivering over 50% lower resistance than conventional tungsten metallization in most cases. Lam reports early adoption is already underway at high-volume 3D NAND fabs in Korea and Singapore, as well as at advanced logic fabs—Korea corresponding to Samsung and SK Hynix, Singapore to Micron.
Business upside lies in process complexity. According to Zacks Research, Lam is currently the only supplier with an ALD molybdenum tool deployed in volume production, serving both foundry and NAND customers. Although molybdenum deposition is slower and more complex, it triples Lam’s available market (SAM) for single-wafer metal deposition at these advanced nodes. Greater process difficulty translates directly into incremental opportunity for equipment vendors.
Entegris Supplies Consumables; Micron Is the Only Pure-Play Memory Stock in U.S. Equities
Applied Materials (AMAT) follows a different path. In February this year, it introduced its Spectral ALD system, substituting molybdenum for tungsten in transistor contacts—specifically targeting the critical interface between transistors and copper interconnect networks—to reduce resistance. The system is already in use at several leading logic foundries. Crucially, Lam’s molybdenum story centers on NAND/DRAM wordlines, while AMAT’s focuses on 2nm GAA logic contacts—different downstream applications, and thus distinct investment narratives.
On the materials side, Entegris (ENTG) supplies the solid-state molybdenum precursor molybdenum dioxide dichloride (MoO₂Cl₂), tailored for DRAM and 3D NAND, alongside its ProE-Vap delivery system. Its thesis rests on the cascading impact of materials switching. According to Entegris, transitioning from copper and tungsten to molybdenum affects precursor selection, polishing pad design, slurry formulation, etch chemistry, filtration—and more. While the process becomes more fragmented, it spans multiple steps.
As for memory manufacturers themselves: Samsung and SK Hynix are not listed on U.S. exchanges. Micron (MU) is the sole pure-play memory stock in U.S. equities—and it holds a head start on molybdenum adoption. Citing Micron NAND Development VP Mark Kiehlbauch, Lam notes that molybdenum metallization enabled Micron to achieve industry-leading I/O bandwidth and storage capacity in its latest NAND generation. Still, Micron is a “user” of molybdenum—not a beneficiary of its sale—and its stock remains driven primarily by memory cycles and HBM demand; molybdenum serves only as a performance enhancer.
Do Molybdenum Miners Benefit? Semiconductor Demand Is Negligible
In theory, molybdenum mining companies could be marginal beneficiaries of this narrative—Freeport-McMoRan (FCX), whose molybdenum output is a byproduct of copper mining, is the relevant U.S.-listed counterpart. But semiconductor-grade molybdenum consumption is minuscule. Per TheElec and industry estimates, Samsung purchased roughly 4 tons of molybdenum last year and about 10 tons this year; SK Hynix began at ~4 tons. Total industry demand is projected to reach only ~80 tons by 2030. Against a global molybdenum market dominated by steel alloys—consuming tens of millions of tons annually—semiconductor demand amounts to a rounding error. Linking miner stock prices to the NAND narrative is causally unsound.
This delineates the true focal point of the “tungsten-out, molybdenum-in” story: SK Hynix’s 375-layer NAND is merely the starting point. The full impact spans three categories—NAND, DRAM, and materials. In this metals transition, equipment and consumables suppliers enjoy higher certainty; memory makers gain performance benefits—not valuation uplift; and raw-material suppliers see virtually no upside.
Disclaimer: This article does not constitute investment advice.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
深潮TechFlow
