Samsung’s 1c (sometimes cross-referenced with the industry-equivalent 1γ/1-gamma) DRAM node is the company’s sixth-generation 10 nm-class process technology. It represents a further shrink within the long-running “10 nm-class” DRAM family (preceded by 1x/1y/1z/1a/1b), targeting roughly the 11–12 nm range in critical dimensions. This places it as one of the densest planar (or near-planar) DRAM processes before the industry transitions more aggressively toward sub-10 nm nodes and structural changes such as 4F² cells or vertical channel transistors.

Naming, Generational Context, and Approximate Scale

DRAM process naming has long been marketing-oriented rather than a strict nanometer measure once the industry entered the 10 nm class around 2016. The sequence for the major players is broadly:

• 1x / early 10 nm-class (~18 nm range in some analyses)
• 1y
• 1z
• 1a / 1α (alpha)
• 1b / 1β (beta)
• 1c / 1γ (gamma) — sixth generation
• 1d and beyond, followed by sub-10 nm efforts (sometimes labeled 0a/10a or similar)

Samsung’s 1c sits at an estimated 11–12 nm class for critical features (wordline/bitline/active pitches and related design rules). Earlier generations were roughly 14 nm-class (1a) and 12–13 nm-class (1b). Shrink factors have slowed; successive generations now deliver more modest density gains (often in the 15–30% bits-per-wafer range) because of fundamental limits in capacitor charge storage, leakage, sensing margins, and pattern fidelity at these scales. Industry analyses note that 6F² cell architectures are approaching practical limits near or below ~10 nm, prompting exploration of 4F² cells, vertical channel transistors (VCT), and new materials.

Micron’s publicly detailed 1γ node (the closest peer) emphasizes >30% higher bits per wafer versus its 1β, speeds up to 9200 MT/s on DDR5, and >20% power reduction, enabled by EUV and next-generation high-κ metal gate (HKMG) CMOS. Samsung’s 1c targets analogous density, speed, and efficiency gains, with particular emphasis on power and thermal advantages for stacked HBM.

Development Timeline and Production Status

• Development completion: Samsung finalized 1c DRAM development and received internal production-readiness approval around mid-2025 (reported early July 2025). This followed a redesign that enlarged die size relative to initial plans, relaxing some peripheral-circuit design rules while protecting the most critical dimensions. The redesign prioritized yield and stability over maximum density, accepting a delay of roughly a year from earlier aggressive targets.
• Yield progression: Early yields were low (sub-30% in some 2024/early-2025 reports). By mid-2025, test yields reached the 50–70% range. By early-to-mid 2026, yields for HBM-related 1c material were reported around 60% and improving, with broader HBM4 package yields climbing toward 80% in some accounts as process and packaging matured.
• Mass production and capacity: Mass production of 1c-based products (including HBM4 core dies) ramped in the second half of 2025 into 2026. Samsung has targeted aggressive capacity expansion—reports indicated goals of tens of thousands of wafers per month initially, scaling toward ~200,000 wafers per month of 1c capacity by the end of 2026 through upgrades at Hwaseong and especially Pyeongtaek (P4 and related lines). The node is used for both conventional DRAM (DDR5, LPDDR, etc.) and as the primary core die for HBM4 and HBM4E.

SK hynix was first to announce a working 1c 16 Gb DDR5 in August 2024 (claiming ~8 Gbps speed, ~11% faster and >9% more power-efficient than its 1b), with volume shipments targeted for 2025. Micron began shipping 1γ samples earlier in the equivalent generation. Samsung’s later but aggressive push on 1c for HBM differentiated it from competitors who initially leaned more heavily on mature 1b material for early HBM4.

Key Technical Features and Manufacturing Challenges

EUV lithography intensity: Samsung has been an early and relatively aggressive EUV adopter in DRAM (starting years earlier). For 1c, reports indicate 6–7 EUV layers (down from earlier internal plans of 8–9) to balance resolution, productivity, and cost/yield. Metal-oxide resists (MOR) have been considered for the finest layers. Higher EUV layer counts improve patterning fidelity for critical features but raise cost and can pressure yields if not fully optimized. Competitors show varying strategies—SK hynix also increasing EUV layers on 1c, Micron more selective/conservative until its 1γ introduction.

Cell and transistor architecture: Still based primarily on the long-standing 6F² buried-channel array transistor (BCAT) / capacitor-on-bitline approach with cylindrical or quasi-cylindrical capacitors and high-κ dielectrics. Scaling pressure is intense: active, wordline, and bitline pitches continue to shrink, sensing margins tighten, and row-hammer and leakage management become harder. Samsung has already demonstrated working dies on a subsequent sub-10 nm “10a” effort that introduces 4F² square cells and vertical-channel transistors (with materials such as InGaO/IGZO explored for leakage control), but 1c itself remains within the established 6F² paradigm.

Performance and efficiency levers:

• Higher bit density (smaller die or more bits per wafer) improves cost structure once yields mature.
• Faster operation and lower operating voltage/power (critical for multi-high HBM stacks and data-center power budgets).
• Thinner dies that facilitate taller HBM stacks (12-Hi, 16-Hi, etc.).
• Design optimizations carried over or refined from the prior 1b platform to reduce risk.

Challenges include atomic-scale process control, EUV stochastic effects and defectivity, capacitor capacitance retention at reduced footprints, thermal management in stacked packages, and the classic density-vs-yield trade-off that prompted Samsung’s die-size enlargement.

Strategic Role, Especially in HBM and the Competitive Landscape

1c is central to Samsung’s HBM recovery strategy. While SK hynix and Micron initially relied more on 1b-class dies for HBM4, Samsung bet on 1c for superior power efficiency, density, and performance differentiation—pairing it with its own advanced logic base dies (4 nm-class foundry process in some reports). This positions 1c as the foundation for HBM4 and HBM4E, with the subsequent 1d node expected later (targeting HBM5E or similar). Capacity expansion at Pyeongtaek and related sites, plus backend packaging focus, aims to close the HBM market-share gap.

Broader applications include high-performance DDR5 for servers/data centers, LPDDR for mobile/AI edge, and GDDR variants. The node’s power and density advantages help address rising energy demands from AI training and inference workloads.

Competitive snapshot (approximate, mid-2020s context):

• Samsung: Aggressive 1c for HBM differentiation; yield maturation and capacity ramp critical.
• SK hynix: Early 1c DDR5 announcement; strong HBM share; also scaling 1c capacity significantly.
• Micron: 1γ leadership in sampling/shipping certain products; more measured EUV adoption historically.

All three face the same physical scaling wall, driving parallel work on materials (high-κ, new channel materials), transistor architecture (VCT), cell layout (4F²), and eventual 3D DRAM concepts.

Outlook and Limitations

1c extends the useful life of conventional 6F² DRAM scaling into the mid-to-late 2020s, buying time for structural innovations. Samsung has already shown early sub-10 nm (≈9.5–9.7 nm class) working dies using 4F² + VCT, with commercialization targeted later (around 2028 in some reports). Yield, cost, and reliability will determine how quickly 1c displaces prior nodes in the bit-supply mix. Exact electrical parameters (speeds, voltages, densities) vary by product (HBM core vs. discrete DDR5/LPDDR) and are typically disclosed at product launch rather than at the pure process-node level.

In summary, Samsung’s 1c / sixth-generation 10 nm-class DRAM is a high-stakes process node optimized for density, efficiency, and HBM leadership. It reflects both the continuing incremental progress possible within planar DRAM and the intensifying need for architectural breakthroughs as traditional scaling saturates. Ongoing yield improvements, capacity ramps, and HBM qualification outcomes will largely define its commercial impact through 2026–2027 and beyond.

---

1) Naming, Generational Context, and Approximate Scale of Samsung’s 1c DRAM Node

The 1c designation sits at the heart of how the DRAM industry communicates process progress once physical dimensions entered the vague “10 nm-class” era. Understanding the naming system, the generational sequence, and the approximate physical scale is essential for comparing Samsung’s technology with SK hynix and Micron offerings and for assessing how close the industry is to the practical limits of conventional planar DRAM.

Industry Naming Conventions

DRAM process nodes abandoned strict nanometer marketing earlier than logic processes. After the mid-2010s, manufacturers adopted sequential letter codes within the “10 nm-class” umbrella rather than claiming precise nanometer values that no longer mapped cleanly to any single critical dimension.

• Samsung and SK hynix use Latin-letter suffixes: 1x → 1y → 1z → 1a → 1b → 1c → 1d …
• Micron switched to Greek letters after exhausting the Latin sequence: 1α (alpha) → 1β (beta) → 1γ (gamma) → 1δ (delta) …

Consequently, Samsung’s 1c and Micron’s 1γ (1-gamma) refer to the same generational step—the sixth generation of 10 nm-class DRAM. The equivalence is widely accepted by industry analysts and is the reason the node is frequently written as “1c / 1γ.”

The full informal label is “sixth-generation 10 nm-class process technology.” Vendors rarely publish exact half-pitch or design-rule numbers; third-party reverse-engineering firms (TechInsights and others) provide the most reliable estimates through physical analysis of production dies.

Generational Sequence Within the 10 nm Class

The progression reflects successive shrinks of the same fundamental 6F² buried-channel-array-transistor (BCAT) cell architecture:

Generation	Samsung / SK hynix	Micron	Approximate Era of Volume Production	Relative Position
1st	1x	—	~2016–2017	Early 10 nm-class
2nd	1y	—	~2018
3rd	1z	—	~2019–2020	First widespread EUV use (Samsung)
4th	1a	1α (alpha)	~2021–2022
5th	1b	1β (beta)	~2023–2024
6th	1c	1γ (gamma)	2025–2026	Current leading-edge node
7th	1d	1δ (delta)	Targeted ~2027

Each step historically delivered roughly 15–30 % higher bits per wafer, improved speed, and lower power, although the magnitude of gains has diminished as physical limits tighten.

Approximate Physical Scale

Public and analyst estimates place Samsung’s 1c critical dimensions in the 11–12 nm range. More precisely:

• Word-line, bit-line, and active pitches are reported in the low-to-mid 11 nm class for the most aggressive variants.
• Earlier nodes for context: 1a ≈ 13–14 nm class, 1b ≈ 12–13 nm class.
• The designation “10 nm-class” is therefore a marketing umbrella rather than a literal half-pitch measurement. True half-pitch continues to shrink, but capacitor aspect ratio, leakage, sensing margin, and pattern fidelity constrain how far conventional 6F² cells can go.

Industry consensus holds that the 6F² planar architecture is approaching saturation near or slightly below 10 nm. Shrink factors (the ratio of linear dimensions from one generation to the next) have risen above 0.9 in recent nodes, meaning each new generation delivers progressively smaller area reductions. This slowdown is the primary reason manufacturers are already demonstrating sub-10 nm experimental dies that abandon pure 6F² layouts in favor of 4F² square cells and vertical-channel transistors.

Why the Naming Matters

The letter-based system allows manufacturers to claim generational leadership without committing to a single, easily comparable nanometer figure. It also masks subtle differences in actual design rules, EUV layer counts, and materials choices between vendors. When Samsung states that its HBM4 core die uses 1c while competitors initially shipped HBM4 on 1b, the difference is both generational (one full shrink) and physical (roughly 1 nm tighter critical dimensions plus associated power and density benefits).

In short, Samsung’s 1c node is the sixth and currently leading production generation of the 10 nm-class DRAM family. It occupies the approximate 11–12 nm design-rule space, sits in direct equivalence with Micron’s 1γ, and represents the final major iteration of conventional 6F² planar DRAM before structural innovations become mandatory for further density gains.

---

2) Development Timeline and Production Status of Samsung’s 1c DRAM Node

Samsung’s sixth-generation 10 nm-class (1c) DRAM process has followed a deliberately cautious yet strategically aggressive path. Early technical hurdles forced a redesign, delaying volume readiness by roughly a year relative to initial internal targets. Once the redesigned process reached production-readiness approval in mid-2025, Samsung rapidly shifted focus to yield ramp, capacity expansion, and qualification for high-bandwidth memory (HBM4), treating 1c as a key differentiator against competitors still relying more heavily on the prior 1b node.

Early Development and Initial Setbacks (2023–Early 2025)

Samsung began serious 1c development shortly after volume production of its fifth-generation (1b) node in May 2023. The company originally aimed to complete development and begin mass production by the end of 2024. Early process integration, however, encountered yield and stability challenges that proved more severe than anticipated.

By the second half of 2024, internal assessments showed that the aggressive die-size shrink and high EUV layer count were limiting manufacturability. Samsung responded with a significant redesign: the core critical dimensions were largely preserved, but peripheral circuitry design rules were relaxed and overall die size was enlarged. The change sacrificed some theoretical bits-per-wafer density in exchange for faster yield learning and greater process margin—an explicit prioritization of stable high-volume manufacturing over maximum density.

During this period, SK hynix publicly announced the industry’s first working 1c 16 Gb DDR5 in August 2024, giving the rival a temporary public-relations lead. Samsung remained comparatively quiet, focusing on the redesign.

Production-Readiness Milestone (July 2025)

In early July 2025, Samsung completed development of the redesigned 1c process and received formal internal production-readiness approval. This milestone confirmed that the node met the company’s criteria for mass-production readiness across performance, reliability, and manufacturability metrics. The achievement came approximately two years after 1b volume production and marked the formal start of the transition from development to production ramp.

At that point, Samsung indicated that HBM4 built on 1c core dies was targeted for mass production in the second half of 2025. Parallel work continued on conventional DRAM products (DDR5, LPDDR variants) that would also migrate to the new node.

Yield Progression (2025–Mid-2026)

Yield has been the central variable determining commercial success.

• Late 2024 / early 2025: Test yields on early 1c material were reported below 30 % in multiple industry accounts.
• Mid-2025: After the redesign and process tuning, wafer-level yields in internal tests climbed into the 50–70 % range.
• Early-to-mid 2026: Yields for HBM-related 1c core dies stabilized around 60 %, with continued improvement. Broader HBM4 package yields (including logic base die, stacking, and bonding) were reported to have risen from below 60 % at the start of mass production toward approximately 80 % by August 2026.

The yield recovery enabled Samsung to begin meaningful customer shipments and to accelerate capacity investments. Die-size enlargement and reduced EUV layer count (settling at 6–7 layers rather than the originally planned 8–9) were credited with improving process window and reducing defectivity.

Mass-Production Ramp and Capacity Expansion (Second Half 2025–2026)

Mass production of 1c DRAM began in the second half of 2025, initially focused on HBM4 core dies and selected high-performance discrete DRAM products. By early 2026, Samsung had commenced volume shipments of HBM4 to major customers, including Nvidia.

Capacity plans have been aggressive:

• Initial production lines were established and expanded at the Hwaseong campus and, more significantly, at Pyeongtaek (particularly P4).
• By the end of 2025, monthly 1c wafer capacity was targeted in the tens of thousands.
• The company has publicly and through industry channels signaled a goal of reaching approximately 200,000 wafers per month of 1c capacity by the end of 2026. This would represent a substantial portion of Samsung’s overall DRAM output and is intended to support both HBM growth and conventional high-density products.

Additional cleanroom and equipment investments at Pyeongtaek (including groundwork for P5 phases) and backend packaging capacity expansions are underway to support the higher stack heights and volumes associated with HBM4 and HBM4E.

Current Status as of August 2026

As of late August 2026, Samsung’s 1c node is in volume production and is the primary process for its HBM4 and HBM4E core dies. Yield continues to improve, capacity is scaling rapidly, and the node is also migrating into broader DDR5 and LPDDR product lines. Samsung has elected to keep 1c as the foundation for both HBM4 and the follow-on HBM4E generation, deferring the subsequent 1d node primarily to later HBM generations (such as HBM5E).

The combination of the mid-2025 readiness milestone, steady yield recovery, and large-scale capacity commitments has positioned 1c as a central element of Samsung’s strategy to regain share in the premium HBM market. Remaining challenges center on further yield maturation, cost reduction through higher volume, and successful qualification of higher-stack and higher-bandwidth HBM variants that rely on the tighter design rules and improved power characteristics of the 1c process.

---

3) Key Technical Features and Manufacturing Challenges of Samsung’s 1c DRAM Node

Samsung’s sixth-generation 10 nm-class (1c) DRAM process represents an incremental but technically demanding extension of conventional planar DRAM scaling. It retains the industry-standard 6F² buried-channel array transistor (BCAT) architecture while pushing critical dimensions, patterning intensity, and materials optimization to new limits. The node delivers measurable gains in density, speed, and power efficiency, yet these improvements come with elevated process complexity, yield sensitivity, and cost pressures that required a mid-course redesign.

Core Architectural Continuity

The 1c node continues the long-standing DRAM cell formula used across the industry for more than a decade:

• 6F² cell layout — each memory cell occupies a rectangular footprint of approximately 3F × 2F, where F is the minimum feature size.
• Buried-channel array transistor (BCAT) with metal-gate word lines.
• Capacitor-on-bitline (COB) structure employing cylindrical or quasi-cylindrical capacitors.
• High-κ dielectric stacks (typically ZrO₂- or HfO₂-based systems, sometimes with additional engineered interlayers) to maintain adequate capacitance as the physical footprint shrinks.

No fundamental change to the one-transistor, one-capacitor (1T-1C) topology occurs at 1c. Structural innovations such as 4F² square cells and vertical-channel transistors appear only in subsequent experimental sub-10 nm nodes.

Critical Dimension and Density Targets

Analyst and industry estimates place the most aggressive critical dimensions of Samsung’s 1c process in the 11–12 nm range. Relative to the preceding 1b node (approximately 12–13 nm class), this constitutes a modest linear shrink. Because area scales with the square of linear dimensions, even a small reduction still yields meaningful bits-per-wafer improvement once yields mature—typically targeted in the 15–25 % range for a full generational step, although Samsung’s deliberate die-size enlargement partially offset the theoretical maximum.

The shrink improves:

• Bit density (more bits per wafer or smaller die for the same capacity).
• Signal propagation distances, supporting higher operating frequencies.
• Operating voltage headroom, contributing to lower power consumption—particularly valuable in multi-high HBM stacks.

Lithography Strategy: EUV Intensity and Trade-offs

Extreme ultraviolet (EUV) lithography is central to 1c patterning. Samsung, an early adopter of EUV in DRAM, settled on 6–7 EUV layers for the production version of 1c. This was a reduction from earlier internal plans that contemplated 8–9 layers.

Key aspects include:

• EUV is applied to the most critical layers (active, word-line, bit-line, and selected contacts or landing pads) to reduce multi-patterning complexity and improve overlay and critical-dimension uniformity.
• The remaining layers continue to rely on argon-fluoride immersion (ArFi) deep-ultraviolet multi-patterning.
• Samsung has evaluated metal-oxide resists (MOR) for the finest features to improve resolution and reduce stochastic defects relative to conventional chemically amplified resists.

Higher EUV layer counts improve pattern fidelity but raise capital intensity, mask costs, and throughput challenges. The decision to moderate the layer count was driven by yield and cost considerations after early process learning revealed excessive defectivity.

Materials and Transistor Enhancements

Beyond pure dimensional scaling, 1c incorporates refinements carried forward and optimized from the 1b platform:

• Continued evolution of high-κ metal-gate (HKMG) technology in peripheral and sense-amplifier transistors to improve drive current and reduce leakage.
• Capacitor dielectric and electrode engineering aimed at maximizing capacitance density while controlling leakage and reliability under higher aspect ratios.
• Process optimizations for storage-node landing pads, bit-line air-gap spacers, and related parasitic-capacitance reduction techniques.
• Design-rule relaxation in non-critical peripheral circuitry (enabled by the enlarged die size) to widen process windows without compromising the core array.

These incremental materials and integration improvements support the dual goals of higher speed and lower power that are essential for both discrete DDR5/LPDDR products and HBM core dies.

Primary Manufacturing Challenges

Several interrelated difficulties define the 1c manufacturing landscape:

Yield sensitivity and process window

Early 1c material exhibited yields below 30 %. Even after redesign, HBM-related core-die yields hovered near 60 % in early 2026, requiring continued process tuning. Stochastic EUV effects, overlay errors, and defectivity at the tightest pitches remain primary yield limiters.

Die-size versus density trade-off

Samsung enlarged the die relative to its original aggressive target. The change improved stability and yield learning speed but reduced theoretical bits per wafer. This strategic compromise prioritized manufacturability and HBM qualification timelines over maximum density.

Capacitor scaling limits

As the cell footprint shrinks, maintaining sufficient capacitance (typically targeted above ~20 fF) while controlling leakage becomes progressively harder. Higher aspect-ratio capacitors increase process difficulty and risk of structural collapse or voiding.

Sensing margin, noise, and reliability

Tighter pitches increase bit-line capacitance and crosstalk. Row-hammer susceptibility, data-retention characteristics, and soft-error rates require additional circuit and process countermeasures that consume area or power.

Cost structure

Elevated EUV usage, more complex multi-patterning sequences, and lower early yields raise the cost per good die. Only high volume and continued yield improvement can restore competitive cost positioning versus the more mature 1b node.

Thermal and mechanical constraints for HBM

When 1c dies are used as the core of HBM4/HBM4E stacks, thinner die profiles, improved power efficiency, and mechanical stability under thermal compression bonding become additional requirements. Any residual process variation that affects die warpage or thickness control directly impacts stack yield.

Summary of Technical Position

Samsung’s 1c node successfully extends conventional 6F² DRAM scaling into the 11–12 nm regime through intensified EUV patterning, refined materials, and a pragmatic redesign that favored yield over pure density. The resulting process delivers the density, speed, and power characteristics needed for next-generation HBM and high-performance discrete DRAM, yet it operates near the practical limits of the planar architecture. These limits explain both the manufacturing difficulties encountered during development and the industry-wide parallel effort to introduce fundamentally new cell structures (4F², vertical-channel transistors) for the nodes that will follow 1c and 1d.

---

4) Strategic Role of Samsung’s 1c DRAM Node, Especially in HBM, and the Competitive Landscape

Samsung’s sixth-generation 10 nm-class (1c) DRAM process is not merely an incremental manufacturing upgrade. It functions as a deliberate strategic instrument aimed at reversing the company’s recent share losses in the high-bandwidth memory (HBM) market and reasserting technology leadership in the AI-driven memory segment. By committing to 1c as the core die for HBM4 and HBM4E—while rivals initially leaned more heavily on the mature 1b node—Samsung accepted higher near-term risk in exchange for differentiated power efficiency, density, and performance that it believes will matter most to leading AI accelerator customers.

Central Role in the HBM Roadmap

HBM has become the highest-value and fastest-growing segment of the DRAM market, driven by the bandwidth and capacity demands of large language models and other AI workloads. In this context, the choice of DRAM process node for the core dies inside an HBM stack directly influences:

• Power consumption and thermal density of the multi-high stack.
• Maximum achievable stack height and capacity.
• Overall bandwidth and energy efficiency of the final HBM product.
• Cost structure once yields mature.

Samsung elected to base both its sixth-generation HBM (HBM4) and the subsequent HBM4E generation on 1c core dies. This decision contrasts with the more conservative approach taken by competitors, who initially qualified and ramped early HBM4 products primarily on fifth-generation (1b-class) material. The 1c node’s tighter design rules (approximately 11–12 nm class versus 12–13 nm class for 1b) enable shorter interconnect distances, lower operating voltages, and thinner dies—advantages that compound across 12-high or 16-high stacks.

Samsung pairs these 1c core dies with advanced logic base dies manufactured on its own foundry processes (reported in the 4 nm class for certain HBM4 variants). The combination is intended to deliver superior power efficiency and integration flexibility relative to competitors that rely on external foundries (for example, TSMC 12 nm or 5 nm base dies) or older internal DRAM processes for the base logic.

The company has further signaled that the following node, 1d, will be reserved primarily for later HBM generations such as HBM5E. This staggered transition gives Samsung time to mature 1d yields while extracting maximum commercial value from the 1c investment across two successive HBM generations.

Competitive Positioning Against SK hynix and Micron

The HBM market in the mid-2020s has been dominated by SK hynix, which captured the majority of Nvidia’s HBM3E demand and established strong early positions in HBM4 qualification. Micron has also secured meaningful share through reliable execution and competitive power characteristics. Samsung’s aggressive 1c strategy is an explicit attempt to close that gap.

Key competitive dynamics include:

• Process-node aggressiveness: Samsung is the only major supplier that has committed its leading-edge 1c node to volume HBM4 production from the outset. SK hynix and Micron have used 1b more extensively for initial HBM4 volume while ramping their own 1c/1γ capacity in parallel for later products or discrete DRAM.
• Power and thermal differentiation: Samsung positions the 1c-based HBM4 as offering meaningful power-efficiency advantages—an increasingly decisive factor as AI clusters face power and cooling constraints.
• Capacity and supply commitment: Samsung has publicly and through industry channels targeted rapid expansion of 1c wafer capacity (aiming toward approximately 200,000 wafers per month by the end of 2026). This scale is intended both to support HBM growth and to improve the overall cost structure of the node.
• Yield and qualification risk: The flip side of the aggressive strategy is elevated execution risk. Early 1c yields lagged, and HBM package yields required significant improvement through 2025 and into 2026. Successful qualification with Nvidia and other major customers remains the critical commercial milestone.

In discrete DRAM (DDR5, LPDDR5X/6, GDDR), the competitive picture is more balanced. All three major suppliers are migrating high-volume products to the 1c/1γ generation. Here the strategic value of 1c lies more in cost reduction through higher bits per wafer and in enabling higher-density modules for servers and AI PCs.

Broader Market and Ecosystem Implications

Beyond pure market-share recovery, the 1c node serves several secondary strategic purposes:

• It provides a technology platform that can be leveraged across mobile, client, and data-center segments, improving overall fab utilization.
• It demonstrates process leadership that supports Samsung’s broader narrative of reclaiming the technology high ground after earlier HBM qualification setbacks.
• It buys time for the development of post-planar technologies (4F² cells, vertical-channel transistors, and eventual 3D DRAM) that will be required once conventional 6F² scaling saturates.

At the ecosystem level, the choice of 1c influences base-die foundry relationships, advanced packaging capacity allocation, and customer qualification cycles. Samsung’s vertical integration (DRAM + foundry logic + packaging) is an asset in this regard, provided the company can continue to improve yields and delivery reliability.

Risks and Open Questions

The strategic bet on 1c carries clear risks. If yields plateau below competitive levels, or if customers prioritize proven supply and lower cost over marginal power advantages, the differentiation may prove insufficient to regain substantial HBM share. Conversely, if Samsung successfully matures the node and secures major long-term HBM4/HBM4E sockets, the investment will have positioned the company strongly for the next multi-year AI memory cycle.

As of August 2026, the 1c node is in volume production for HBM4, capacity is scaling, and yields are improving. The ultimate measure of its strategic success will be the share of leading-edge AI accelerator memory that Samsung captures over the 2026–2028 period and the degree to which 1c-based products become the reference for power-efficient, high-density HBM.

---

5) Outlook and Limitations of Samsung’s 1c DRAM Node

Samsung’s sixth-generation 10 nm-class (1c) DRAM process extends conventional planar scaling into the 11–12 nm regime and serves as the technological foundation for the company’s HBM4 and HBM4E ambitions. Its commercial trajectory through 2026–2028 will be shaped by continued yield maturation, capacity realization, customer qualification outcomes, and the broader physical limits of the 6F² architecture. While the node is already in volume production and scaling rapidly, several structural and competitive constraints will determine how long it remains the leading-edge workhorse and how effectively it restores Samsung’s position in the premium memory market.

Near-Term Outlook (2026–2027)

As of August 2026, 1c is firmly in the volume-production phase. Key expectations for the coming 12–18 months include:

• Yield and cost improvement: Wafer-level yields for HBM-related 1c material, which stabilized near 60 % earlier in 2026, are expected to continue rising. Package-level HBM4 yields have already approached 80 % in some reports. Further gains will lower the cost per good bit and improve competitiveness against more mature 1b-based products from rivals.
• Capacity realization: Samsung is targeting roughly 200,000 wafers per month of 1c capacity by the end of 2026 through expansions at Pyeongtaek (P4 and early P5 phases) and conversions at Hwaseong. Achievement of this scale would make 1c a substantial portion of the company’s overall DRAM output and support both HBM growth and high-density discrete products.
• HBM market impact: Successful long-term supply agreements with major AI accelerator customers (particularly Nvidia) remain the decisive commercial test. If 1c-based HBM4 and HBM4E deliver the projected power-efficiency and density advantages at competitive cost and reliability, Samsung can regain meaningful share in the highest-value segment of the DRAM market.
• Product migration: Beyond HBM, 1c is expected to become the mainstream node for high-performance DDR5, LPDDR, and selected GDDR products, improving overall fab utilization and bit-supply growth.

Industry analysts generally view 1c as the last major high-volume node based purely on evolutionary 6F² scaling. It is therefore positioned to serve as the primary advanced DRAM process for at least two HBM generations (HBM4 and HBM4E) before the subsequent 1d node assumes a larger role.

Medium-Term Outlook (2027–2028 and Beyond)

Looking further ahead, the 1c node faces a natural transition:

• Samsung plans to introduce the seventh-generation (1d) node for later HBM variants such as HBM5E, while continuing to extract value from 1c in both HBM and conventional DRAM.
• Parallel development of sub-10 nm processes (sometimes labeled 10a or 0a) that incorporate 4F² square cells and vertical-channel transistors is already underway. Working dies have been demonstrated, with mass-production targets currently projected around 2028. These structural changes are required because linear shrinks within the 6F² framework are delivering diminishing returns.
• The broader industry is expected to follow a similar path: maximize the economic life of 1c/1γ-class technology while preparing the materials, transistor, and cell-architecture innovations needed for the next scaling era.

If Samsung executes well on capacity and yield, 1c should remain a high-volume, high-margin node through at least 2027–2028, after which its relative importance will gradually decline as 1d and early sub-10 nm production ramps.

Fundamental Limitations

Despite its commercial importance, the 1c node operates near several hard physical and economic boundaries that limit its ultimate potential.

Scaling saturation of the 6F² architecture

Linear shrink factors have risen above 0.9 in recent generations, meaning each new node delivers progressively smaller area reductions. Capacitor charge storage, leakage control, sensing margin, and pattern fidelity become increasingly difficult to maintain. Industry analyses indicate that conventional 6F² cells are approaching practical limits near or slightly below 10 nm. This is the primary reason manufacturers are already investing in 4F² layouts and vertical-channel transistors.

Yield and cost structure

Even after the mid-development redesign that enlarged die size and moderated EUV layer count (to 6–7 layers), early yields were low and required substantial recovery effort. High EUV intensity raises capital and mask costs. Only sustained high volume and continued yield improvement can bring the cost per bit into a fully competitive range versus prior nodes.

Power, thermal, and reliability constraints in HBM

While 1c offers power-efficiency advantages relative to 1b, multi-high HBM stacks still face significant thermal density and mechanical challenges. Any residual process variation that affects die thickness, warpage, or leakage directly impacts stack yield and long-term reliability under AI workload conditions.

Competitive response risk

SK hynix and Micron are also ramping their equivalent 1c/1γ nodes and possess strong positions in HBM qualification and supply reliability. If competitors close the power or density gap while offering more predictable delivery, Samsung’s technology differentiation may prove insufficient to shift large volumes of demand.

Dependency on external ecosystem factors

Success depends not only on the DRAM process itself but also on base-die logic technology, advanced packaging capacity, customer qualification cycles, and the overall AI accelerator demand trajectory. Delays or shortfalls in any of these areas can blunt the strategic impact of the 1c investment.

Balanced Assessment

Samsung’s 1c node is a capable and strategically important process that successfully pushes conventional DRAM scaling one more generation while enabling a differentiated HBM offering. Its outlook is positive in the near term provided yields continue to improve and capacity targets are met. However, it is also a transitional technology. The same physical constraints that made 1c difficult to develop are already driving the industry toward architectural change. The ultimate measure of the node’s success will be how effectively Samsung converts its process investment into sustained HBM market share and profitability before the next wave of cell-structure and transistor innovations becomes mandatory.

---