Samsung FMS 2026: zHBM, 400+ Layer V10 BV-NAND & zNAND-O Unveiled for Next-Gen AI Memory, Highlighting IDM Advantage

1. FMS 2026: Samsung's Stage for AI Memory Innovation

This section provides an overview of the Future of Memory and Storage (FMS) 2026 event, detailing Samsung Electronics' significant presence, keynote address, and exhibition concept, which served as the backdrop for the groundbreaking zHBM and V10 NAND announcements.

Keynote Highlights: Driving 3D Innovations

The Future of Memory and Storage (FMS) 2026 took place from August 4 to 6, 2026, at the Santa Clara Convention Center in California, USA, serving as a critical platform for industry leaders.
On the event's opening day, August 4, at 11:40 AM local time, Samsung executives took the stage for a keynote address titled, "Driving the wave of AI Revolution: 3D Innovations in Memory & Storage Architecture."
The presentation was delivered by Jin-Yup Lee, Vice President of Flash Development, and Kyeong-Ryun Kim, Senior Vice President of DRAM Development.
They presented a detailed technical vision focused on maximizing AI system performance and enhancing power efficiency to meet the rapidly increasing demand for High-Performance Computing (HPC) and advanced AI infrastructure.

Immersive AI Cloud Booth Experience

Samsung Electronics participated in FMS 2026 to present its comprehensive vision for next-generation memory technology specifically tailored for the AI era.
The company's exhibition booth, approximately 20 pyeong in size, was uniquely designed to resemble an AI cloud server, creating an immersive experience for attendees.
Inside, the space was strategically organized into three distinct zones: 'Highlight,' 'Cloud AI,' and 'Edge AI.'
Across these zones, Samsung showcased about 30 different products, highlighting its extensive memory portfolio spanning DRAM, NAND, and integrated storage solutions.

Event Detail	Information
Event	FMS (Future of Memory and Storage) 2026
Dates	August 4 - 6, 2026 (local time)
Location	Santa Clara Convention Center, California, USA
Samsung Booth Size	Approximately 20 pyeong
Products Exhibited	About 30 products (DRAM, NAND, storage)
Keynote Speakers	Jin-Yup Lee (VP, Flash Development) & Kyeong-Ryun Kim (SVP, DRAM Development)

2. zHBM: Samsung's Vertical Memory Leap for AI Accelerators

Architectural Innovations and Benefits

At FMS 2026, Samsung unveiled mock-ups of zHBM, a next-generation 3D memory architecture poised to succeed HBM4.
This concept model introduces a revolutionary design that stacks HBM vertically and directly on top of AI accelerators, or xPUs.
By eliminating the traditional interposer and minimizing the data transfer distance between the processor and memory, the zHBM architecture is engineered to significantly increase both bandwidth and power efficiency.
This direct-stacking approach is designed to support the ultra-high-speed data processing demanded by large-scale AI model training and inference.

Unprecedented Performance and Efficiency

The direct-on-chip architecture of zHBM translates into dramatic gains over the HBM5 standard.
Next-generation interface systems utilizing zHBM can deliver up to eight times higher performance compared to HBM5.
Simultaneously, it improves power efficiency by up to three times and achieves more than ten times the memory density.
Critically for high-power AI accelerators, zHBM also reduces thermal resistance by more than half, a crucial improvement for managing heat in densely packed systems.

Metric	Baseline Comparison	zHBM Advantage
Performance	HBM5	Up to 8x higher
Power Efficiency	HBM5	Up to 3x improvement
Thermal Resistance	HBM5	Reduced by more than 50%
Memory Density	HBM5	More than 10x higher

Customization for AI Acceleration

Beyond its raw performance, zHBM is presented as a customizable 3D memory solution.
The architecture allows for the addition of custom IPs into the interlayer between the memory and the AI accelerator.
This unique feature enables customers to optimize product characteristics to meet specific needs, whether that involves memory capacity expansion or direct accelerator performance enhancement.
Samsung emphasized that it plans to maximize zHBM performance through close collaboration with its customers.
Following the roadmap preview at FMS 2026, commercial samples are targeted for late 2026 or early 2027.

3. zNAND-O: High-Performance NAND Tailored for On-Device AI

Alongside its other major memory announcements, Samsung offered a glimpse into its roadmap for specialized storage by unveiling zNAND-O, a next-generation high-performance NAND flash solution currently under development.
As a concept model and roadmap preview, the company showcased mock-ups for the industry's first time at FMS 2026, signaling a clear focus on the rapidly growing edge computing market.

Leveraging V-NAND and TSV Technology

The foundational design of zNAND-O is a 3D packaged product built by combining two of Samsung's core competencies: its established V-NAND vertical stacking technology and the advanced interconnect capabilities of TSV (Through-Silicon Via) technology.
This sophisticated construction allows for the stacking of either 4-layer or 8-layer semiconductor chips into a single, compact package.
By leveraging TSV, zNAND-O can create direct vertical electrical connections between the stacked chips, which is critical for achieving the high bandwidth and low latency required for its target applications.

Optimized for On-Device AI Environments

The "-O" in the zNAND-O product name explicitly signifies its optimization for On-device AI environments.
This solution is engineered to support the high performance necessary in scenarios requiring real-time processing of large-scale data directly on a device, without relying on the cloud.
To meet these demands, zNAND-O offers three key advantages: high spatial efficiency, allowing powerful storage to fit into small form factors; high data loading bandwidth, for quickly loading large AI models and datasets; and fast response speeds, to enable rapid inference and decision-making.
While currently in the concept phase, Samsung is targeting the release of commercial samples for late 2026 or early 2027.

4. V10 BV-NAND: Scaling New Heights with 400+ Layers

As part of its next-generation memory roadmap, Samsung offered a first look at its 10th-generation V-NAND, the V10 BV-NAND, during its presentation at FMS 2026.
This new memory represents a significant leap in vertical stacking, pushing the boundaries of what is possible in NAND flash architecture.
While the models shown were concept mock-ups as part of a roadmap preview, Samsung has targeted late 2026 or early 2027 for the availability of commercial samples.

Ultra-High Stacking with 400+ Layers

The defining characteristic of the V10 BV-NAND is its industry-leading layer count, achieving an ultra-high stack of 400 or more vertical layers.
This feat is made possible by combining two key manufacturing techniques: wafer bonding and an advanced 3 Stack technology.
By successfully integrating these processes, Samsung can construct taller and denser memory structures, paving the way for substantial increases in storage capacity within the same physical footprint.

Enhanced Performance and Density

The move to a 400+ layer design delivers significant gains across multiple key metrics.
Compared to the previous V9 generation, the V10 BV-NAND improves memory density by approximately 58%.
This leap in density is complemented by improvements in both performance and power efficiency.
The new generation features enhanced data read/write performance and a higher I/O speed than its predecessor.
Furthermore, V10 is designed for greater power efficiency, with one source reporting a 2.5x improvement over previous baselines.
These advancements are specifically engineered to meet the demands of the high-performance and high-capacity NAND market.

Metric	V9 V-NAND (Previous Gen)	V10 BV-NAND (Announced at FMS 2026)
Vertical Layers	Baseline	400+ Layers
Memory Density	Baseline	~58% Improvement
Power Efficiency	Baseline	2.5x Greater Efficiency (per source)

5. Samsung's Expansive AI Memory Portfolio and Roadmap

This section delves into the broader portfolio of next-generation AI memory and storage solutions Samsung exhibited at FMS 2026, providing the wider context for the headline announcements of zHBM and V10 V-NAND by showcasing the company's comprehensive strategy and industry leadership.

HBM Innovations: From HBM4E to HBM5

Samsung's showcase solidified its leadership in the high-bandwidth memory market, highlighting significant milestones already achieved in 2026.
The company successfully mass-produced and shipped HBM4 in February 2026, a product built on its advanced 1c DRAM and 4nm base die technology.
Building on this momentum, Samsung supplied samples of its next evolution, HBM4E, to customers in May 2026.
At FMS 2026, HBM4E was positioned as offering the industry's highest performance and highest capacity.
To substantiate this claim, Samsung displayed both HBM4E wafers and a next-generation AI platform powered by the new memory.
Looking to the future, the company also revealed a mock-up of its HBM5 solution.
This next-generation memory is designed for even higher bandwidth and capacity, and notably applies a new thermal management technology called HPB (Heat Path Block) to handle the increased performance demands of future AI systems.

LPDDR5X-PIM: In-Memory Computing for Efficiency

A key highlight of Samsung's exhibit was the LPDDR5X-PIM (Processing-In-Memory), a product developed by the company for the world's first time.
This groundbreaking memory solution is engineered to perform calculations directly within the memory, a paradigm shift from traditional computing architecture.
By processing data where it is stored, LPDDR5X-PIM optimizes data movement between memory and processor, which in turn drastically improves power efficiency.
This capability is crucial for accelerating AI workloads, particularly in power-constrained environments.

Comprehensive Storage Solutions for AI

Samsung's vision for AI infrastructure extends beyond DRAM to encompass a full range of storage solutions.
The company's exhibit featured next-generation memory and storage solutions optimized for the demanding environments of high-performance computing and AI data centers.
This included key enterprise SSD products such as the PM1763 and BM1773, rounding out a comprehensive portfolio designed to address every layer of the AI data pipeline.

Product	Key Innovation	Primary Benefit
HBM4E	Industry's highest performance & capacity; based on 4nm process	Powering current-generation high-performance AI accelerators
HBM5 (Mock-up)	Heat Path Block (HPB) for thermal management; Higher bandwidth	Enabling future AI hardware with superior heat dissipation and performance
LPDDR5X-PIM	In-memory computing (performs calculations within memory)	Optimizes data movement and improves power efficiency for AI tasks

6. Samsung's IDM Advantage: Powering AI Innovation End-to-End

While the groundbreaking zHBM and 400-layer V10 V-NAND captured headlines at FMS 2026, the strategic foundation enabling these innovations is equally significant.
Samsung's presentation highlighted that its next-generation memory roadmap is deeply rooted in its comprehensive, End-to-End solution capabilities as an Integrated Device Manufacturer (IDM).
This integrated structure is central to the company's plan to strengthen its semiconductor technology leadership in the AI era.

The One-Stop Solution for AI Semiconductors

Samsung stands as a comprehensive semiconductor company, uniquely positioned in the global market as the only provider of a 'one-stop solution' for AI chips.
This all-inclusive model encompasses every critical stage of semiconductor production: memory, logic, foundry, and advanced packaging.
By offering integrated services that span the entire lifecycle from initial product design to final mass production, Samsung provides a seamless and holistic manufacturing ecosystem.
The company's strategy leverages this unique position, combining next-generation memory technology innovation with differentiated solutions across its foundry and packaging divisions to maintain its competitive edge.

Accelerating Custom AI Development

For companies developing specialized hardware, Samsung's IDM model offers a powerful advantage by directly supporting the implementation of custom AI semiconductors.
Engaging with a single partner that manages memory, logic, and assembly streamlines the entire development process, leading to a significant reduction in development time.
Furthermore, this integrated approach allows for co-optimization across different components, resulting in superior performance and greater power efficiency for the final AI chip.
This ability to deliver a complete, optimized solution from start to finish is the core of Samsung's strategy for driving AI infrastructure innovation.