Back to List
Dell and Kioxia Achieve 10 PB Storage Density in 2RU Server Using High-Capacity LC9 QLC SSDs
Industry NewsDellKioxiaNVMe SSD

Dell and Kioxia Achieve 10 PB Storage Density in 2RU Server Using High-Capacity LC9 QLC SSDs

Dell and Kioxia have announced a significant milestone in storage density, successfully integrating nearly 10 petabytes (PB) of data into a compact 2RU server. By utilizing Kioxia’s LC9 high-capacity QLC SSDs, Dell has populated its PowerEdge R7725xd server with 40 E3.L form factor drives, each offering 245.76 TB of capacity. This configuration results in a total system capacity of 9.8 PB. Powered by AMD EPYC 9005 processors and supporting high-speed connectivity via five 400 Gbps NICs, the system is designed to meet the scaling demands of AI infrastructure, data lakes, and massive ingestion streams. This advancement represents a major shift in data center efficiency, potentially allowing for 196 PB of storage within a single rack while significantly improving total cost of ownership (TCO).

Hacker News

Key Takeaways

  • Unprecedented Density: Dell and Kioxia have demonstrated a 9.8 PB all-flash storage solution housed within a slim 2RU (Rack Unit) chassis.
  • Advanced SSD Technology: The system utilizes 40 Kioxia LC9 E3.L form factor NVMe SSDs, with each individual drive boasting a capacity of 245.76 TB.
  • High-Performance Compute: The storage server is built on the Dell PowerEdge R7725xd platform, powered by AMD EPYC 9005-series processors.
  • Massive Throughput: To handle the immense data capacity, the server supports up to five 400 Gbps Network Interface Cards (NICs) for rapid data transfer.
  • Rack-Scale Efficiency: This technology enables a theoretical density of 196 PB per rack when utilizing twenty of these 2RU servers.

In-Depth Analysis

Hardware Synergy: Dell PowerEdge R7725xd and Kioxia LC9

The collaboration between Dell and Kioxia centers on the integration of the LC9 high-capacity QLC (Quad-Level Cell) SSD into Dell's enterprise server lineup. The Dell PowerEdge R7725xd serves as the foundation for this high-density feat. By leveraging the E3.L form factor, which is optimized for thermal efficiency and capacity in enterprise environments, Dell is able to fit 40 of Kioxia's 245.76 TB NVMe SSDs into a 2RU space.

This density is made possible by the evolution of QLC NAND technology, which allows for higher bit density per cell compared to traditional TLC (Triple-Level Cell) drives. The use of the AMD EPYC 9005-powered platform ensures that the server has the necessary PCIe lanes and processing power to manage 40 high-capacity NVMe drives simultaneously. The inclusion of five 400 Gbps NICs is a critical design choice, ensuring that the massive 9.8 PB of stored data does not become bottlenecked during high-speed ingestion or retrieval tasks.

Scaling AI Infrastructure and Data Lakes

According to executives from both Dell and Kioxia, the primary driver for this level of density is the burgeoning demand for AI infrastructure. Arun Narayanan, SVP of Compute and Networking at Dell, emphasized that the combination of the PowerEdge R7725xd and Kioxia’s SSDs provides the storage density and power efficiency required to scale AI workloads without sacrificing performance.

For enterprise customers, this means the ability to deploy massive ingestion streams and scale data lakes with a significantly reduced physical footprint. Neville Ichhaporia of Kioxia America noted that these servers allow for large-scale backups and data management in a fraction of the space previously required. This reduction in footprint directly translates to improved Total Cost of Ownership (TCO), as data centers can store more data per square foot while potentially reducing power and cooling overhead relative to lower-density configurations.

The Competitive Landscape of High-Capacity Flash

Kioxia is not alone in the pursuit of the 256 TB-class SSD market. The industry is seeing a surge in high-capacity flash development from major players including Micron (with the 6600 ION), Sandisk (UltraQLC SN670), and SK Hynix (AIN D) along with its subsidiary Solidigm. This collective push toward ultra-high capacity signifies a broader industry trend where flash storage is increasingly positioned as a replacement for traditional Nearline Hard Disk Drives (HDDs).

Furthermore, reports indicate that Scality is working to support future nearline-class SSDs from Samsung. These upcoming Samsung drives are being viewed as "HDD killers," with roadmaps suggesting capacities could eventually reach 1 PB per drive. The current Dell-Kioxia 10 PB (9.8 PB) server is a tangible step toward this future, proving that all-flash arrays can now compete with and exceed the density of traditional mechanical storage systems.

Industry Impact

The introduction of a 10 PB 2RU server marks a pivotal moment for the storage industry. By enabling 196 PB of storage in a single rack, Dell and Kioxia are effectively redefining the spatial requirements of the modern data center. This level of density is particularly critical for AI training environments, where massive datasets must be stored locally to ensure low-latency access for GPU clusters.

As flash technology continues to scale toward the 1 PB per drive milestone, the economic and operational arguments for maintaining large-scale HDD arrays are likely to diminish. The shift toward high-capacity QLC SSDs suggests that the industry is prioritizing density and power efficiency as the primary metrics for next-generation storage infrastructure. This trend will likely accelerate the obsolescence of traditional hard drives in high-performance and high-density enterprise applications.

Frequently Asked Questions

Question: How much total storage does the Dell PowerEdge R7725xd provide in this configuration?

In this specific configuration using Kioxia LC9 SSDs, the server provides approximately 9.8 PB of storage capacity within a 2RU chassis.

Question: What specific SSDs are used to achieve this 10 PB capacity?

The system uses 40 Kioxia LC9 E3.L form factor NVMe SSDs. Each of these QLC-based drives has a capacity of 245.76 TB.

Question: What are the primary use cases for such a high-density storage server?

The system is designed for scaling AI infrastructure, managing massive data ingestion streams, effortlessly scaling data lakes, and handling large-scale backups with a minimal physical footprint.

Related News

Inside the Architecture of vLLM: A Comprehensive Breakdown of High-Throughput LLM Inference Systems in 2025
Industry News

Inside the Architecture of vLLM: A Comprehensive Breakdown of High-Throughput LLM Inference Systems in 2025

This technical analysis explores the architecture of vLLM, a state-of-the-art high-throughput Large Language Model (LLM) inference system. Based on the V1 engine as of August 2025, the breakdown details the core components that enable efficient inference, including PagedAttention, continuous batching, and advanced scheduling. The article outlines the system's progression from a fundamental offline engine to a sophisticated, multi-GPU serving layer capable of handling concurrent web traffic. Key features such as chunked prefill, prefix caching, and speculative decoding are highlighted as essential for optimizing performance. This overview provides a high-level mental model for developers and researchers interested in the evolution of LLM engines and their role in modern AI infrastructure.

Jony Ive and OpenAI Collaborating on Hockey Puck-Sized Smart Speaker Expected to Launch in 2027
Industry News

Jony Ive and OpenAI Collaborating on Hockey Puck-Sized Smart Speaker Expected to Launch in 2027

Former Apple design chief Jony Ive is reportedly collaborating with OpenAI to develop a new AI-driven hardware device. According to reports from Bloomberg’s Mark Gurman, the device is described as a battery-powered smart speaker without a display. It features a unique doughnut-shaped design roughly the size of a hockey puck. Slated for a 2027 release, the gadget is expected to retail for over $300. This collaboration marks a significant move for OpenAI as it ventures into dedicated consumer hardware, leveraging Ive's renowned design philosophy to create a screenless interface centered on artificial intelligence. The device aims to provide a unique aesthetic and functional experience distinct from current market offerings.

AMD Acquires AI Startup Taalas to Boost Inference Performance by Etching Models Directly into Silicon
Industry News

AMD Acquires AI Startup Taalas to Boost Inference Performance by Etching Models Directly into Silicon

AMD has announced the acquisition of Toronto-based AI chip startup Taalas, a strategic move aimed at challenging Nvidia's dominance in the AI hardware sector. Taalas distinguishes itself through a radical approach to inference: instead of relying on traditional High Bandwidth Memory (HBM) to store model weights, the company "etches" these weights directly into the silicon. This process creates what are termed Model-Specific Integrated Circuits (MSICs). Early benchmarks of Taalas' HC1 test chip, manufactured on TSMC's 6nm process, demonstrated the ability to serve Meta’s Llama 3.1 8B at a staggering 16,960 tokens per second. This performance represents a 48x increase over standard Nvidia GPUs and an 8.5x improvement over Cerebras accelerators. The acquisition is intended to provide faster and more cost-effective "premium" inference services for AI agents and code assistants.