High-capacity NVMe SSD for AI applications

KIOXIA Europe has developed its new KIOXIA LC9 Series 122.88TB NVMe SSD in a 2.5" form factor – the first SSD built with the company's BiCS FLASH generation 8 3D flash memory technology 2Tb QLC die. The series, which is under development, will be showcased at various upcoming conferences.

As AI systems become increasingly sophisticated and data volumes grow, enterprises need storage solutions to keep pace with the complex demands of modern workloads.

High-capacity drives are critical for certain phases of the AI process, including LLMs, training and storing vast datasets, vector databases and the rapid retrieval of information for inference and fine-tuning. Designed for generative AI applications, the new enterprise-class drive is built for high capacity and provides a PCIe 5.0 interface with dual-port capability for fault tolerance or connectivity to multiple compute systems.

These high-capacity QLC-based SSDs are ideal for deploying with hybrid cloud and multi-cloud systems. High-capacity SSDs feed training and inference data to AI server systems through these cloud configurations.

This new SSD complements the recently announced KIOXIA AiSAQ technology, which improves scalable RAG (Retrieval Augmented Generation) performance by storing vector database elements on SSDs instead of costly, limited DRAM. Also, it enhances system and rack-level efficiency with higher storage density and lower power consumption per TB compared to lower-capacity SSDs.

"AI workloads are stretching the capabilities of data storage, asking for larger capacities and swifter access to the extensive datasets found in today's data lakes, and KIOXIA is ready to offer the necessary advanced technologies including 2Tb QLC BiCS FLASH generation 8 of 3D flash memory, CBA and the complimenting AiSAQ," said Axel Störmann, VP and chief technology officer for SSD and Embedded Memory products at KIOXIA Europe GmbH.

"This new LC9 Series NVMe SSD is an instrumental KIOXIA product expansion that will support AI system developers' needs for high-capacity storage, high performance, and energy efficiency for applications such as AI model training, inference, and Retrieval-Augmented Generation on a vaster scale."