MYBIGGAMING
LIVE
Agothic
Hardware 15 October 2026 • 2 min read •

KIOXIA CM9-DLC: New PCIe-5.0-SSDs with Liquid Cooling and up to 30.72 TB

KIOXIA unveils the CM9-DLC series: enterprise PCIe 5.0 SSDs optimized for direct liquid cooling, designed for AI servers and high-density data centers.
Author: Гика • PC •
KIOXIA CM9-DLC: New PCIe-5.0-SSDs with Liquid Cooling and up to 30.72 TB

KIOXIA has introduced the CM9-DLC series, a new line of enterprise SSDs designed for AI servers and high-density data center environments. The standout feature is support for Direct Liquid Cooling, allowing these drives to be integrated into the liquid cooling loops of modern server racks via Cold Plates.

The CM9-DLC series is built upon the established CM9 family but features a modified thermal design. This construction enables heat transfer to an external cold plate through thermally coupled surfaces, meaning the liquid itself does not need to come into direct contact with the electronic components. This approach aims to reduce fan power consumption and optimize space in fully liquid-cooled AI racks.

The drives utilize the E3.S enterprise form factor, which offers superior thermal management, hot-swap capabilities, and mechanical integration compared to standard M.2 drives. Additionally, KIOXIA supports dual-port configurations, enabling access through different host connections for high-availability storage systems.

Performance and Endurance Tiers

The series is offered in two distinct endurance classes based on the workload requirements:

  • Read-Intensive: Optimized for read-heavy applications, offering up to 30.72 TB of capacity and 1 DWPD (Drive Write Per Day).
  • Mixed-Use: Designed for higher write demands, providing up to 25.6 TB of capacity and 3 DWPD.

Leveraging the PCIe 5.0 interface and NVMe 2.0 protocol, the drives deliver sequential read speeds of up to 14.8 GB/s and sequential write speeds of up to 11 GB/s. Random read performance is rated at up to 3.4 million IOPS (4-KiB, QD512), while random write reaches 800,000 IOPS (4-KiB, QD512).

Role in AI Workloads

A significant use case for this series is SSD Context Caching. In Large Language Model (LLM) processing, massive amounts of temporary data—such as the Key-Value cache in Attention mechanisms—require significant storage. Since HBM on AI accelerators is both expensive and capacity-limited, high-speed enterprise SSDs can act as a secondary storage tier, helping to manage large context windows by offloading less time-critical data from the GPU memory.

Currently, KIOXIA is shipping samples to selected customers for functional testing. Official retail pricing, precise power consumption figures, and final production specifications have not yet been announced.

Article author

Гика

Cookies

We use cookies

The site uses technically necessary cookies for sign-in, security, storing preferences, and correct page operation. Analytics and other non-essential technologies should only be enabled after your consent. You can read more about data categories, purposes, and retention in the privacy policy.

Cookie policy
Quick actions