Samsung Electronics announced on July 8 that it has begun mass production of its next-generation enterprise-grade high-performance solid-state drive (eSSD) 'PM1763,' optimized for artificial intelligence (AI) infrastructure.
The PM1763 is based on the new 'PCIe 6.0' transmission standard, which doubles the data transfer bandwidth compared to its predecessor. It incorporates Samsung's latest 9th Gen V-NAND and a new 4nm controller, significantly enhancing product performance and power efficiency.
The product will be available in three capacities: 4 terabytes (TB), 8TB, and 16TB. Among these, the 16TB model boasts industry-leading performance. It achieves a maximum sequential read speed of 28,400 megabytes (MB) per second and a write speed of 21,900MB, which is twice as fast as the previous model (PM1753). This speed allows for the transfer of a massive language model (LLM) of approximately 40 gigabytes (GB) in just 1.4 seconds, drastically reducing data latency and enhancing AI task efficiency.
Heat management and eco-friendly performance have also been improved. The PM1763 is optimized for liquid cooling environments recently adopted in AI data centers. It utilizes a 'Direct-to-Chip' (D2C) method that attaches cooling plates directly to the chips, enabling long-term operation under heavy workloads without performance degradation. Power efficiency has also improved by more than 1.8 times compared to the previous model, significantly reducing operational costs for data centers.
Additionally, the PM1763 includes security solutions designed for the quantum computing era. It applies the 'PQC' encryption algorithm to prevent hacking using quantum computers and supports 'TDISP' technology to block unauthorized external access in virtualized environments, enhancing data security.
Choi Jae-seok, Senior Vice President of Samsung Electronics' Memory Business Planning Team, stated, "The PM1763 meets the requirements of global customers' next-generation AI platforms based on industry-leading performance and has completed product validation. It will be a key solution to support the efficient operation of our customers' AI models."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.