Samsung has revived its Z-NAND memory technology, which had been quiet for some time. The company is developing new generation Z-NAND memory specifically for AI data centers. It offers up to 15 times higher performance compared to traditional NAND memory solutions. It is also claimed to reduce energy consumption by up to 80 percent.
Samsung is making headlines with its Z-NAND memory
At the heart of the newly developed Z-NAND technology is a system called GPU-Initiated Direct Storage Access (GIDS), which allows GPUs to directly access storage units. This system eliminates the need for CPU and DRAM, reducing data access times and latency throughout the system. A Samsung memory executive stated that Z-NAND is specifically optimized for AI applications and high-volume data processing with GPUs.
Z-NAND technology was first introduced in the mid-2010s as a direct competitor to Intel’s 3D XPoint (Optane) technology. The first-generation version attracted attention with its modified V-NAND architecture operating in SLC mode. Reducing the page size to 2 to 4 KB resulted in significant improvements in data read and write speeds.
The new generation Z-NAND is reported to offer significantly higher performance compared to NVMe SSDs. In addition to its speed, the significant energy savings it offers in data centers could pave the way for wider adoption of this technology in line with the growth of AI infrastructures.
However, cost will be a determining factor in the product’s market adoption. It is also known that PCIe 6-based SSDs are in development. This raises questions about how Z-NAND will fare in the competitive landscape.
Samsung’s revival of Z-NAND has the potential to offer an alternative solution in AI infrastructures where data access speeds are critical. Technical advantages such as high performance, low latency, and energy efficiency make this technology attractive for both large-scale data centers and specialized AI systems.
{{user}} {{datetime}}
{{text}}