SK Hynix anticipates that the expansion of the artificial intelligence (AI) inference market will create new demand for NAND flash memory.
During a conference call on third-quarter earnings held on the morning of October 28, SK Hynix stated, "The proliferation of agentic AI and large-scale data processing is rapidly increasing temporary storage data," adding, "There are limitations in capacity and cost when storing all this data in high-bandwidth memory (HBM) and DRAM."
The company emphasized that, "It is becoming increasingly important to store data based on usage frequency across HBM, DRAM, and solid-state drives (SSD)," and noted that NAND will face new demand opportunities for storing and reusing data required for AI inference.
Specifically, SK Hynix pointed out that "SSDs are already being used for temporary data storage in some AI systems," and projected that their role will expand as shared storage for large-scale data among multiple servers and graphics processing units (GPUs).
SK Hynix also mentioned that it is pursuing technologies to optimize data storage locations and develop next-generation memory, stating, "In the medium to long term, we will respond to new memory demands through high-bandwidth flash (HBF) that combines NAND's large capacity and cost efficiency with high data transfer speeds."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.