As artificial intelligence (AI) servers transition from being GPU-centric to storage-focused in the enterprise AI era, Samsung Electronics has identified the growing role of solid-state drives (SSDs) as a core component of AI infrastructure.
On July 8, Samsung introduced its enterprise SSD (eSSD) PM1763 through its newsroom, detailing the product's future role in data centers and the direction of next-generation technology.
According to Samsung, as generative AI and AI agent services proliferate, AI servers are entering a phase where relying solely on GPU performance is no longer sufficient for competitiveness. The rapid increase in the volume of data utilized by AI models has made the input/output speed and data supply capacity of storage devices critical factors influencing overall system performance.
In enterprise AI environments, the emphasis is shifting from training large language models (LLMs) to inference. During the inference process, it is crucial for GPUs to receive data without delay, making the processing speed and latency of SSDs key variables that determine the response time of AI services.
Samsung explained that next-generation enterprise SSDs must not only transmit large volumes of data at ultra-high speeds but also possess high parallelism and stability to handle multiple AI computation requests simultaneously. This capability can reduce GPU wait times and enhance resource utilization in AI servers.
Industry experts believe that as competition in AI semiconductors expands from a GPU focus to include high-bandwidth memory (HBM), CXL, and enterprise SSDs, the importance of storage technology will continue to grow.
Looking ahead, Samsung plans to continuously advance its enterprise SSD technology in line with the expansion of AI data centers and strengthen the development of solutions that enhance the performance and energy efficiency of AI infrastructure.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.