Wednesday, September 16, 2026
AI 인프라 · 뉴스 & 분석
반도체·하드웨어리포트
반도체·하드웨어 · 리포트

Micron announced advanced 512GB DDR5 RDIMM modules with speeds up to 9,200 MT/s for next-generation datacenters, enabling higher-bandwidth memory for AI servers.

DRAM density and bandwidth improvements reduce the CPU-memory bottleneck in AI servers; higher capacity and speed enable larger batch sizes and higher-throughput inference workloads.
업계 전문지Slicast · 2026년 9월 15일 18:11 UTC · 글로벌 · 출처: HPCwire
중요도 63

Micron Technology has announced the successful demonstration of the world's first 512GB DDR5 RDIMM (registered DRAM module), a milestone that reinforces its leadership in high-capacity server memory. The module delivers speeds up to 9,200 MT/s and reduces operating power by more than 60% compared to using four 128GB RDIMMs, enabling enterprise data centers and cloud providers to support demanding AI, analytics, in-memory database, and simulation workloads with improved efficiency.

The unprecedented capacity is achieved through Micron's advanced vertical interconnect packaging innovations, which stack DRAM dies and interconnect them via through-silicon vias (TSVs) to maximize density within individual DRAM packages while maintaining the performance required by modern data center architectures. A single 24-slot dual-socket server can now contain up to 12TB of DDR5 DRAM.

"Micron continues to set the pace for memory technology, delivering a new class of ultra-dense, high-performance server memory that helps customers keep larger datasets closer to the compute engines that need them," said Raj Narasimhan, senior vice president and general manager of the Cloud Memory Business Unit at Micron. "A 512GB RDIMM enables multi-terabyte servers, supporting larger AI and database workloads, greater virtualization density and improved power efficiency, all within existing server footprints."

Both AMD and Intel are actively validating the module across next-generation server platforms. "AMD solutions continue to advance data center performance, scalability and efficiency as AI and data-intensive workloads reshape infrastructure requirements," said Amit Goel, corporate vice president, Compute and Enterprise AI Platform Solutions Engineering at AMD. "Our close engineering collaboration with Micron brings compute and memory innovation together to help customers realize the full value of AMD-powered platforms."

Intel's Karin Eibschitz Segal, general manager of platform and system engineering in the Data Center Group, added: "The growth of AI and other data-intensive workloads is placing new demands on server infrastructure, making memory capacity and bandwidth increasingly important to overall system performance. Intel is working closely with Micron to validate its 512GB DDR5 RDIMM and help enable future server platforms designed to meet the growing demands of next-generation data center workloads."

The proliferation of large language models, agentic AI, real-time inference, and high-core-count CPU workloads is driving demand for greater server memory capacity, higher bandwidth, and improved power efficiency. The 512GB RDIMM addresses these demands by enabling larger datasets to remain resident in main memory, reducing reliance on slower storage tiers and minimizing costly data movement. For memory-bound workloads such as Spark Support Vector Machines-based data analytics, the 512GB RDIMM can deliver up to 1.4 times higher performance compared to 256GB DDR5 configurations. The module also provides significant advantages for memory-intensive databases and caching platforms, including RocksDB and Redis, improving throughput, increasing concurrency, and enabling more efficient dataset scaling.

Micron expects 512GB RDIMMs to enter volume production in the second half of 2027, aligned to customer needs.

원문 보기
Micron announced advanced 512GB DDR5 RDIMM… · Slicast