Urgent.News

What's breaking now, across thousands of outlets.

Tech

Samsung Advances SOCAMM2 Memory as Samsung Electro-Mechanics Expands NVIDIA Partnership

Competition in artificial intelligence infrastructure is expanding beyond graphics processing units (GPUs) and high-bandwidth memory (HBM) to system memory, semiconductor package substrates, and power management components. As AI evolves toward agentic systems capable of executing complex tasks autonomously, server architecture and overall system efficiency are becoming increasingly…

Samsung Electronics has introduced SOCAMM2, a next-generation server memory module designed for the evolving AI infrastructure. This technology combines low-power DRAM (LPDDR) with a removable module architecture, enabling memory modules to be replaced without altering the motherboard. As agentic AI systems require increased memory capacity, bandwidth, power efficiency, and space, SOCAMM2 offers flexibility for heatsink placement and airflow management, potentially improving data center maintenance and server lifecycle management.

Samsung Electro-Mechanics has strengthened its partnership with NVIDIA, extending cooperation into multilayer ceramic capacitors (MLCCs). The companies signed a multiyear supply agreement, with Samsung Electro-Mechanics supplying substrates for NVIDIA's next-generation GPUs and high-speed networking products. The expansion aims to support NVIDIA's AI platforms and increase Samsung Electro-Mechanics' visibility into long-term demand, potentially accounting for a significant share of its total package substrate production capacity by 2028.

MLCCs play a crucial role in power stability for AI servers, as they stabilize voltage and suppress electrical noise. Samsung Electro-Mechanics and NVIDIA's collaboration could lead to a stronger global supply chain for AI infrastructure, with Samsung Electro-Mechanics contributing its expertise in substrate processing and high-reliability MLCCs. NVIDIA aims to expand its manufacturing network, including with Samsung Electro-Mechanics, Kinsus, TSMC, Amkor, and Siliconware Precision Industries (SPIL).

Written by urgent.news from Korea IT Times's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at koreaittimes.com →

More in Tech

More from Friday 20 November →