Samsung Electronics plans to introduce a 400-layer vertical NAND (V NAND) chip by 2026 to address rising demand for high-capacity storage in AI servers, according to The Korea Economic Daily on October 28. Using bonding technology, Samsung will stack memory cells and circuits separately on different wafers, addressing heat and capacity issues in chips exceeding 300 layers. The "Bonding Vertical NAND" (BV NAND) promises 1.6 times greater storage density, suitable for SSDs in AI data centers. Samsung also aims to launch 1,000-layer NAND by 2030 and faster V11 NAND by 2027.
Global server shipments are expected to grow 5.8% year-on-year to 13.6 million units in 2024, according to Market Intelligence & Consulting Institute (MIC). Dell, HPE, and Supermicro are projected to become the top three global server brands, driven by U.S. chip restrictions, increasing enterprise AI server demand, and recovering server procurement. MIC estimates AI servers will constitute nearly 25% of global server shipments by 2027.
Samsung is also collaborating with SK Hynix to standardize Low Power Double Data Rate 6 (LPDDR6)-Processing in Memory (PIM) technology, reports Business Korea. This collaboration seeks to enable low-power, AI-optimized memory for on-device AI systems, which improve performance and reduce power consumption. The global on-device AI market is projected to grow 37.7% annually, reaching USD 173.9 billion by 2030, according to the same source.