New Delhi, August 5(APAC Media): Samsung has introduced its next-generation AI memory technology to improve the speed and efficiency of AI systems.
The company made the announcement at the Future of Memory and Storage (FMS) 2026 conference in California. Samsung unveiled its new V10 Bonding V-NAND (BV-NAND) prototype and also shared new memory designs that are being developed for future AI servers. The new technology is designed to handle the growing amount of data used by AI applications, especially AI inference, where AI models quickly process and generate responses.
New Memory Built For AI
Samsung said its V10 BV-NAND uses a new wafer-bonding design with more than 400 memory layers. The company claims it offers about 58% higher storage density than its previous V9 NAND technology. It also improves read, write and data transfer speeds while using power more efficiently.
Samsung also introduced two new memory concepts called zHBM and zNAND-O. Unlike current high-bandwidth memory, which sits next to AI processors, the new design places memory directly above AI chips. This can help move data faster while reducing heat and power use.
Why It Matters
As AI models become larger, they need faster and more efficient memory to process huge amounts of data. Samsung said its new memory design could deliver more than 10 times the memory density of today’s HBM5 technology and improve energy efficiency by up to three times. The company has not announced when these technologies will be available in commercial products.
The announcement comes as demand for AI infrastructure continues to grow worldwide. Chip companies are now focusing on building faster and more power-efficient memory to support the next generation of AI servers and data centres.
