Samsung has unveiled a new generation of memory technologies designed for artificial intelligence systems and data centers, including a breakthrough architecture that places memory chips directly on top of AI processors. The company says the new design, known as zHBM, could deliver performance up to eight times greater than expected from next-generation HBM5 memory.

Unlike current systems, where memory sits beside the processor, zHBM stacks memory directly above it, dramatically shortening the distance data must travel. Samsung says the design boosts bandwidth, reduces power consumption, and speeds up processing for large AI models. The company also claims the technology can increase memory density by more than tenfold, improve energy efficiency by up to three times, and cut thermal resistance by more than half.

Samsung also introduced zNAND O, a high-speed storage solution designed for real-time AI applications, and V10 BV NAND, a next-generation flash memory featuring more than 400 layers. The company said the new NAND technology increases storage density by roughly 58% while improving read, write, and data transfer speeds.

In addition, Samsung outlined its roadmap for future HBM4E and HBM5 memory, as well as technologies that allow certain processing tasks to be performed inside the memory chip itself, reducing the need to repeatedly transfer data between processors and memory.

Most of the technologies remain in the prototype stage, and Samsung has not announced a commercial release date for zHBM or zNAND O. While consumers are unlikely to see an immediate impact on personal devices, the innovations could significantly improve the speed, efficiency, and cost of AI services in the years ahead.

(YWN World Headquarters – NYC)