- SK Hynix says it has become the first supplier to mass-produce NAND flash memory with more than 300 stacked layers, using a breakthrough stacking process.
- The 321-layer product promises higher speed and energy efficiency and is aimed at AI applications and high-density storage markets.
What happenedNew NAND milestone in production
South Korea’s SK Hynix has begun mass production of its 321-layer 3D NAND flash memory, marking the first time in the industry that more than 300 layers of NAND have been stacked in a commercially produced product — a major leap from its previous record of 238 layers.
The new technology uses a refined “3 plugs” process that connects multiple vertical conductors through optimised manufacturing steps and innovative low-stress materials to prevent wafer warping. This stacking breakthrough helps increase throughput and yield while reducing manufacturing hurdles associated with extremely tall NAND structures.
Compared to its predecessor, the 321-layer NAND delivers roughly a 12 per cent improvement in data transfer speed and a 13 per cent boost in read performance, along with more than a 10 per cent gain in power efficiency.
SK Hynix plans to supply these products to customers starting in the first half of next year, with applications ranging from high-capacity solid-state drives (SSDs) for data centres to storage that feeds next-generation AI systems.
Also Read: SK Hynix commits nearly $13B to advanced AI memory packaging plant in South Korea
Also Read: SK Hynix ramps up chip packaging capacity in South Korea
Why it’s important
Memory demand is rising sharply as artificial intelligence, cloud computing and big data workloads drive a need not just for faster processing but for denser, more efficient storage. High-layer NAND stacks help achieve that by allowing more memory cells per unit area without proportional increases in cost or power draw.
SK Hynix’s achievement not only strengthens its competitive position against rivals such as Samsung and Micron but also highlights how memory makers are innovating beyond conventional scaling limits to support burgeoning data-intensive applications.
The new NAND generation may help address part of the global memory supply shortage that has affected the semiconductor sector, driven in part by allocation of capacity toward AI-focused products.
As SSDs based on the 321-layer architecture roll out, enterprises and cloud providers could benefit from improved storage density and efficiency, which are key to reducing costs per terabyte and enabling next-generation AI workloads at scale.
