Samsung is expected to more than double output of its HBM4 and HBM4E DRAM

Samsung is preparing to more than double production of its HBM4 family next year by dramatically scaling glass-carrier use and shifting its shipment mix toward higher-value HBM4/HBM4E parts. Overall HBM wafer input is projected to rise to about 250,000 wafers per month next year from roughly 180,000 this year, while outsourced glass-carrier cleaning is set to increase to 50,000 sheets a month from 20,000 this year (10,000 last year). That 2.5-fold jump in carrier demand, together with changing product mix, means HBM4 and HBM4E will grow from about 40% of shipments this year to roughly 80% next year, producing at least a twofold increase in HBM4-family output.

The scale-up centers on 12-layer and higher stacks, where wafer-thinning and warpage control become critical; glass carriers are temporarily bonded to the wafer underside to prevent bending or cracking during grinding and drilling. Samsung already began mass shipments of sixth-generation HBM4 (10nm-class logic with a 4nm base die) and has provided 12-layer HBM4E samples to customers including Nvidia. With sixth- and seventh-generation chips slated to make up the bulk of next year’s volume, the push targets AI accelerator demand and positions HBM4 as Samsung’s core high-value memory product.