Samsung is reportedly planning to use its 2 nm process for the base die of HBM4E, its 7th-generation high-bandwidth memory. This move comes just a month after shipping the industry's first commercial HBM4 and represents a significant leap in memory technology. The base die, which previously handled only power and signal control, now performs compute tasks directly, making the process node crucial.

For HBM4, Samsung already led with 4 nm logic base dies from its foundry, outpacing SK hynix's 12 nm process sourced from TSMC. Moving to 2 nm for HBM4E would further extend this advantage, improving power efficiency, thermal management, and area utilization. The redesign of the HBM4E power delivery network to handle increased power bumps from 13,682 to 14,457 within the same footprint demonstrates the complexity of this advancement.

Standard HBM4E is expected mid-year, with custom products following in the second half. This push also benefits Samsung Foundry's fab utilization, particularly at the new Taylor facility in Texas where equipment installation is underway. With TSMC planning 3 nm for custom HBM4E and SK hynix developing its own variant, Samsung's 2 nm move positions it to maintain its competitive edge in the high-bandwidth memory market.