Samsung Foundry Offers 5nm Process as 4nm Capacity Reaches Full Load
Samsung Foundry's 4nm production capacity is reportedly fully loaded due to surging orders for HBM4 base dies and Groq LPUs. In response, the company is offering its 5nm process as an alternative for chip design clients seeking advanced node capacity. This highlights the intense demand for AI hardware and high-bandwidth memory, which is straining advanced semiconductor manufacturing capacity. It also shows how foundries are repurposing older but related nodes like 5nm to alleviate supply chain bottlenecks. Since Samsung's 4nm and 5nm belong to the same node family, much of the 5nm capacity had previously migrated to 4nm. However, Samsung retained the SF5A (5nm) process for automotive clients and is now expanding its use to AI and HPC applications.
## BACKGROUND
High Bandwidth Memory (HBM) is a specialized 3D-stacked DRAM interface used in high-performance AI accelerators, where the "base die" acts as the foundational control layer. Groq is an AI chip startup known for its Language Processing Unit (LPU) architecture, which requires advanced foundry nodes for manufacturing. Samsung's SF5A is a specialized automotive-grade 5nm process designed for high reliability.