SEMIFIVE, a South Korean custom semiconductor company, has announced the start of mass production for Bertha, a data center AI inference accelerator developed for fellow South Korean firm HyperAccel. The announcement was made on September 8, 2026.
Manufacturing on Samsung Foundry's 4nm Process
Bertha is being manufactured using Samsung Foundry’s advanced 4nm process node. This marks SEMIFIVE’s first large-scale mass production project on this cutting-edge technology, representing a significant milestone for the company.
Production Volumes Expected to Grow
Following the initial production contract with HyperAccel, SEMIFIVE projects that mass production volumes will expand steadily through subsequent purchase orders. This indicates a strong and ongoing demand for the AI inference chip in data center applications.
