A US push for additional Korean semiconductor manufacturing investment has entered bilateral negotiations just as the first project under South Korea's US$200 billion strategic-investment mechanism approaches selection.
Nvidia is reportedly shifting its Rubin Ultra AI accelerator toward an eight-high high-bandwidth memory (HBM) configuration, down from an earlier 12-high design, as rising memory costs put greater emphasis on bandwidth efficiency and overall system economics.
Montage Technology is accelerating its move beyond traditional DDR memory interfaces, with CXL 3.2, DDR6 and PCIe products positioning the Chinese interconnect chipmaker deeper in AI data center infrastructure.
