The question hanging over the Gangnam gathering was straightforward: why was Nvidia’s largest supplier of high-bandwidth memory (HBM) for AI chips, absent?