➀ SK Hynix introduced its in-memory computing advancements for LLM inference with AiMX-xPU and LPDDR-AiM, focusing on enhanced power efficiency and speed. ➁ The company showcased a GDDR6 Accelerator-in-Memory card utilizing Xilinx Virtex FPGAs and GDDR6 AiM package. ➂ SK Hynix aims to scale memory capacity from 32GB to 256GB per card and explore on-device AI applications.