Micron unveils EIM memory for physical AI, CXMT details 4F² DRAM plans

Micron presented its Edge Inference Memory (EIM), which stacks LPDDR atop a logic die to cut latency for physical AI, at SEDEX 2026 in Seoul, targeting mass production in the second half of 2028. China's CXMT reported progress on its 4F² architecture and SAQP process and forecast 725–741% year-on-year revenue growth for the first three quarters.

• EIM stacks LPDDR on a logic die to cut latency between AI chips and memory
• Micron targets mass production of EIM in the second half of 2028
• CXMT adopted SAQP and advanced its 4F² architecture, shipping LPDDR6 in August
• CXMT expects 725–741% year-on-year revenue growth for Q1–Q3