Chinese MRAM chipmaker ICY Tech rethinks HBM for rack-scale AI inference
DIGITIMESen

China-based ICY Tech has introduced its uHBM and uLPU AI inference architectures, extending persistent MRAM computing from validation silicon into a product roadmap spanning compute-memory dies, chips, modules, 2U trays, and racks.
This is a short summary published by AI Global Wire. The full article is owned and hosted by DIGITIMES — open it there to read it in full.
Read the full story at DIGITIMESRelated AI news
- Sources: Cognition is raising around $1B at a ~$47B valuation, up from $26B in May; Cognition received nearly $10B in investor interest (Bloomberg)Techmeme · September 2, 2026
- 戴爾 AI 伺服器績效超優+調高全年財測 盤後飆逾 7%TechNews (TW) · September 2, 2026
- US delivery firm OneRail launches Nvidia-powered AI platformTech in Asia · September 2, 2026
- Palo Alto 財報優 看好 AI 帶動資安需求升溫TechNews (TW) · September 2, 2026
- UMC targets 12-inch TFLN to break 400G SiPh bottleneckDIGITIMES · September 2, 2026
- Apple's M6 and M5 Ultra break rule tying process to chip strengthDIGITIMES · September 2, 2026