[Huawei launches OceanStor M900 AI memory storage] "Science and Technology Innovation Board Daily" reported on the 17th that at today's Huawei Full Connection Conference 2026, Huawei launched an AI memory storage-OceanStor M900 based on Lingqu. Agents and long sequences require a multi-layer KV Cache architecture. For the Agent inference system, Huawei has proposed a multi-layer KV cache architecture and launched Huawei's L3.5 layer PB-level KV cache based on Lingqu UnifiedBus and supporting "one-hop direct connection" - OceanStor M900 cluster. At the same time, M900 uses mixed media and optimized retention algorithms, which can increase SSD read and write life by 16 times, ensuring KV hits and long-term stability from the bottom up. (Kechuangban Daily reporter Huang Xinyi)