May. 12 at 9:52 PM
The Five Layers of Memory
1. Near Memory:
Sk hynix, Samsung ,
$MU
2. Main Memory:
Sk hynix, Samsung, CMXT,
$MU
3. Expansion Memory:
SK hynix, Samsung,
$MU, ALAB, MRVL ,MCHP ,RMBS
4. Contexted Memory:
SK hynix, Samsung, Kioxia,
$MU ,
$WDC ,
$SNDK,
$SIMO
Contexted Memory is the "Long-term Memory" switch for AI! 🚀 As long-context models explode in 2026,
$SNDK and
$WDC are reshaping the hierarchy with high-speed NAND. Notably,
$SIMO’s controller tech is vital for LLMs to retrieve history in milliseconds.
While
$MU dominates HBM, the cost-efficiency and capacity at this layer offer more "alpha" for Edge AI. Who’s your pick for the AI storage silver medalist? The surging
$SNDK or the controller king
$SIMO? Let’s debate below! 👇
