Specifications
Vendor
SK hynix
Product
SOCAMM2
Type
Small Outline Compression Attached Memory Module
Memory Technology
LPDDR5X-class low-power DRAM
Target Platform
AI servers and accelerator platforms
Form Factor
Compact module mounted lying flat against the board
Advantage vs RDIMM
Higher capacity per unit volume and lower power per bit
Cooling
Improved thermal path versus vertical DIMM sockets
Serviceability
Replaceable module rather than soldered-down memory
Status
Previewed / sampling with AI server platforms
Application
AI inference and training servers, edge AI appliances
Why AI servers are moving off vertical DIMMs
SOCAMM exists because the vertical RDIMM socket has become a mechanical problem in dense AI servers. A DIMM stands perpendicular to the board, which consumes vertical air space exactly where the accelerator cold plates want it, and it forces the memory subsystem to run at full DDR5 signalling power. SOCAMM lays the module flat instead, freeing the airflow path and letting the platform use LPDDR — which delivers more bandwidth per watt than registered DDR5 at the capacities AI servers need.
SOCAMM2 is the second generation of that idea and the one SK hynix showed alongside its 16-layer HBM4, SOCAMM2 and LPDDR6 roadmap preview. The serviceability argument matters as much as the density: soldered LPDDR would be cheaper and denser still, but a field-failed module could not be replaced without a board swap. SOCAMM keeps memory field-replaceable while capturing most of the power and volume benefit of embedded LPDDR.
SK hynix products are supplied by QS Compute with B2B pricing, global shipping and short sample lead times for AI data center, edge deployment and industrial integration projects.
Request a Quote — SK hynix SOCAMM2 AI Server Memory Module — Low-Power LPDDR5X Form Factor (SK hynix SOCAMM2)
QS Compute — Global B2B AI computing hardware supply, 15-day sample lead time.
Get Your Quote →