AMD and Qualcomm test SOCAMM to solve AI memory issues

SOCAMM memory is breaking out of NVIDIA orbit as AMD and Qualcomm chase fixes for AI memory choke points.

SOCAMM stops being NVIDIA-only

• SOCAMM starts life as a NVIDIA-focused memory play.
• AMD and Qualcomm start circling the same tech.
• Agentic AI workloads push memory limits hard.

What SOCAMM actually is

• The design pulls from LPDDR-style DRAM.
• The modules stay removable, not soldered down.
• It complements HBM rather than replacing it.

Why AI systems care

• Agentic AI needs fast short-term memory.
• SOCAMM allows massive per-CPU capacity.
• Millions of active tokens stay resident.

Different design path from NVIDIA

• AMD and Qualcomm test square module layouts.
• DRAM chips sit in dual-row arrangements.
• The goal is cleaner power control.

Power management changes

• PMIC shifts directly onto the SOCAMM module.
• Boards lose some power circuitry complexity.
• Stable high-speed operation becomes easier.

Performance and efficiency tradeoffs

• Bandwidth trails behind HBM.
• Power draw stays relatively low.
• The balance still works for memory-heavy tasks.

What comes next

• NVIDIA plans SOCAMM 2 with Vera Rubin clusters.
• AMD and Qualcomm evaluate future AI racks.
• Wider adoption looks likely if tests hold.