NVIDIA walks through five ways to tackle the AI memory wall, from NVFP4 to CMX
NVIDIA Developer · youtube · 2026-11-06
A new Inside the Architecture episode from NVIDIA Developer explains how to get more useful work from every byte of memory across the AI factory:
- NVFP4: cuts model weight memory while preserving accuracy
- NVLink + expert parallelism: distributes model weights across GPUs
- NVIDIA Dynamo: manages and reuses KV cache across inference workers
- Vera: speeds tool execution and connects host memory with Rubin GPUs
- NVIDIA CMX: reuses context across servers to reduce repeated computation
The core message: memory efficiency depends on the full system, not any single component.
Related event: NVIDIA Details Five System-Level Ways to Push Past the AI Memory Wall(3 posts)→
More from Infra
- SpaceX data center pledges 40% power discounts and billions in community benefits — Scobleizer · 2026-11-06
- Mystery seller seeks to dump $500 million of data center loans — dinabass · 2026-11-06
- Collinear AI runs full-scale agentic RL on Together Post-Training with self-onboarding — togethercompute · 2026-11-06
- SemiAnalysis: Google TPUv7 Delivers Up to 50% Better Performance-per-Dollar Than B200/B300 for Inference — vllm_project · 2026-11-06
- Meta President: new US data center build-outs will be 'very challenging' — SumitGup · 2026-11-06
- Veteran investor: memory stocks aren't cheap even at P/E of 3 as LTAs are meaningless — bookwormengr · 2026-11-06