HP ZGX Fury hands-on: 748GB memory and GB300 make a team-sized local AI box
Sam Witteveen · youtube · 2026-10-30
Sam Witteveen reviews the HP ZGX Fury, the bigger sibling of the DGX Spark (ZGX Nano). Specs include NVIDIA's GB300 chipset, 748GB of unified memory, and 20 petaflops — pitched as a mini data center sitting next to your desk that a whole team can share for local LLM inference.
The video covers:
- How the Fury's unified memory works and what separates it from DGX Station
- The HP ZGX Runtime (ZRT) and HP Z Toolkit software stack, Red Hat Enterprise certified
- Benchmarks and target use cases: running agentic systems locally and on-prem AI
Aim at teams that need more memory than a DGX Spark but don't want their data in the cloud.
More from Infra
- AI scarcity moved to atoms: memory, GPU and wafer margins hit 58%-80% — sujingshen · 2026-10-31
- Anthropic's S-1 risk factors: compute tops the list at 13.4%, edging out regulation — ayushtweetshere · 2026-10-31
- TU Eindhoven shows glass interposer with BGA below and MPO optical port on top — jwt0625 · 2026-10-31
- AI Boom Triggers Massive PCB Shortage With Orders Booked Until 2030 — chemist_slime · 2026-10-31
- Beff Jezos doubles down on Cerebras: GPU backends feel 'stupid' for frontier-scale fast inference — beffjezos · 2026-10-31
- Intel Arc B580/B70 ComfyUI Benchmarks: omnixpu Beats Native pytorchxpu — Willing-Grade5566 · 2026-10-31