QDNA, a research notebook on on-premise AI servers and sovereign, local, hosted or hybrid platforms
Guidance for choosing hardware, open models and architecture, then framing compliance.
Sovereign AI platform and organisational digital twin
The harness, memory and agents your teams work with. This layer runs on the hardware platform below.
- Agent harness
- Memory
- Vector databases
- Agents
- Assisted coding
- Tests
- Services
- Business roles
- Organisational digital twin
Hardware platform
From DGX Spark to GB300 NVL72: size, choose, price.
- DGX Spark
- DGX Station
- HGX B300
- GB300 NVL72
Image: NVIDIA
Sites published by QDNA
- gironde.qdna.fr Family outings, weather and air quality
- mmr.qdna.fr MMR 2035 dashboard
- apprendre-ia.qdna.fr Free AI course
- media.qdna.fr HLS video streaming (external site)
Reference sheets and guides
- AI server prices Guide by tier, 2026
- DGX Station Hardware sheet
- GB300 NVL72 Hardware sheet
Latest articles
- Tutorial: operating a digital organisational twin Published on
- LLM inference engines compared in 2026 Published on
- Sovereign AI inference: 8 × RTX PRO 6000 vs 8 × H200 SXM Published on