Hypercube
Sovereign compute estate
| Family | Reference Envelope | Best Fit |
|---|---|---|
| FamilyHYPERQ Nano Ultra | Reference EnvelopeDevice class | Best FitLocal, mobile and disconnected AI |
| FamilyHYPERQ Nano | Reference Envelope2 to 8 accelerators, 5 to 18 kW | Best FitVision, robotics, secure RAG |
| FamilyHYPERQ Micro | Reference Envelope8 to 16 accelerators, 20 to 60 kW | Best FitPrivate inference and fine tuning |
| FamilyHYPERQ Compact | Reference EnvelopeUp to 72 accelerators, 0.12 to 0.5 MW | Best FitRegional and industrial infrastructure |
| FamilyHYPERCUBE Nano and Core | Reference EnvelopeSub 1 MW; 280 to 350 kW | Best FitEnterprise and regulated private AI |
| FamilyHYPERCUBE Prime | Reference EnvelopeAbout 400 kW class | Best FitProduction AI and institutional services |
| FamilyHYPERCUBE Sovereign | Reference EnvelopeAbout 216 devices, 420 to 480 kW | Best FitGovernment and national platforms |
| FamilyHYPERCUBE SuperPod and Titan | Reference Envelope1.8 MW to 20 MW+ | Best FitSupercomputer to campus scale AI |
Source: HYPERCUBE Product Catalogue 2026, reference envelopes and concept visualisations. Reference configurations are engineering illustrations and do not constitute an offer. Accelerator quantities, memory, performance, power, cooling, certifications and availability depend on final engineering, silicon generation, software stack, ambient conditions, facility architecture and country specific approvals. Final configuration is established through signed engineering.