Enterprise AI infrastructure configurator
Design your private AI infrastructure
Step 1 of 11
1 · Organisation size
How many employee accounts will the environment serve? This sets registered capacity; simultaneous users are estimated separately at the next steps.
AI capacity dashboard
Meets workload| Employee accounts | 5,000 |
|---|---|
| Recommended concurrent users | 400 (8% at peak) |
| GPUs | 16 × NVIDIA H200 NVL |
| GPU memory | 2256 GB aggregate |
| Supported models | Llama 3.2 3B, Llama 3.1 8B, Gemma 3 12B, Qwen2.5 14B, Qwen2.5 32B, Llama 3.3 70B, gpt-oss 120B (MoE, 5B active), Llama 3.1 405B |
| Estimated throughput | 30,778 tokens/s (theoretical max 42,578) |
| Daily token capacity | 598.32 M vs demand 35 M generated |
| Monthly token capacity | 13.16 bn |
| Deployment | On-Premises Private AI Data Centre |
Illustrative until validated by benchmarking: Llama 3.3 70B, Int4, 8,192-token context, batching to 64 concurrent sequences.