Enterprise product packages
Predefined packages, calculated from real hardware
Each package shows its actual hardware configuration and the employee capacity, concurrency and AI capacity derived from it. Capacities are computed by the same estimator used in the designer, never typed in as marketing numbers.
Private AI Starter
25–100 employees
A private assistant and knowledge search for small and medium organisations.
15
Concurrent users
at peak
- Hardware
- 2 × GeForce RTX 5090 (64 GB) in 1 × AI Workstation Tower
- Reference model
- Llama 3.1 8B at Int4
- Daily token capacity
- 85.51 M expected · 1.02 bn theoretical
- Deployment
- On-Premises Private AI Data Centre
- Facilities
- 2.1 kW IT load · 10 kVA online double-conversion UPS · Precision air conditioning unit, 10 kW
- Monitoring
- Remote
- High-end consumer GPU option
- Private AI assistant
- Internal knowledge search
- Basic employee access management
- Remote monitoring
$139,640 one-off incl. deployment & tax
then $0.00/year recurring
Start from thisConsumer GPUs have no ECC memory and consumer warranty terms; suitable for cost-sensitive deployments with tolerant workloads.
Private AI Professional
100–500 employees
Multiple models and document intelligence for growing organisations.
50
Concurrent users
at peak
- Hardware
- 4 × RTX PRO 6000 Blackwell Server Edition (384 GB) in 1 × 4U 8-GPU PCIe Server
- Reference model
- Qwen2.5 32B at Int4
- Daily token capacity
- 104.16 M expected · 600.55 M theoretical
- Deployment
- On-Premises Private AI Data Centre
- Facilities
- 4.5 kW IT load · 10 kVA online double-conversion UPS · Precision air conditioning unit, 10 kW
- Monitoring
- Remote
- Multiple GPU configuration
- More concurrent employees
- Internal document intelligence
- Multiple AI models
- Advanced monitoring
- Optional physical monitoring station
$300,944 one-off incl. deployment & tax
then $9,600/year recurring
Start from thisPrivate AI Enterprise
500–2,500 employees
Dedicated, highly available AI infrastructure for large organisations.
120
Concurrent users
at peak
- Hardware
- 8 × NVIDIA H200 NVL (1128 GB) in 2 × 4U 8-GPU PCIe Server
- Reference model
- Llama 3.3 70B at Int4
- Daily token capacity
- 303.76 M expected · 1.84 bn theoretical
- Deployment
- On-Premises Private AI Data Centre · high availability
- Facilities
- 8.3 kW IT load · 20 kVA online UPS with extended battery · Precision air conditioning unit, 10 kW
- Monitoring
- Remote + Physical Monitoring Station
- High-performance multi-GPU servers
- Dedicated AI infrastructure
- Multiple departments
- Advanced access controls
- Multi-model inference
- High availability architecture
- Physical and remote monitoring
$940,154 one-off incl. deployment & tax
then $24,000/year recurring
Start from thisPrivate AI Data Centre · Custom
2,500+ employees or specialised workloads
Fully customised, multi-rack, multi-site architecture.
400
Concurrent users
at peak
- Hardware
- 16 × NVIDIA H200 NVL (2256 GB) in 2 × 4U 8-GPU PCIe Server
- Reference model
- Llama 3.3 70B at Int4
- Daily token capacity
- 598.32 M expected · 3.68 bn theoretical
- Deployment
- On-Premises Private AI Data Centre · high availability
- Facilities
- 13.1 kW IT load · 20 kVA online UPS with extended battery · Precision air conditioning unit, 20 kW
- Monitoring
- Remote + Physical Monitoring Station
- Fully customised architecture
- Multiple racks
- Enterprise GPU infrastructure
- Dedicated networking and storage
- Redundant power and cooling
- Multi-site deployment
- Custom security architecture
$1,630,844 one-off incl. deployment & tax
then $24,000/year recurring
Start from this
Assumptions: 20 prompts per employee per day, 600 input + 350 output tokens, 8k context, working-day utilisation 60%. Region US; enterprise hardware multiplier 2×.