Yes, you can start small and scale when needed. Our configurations are designed as modular building blocks: from the entry-level Server S1 up to the AI data center.
Our configurations
- S1: The entry point for small teams
- S2: Enterprise AI for your whole team (2x NVIDIA RTX PRO 6000 Blackwell, 192 GB VRAM)
- DC4 and DC8: Data center configurations for high load and many users
Capacity grows in 2-GPU steps. Each configuration can run on its own or be combined with others, without changing your applications. Details on the server page.
The software makes scaling easy
The onprem.ai software, the operating system for enterprise AI in your own data center, manages all servers centrally:
- Add multiple servers: configurations combine easily
- Load balancing: load is distributed automatically
- Central management: control all systems from one interface
- No data migration: new servers are added, existing integrations stay unchanged
Scaling scenarios
Example 1: Start with S1 for a pilot project, later expand to S2 for the whole team
Example 2: Two sites with different needs: one server each, managed centrally together
Example 3: Engineering team with high token consumption: go straight to S2 or a DC configuration
Which configuration fits your team is shown by the cost calculator: it derives the GPU demand from team size and usage.
Next steps
- See the servers and choose a configuration
- Open the cost calculator and determine your demand
- Contact us for an individual scaling strategy
Sources and further information:
- How much does it cost to run AI yourself? (pricing model)
- On-premise AI for SMEs (hardware recommendations)