Member of Inception Program

How much does it cost to run AI yourself?

The costs consist of three predictable building blocks: hardware, license, and adoption. You pay for your infrastructure, not for tokens, API calls, or user counts.

The three building blocks

1. Hardware (one-time or monthly)

You buy the servers or rent them. When you buy, the hardware is yours: it retains a residual value and can later be sold or repurposed. Entry starts with the Server S1 and grows in 2-GPU steps up to the AI data center (DC8). Important: GPU prices are currently volatile, you will receive binding figures in your quote.

2. License (yearly per GPU)

The software license covers security updates, new features, and continuously optimized AI models, optionally with remote monitoring and support. After the first year you decide whether to renew, your infrastructure stays operational either way.

3. Adoption (time and materials)

Integration, custom development, consulting, and training are billed by time and effort.

Prefer renting over buying?

With Managed On-Premises you get hardware, operations, and monitoring from a single source, from €1,490 per GPU per month. The servers sit in your infrastructure, we handle the technical operation remotely.

Run your own scenario

How on-premise compares to cloud GPUs over three years depends on your team size and usage. Our interactive cost calculator shows total costs, break-even, and the residual value of the hardware:

Open the cost calculator

A note on cloud costs

Cloud prices are not guaranteed to stay stable. Token models, limits, and model upgrades tend to drive costs up over time. On-premise is predictable, independent, and not tied to users or tokens.

Next steps