Inspect the rack.
Our own hardware, our own stack. This is a real Kepler rack — Supermicro nodes for control plane, compute, storage and (soon) GPU. Click a hotspot to zoom in and see what each layer does.
A Planet, down to the metal.
Every layer in the rack maps to a part of the platform. We own the servers and the stack; the renewable power and cooling belong to the colo facility in Falkenberg.
Control plane
Runs the Kepler layer: Laravel/Octane, Temporal, OpenBao, Logto. HA with quorum.
Nova hypervisors
AMD EPYC + NVMe. Your VMs run here, isolated per tenant.
Cinder block
NVMe block volumes, replicated. Snapshots and encryption at rest.
Roadmap
Reserved capacity for AI/ML — launching after V1. Shown as coming.
6× L40S already in the rack.
The GPU fleet is live and bookable today — single-tenant, on EU soil. The whole card is yours, never shared.
Want to see it for real?
For customers and procurements we book site visits in Falkenberg. Bring the team.