Two European answers to Kubernetes on Hetzner
Hetzner gives you fast, low-cost servers and leaves Kubernetes to you. Cloudfleet and Ankra are both European companies that close that gap, and both run their control planes in Europe.
Cloudfleet runs the Kubernetes control plane for you. Its nodes come up automatically on Hetzner, AWS and GCP, and any Linux server, from Proxmox to a dedicated machine in your rack, can join the same cluster. It is a clean, focused product with clear pricing.
Ankra builds the whole cluster on your servers and then runs what lives on it. Applications go from repository to release, platform add-ons stay current, vulnerabilities are ranked, cost is estimated, data is backed up and the AI helps operate it.
Cloudfleet gives you a well-run cluster. Ankra gives you the cluster and the platform work on top of it.
How the cluster is built
Cloudfleet hosts the control plane in a region you pick when you create the cluster. On the Basic plan it is shared with other customers, and Pro adds high availability with a 99.95% SLA. Workers are auto-provisioned on Hetzner, AWS and GCP based on demand. UpCloud, Exoscale, Scaleway and OVH were in private preview in May 2026. Mixing clouds and on-premises servers in one cluster is its standout feature.
Ankra builds kubeadm or k3s on Hetzner, with control plane, workers, private network and bastion in your own project. The same applies on OVHcloud, UpCloud, DigitalOcean, AWS EC2 and Proxmox VE. Ankra handles upgrades, node groups and power schedules, with node autoscaling on the five public clouds, and can stop a cluster outside working hours to save compute. It also creates EKS, GKE, AKS and three more provider services, and imports any conformant cluster. One Ankra cluster stays within one provider, except where Cluster Mesh connects UpCloud and Proxmox clusters.
What runs on the cluster
Cloudfleet stops at a running cluster with a registry and a chart marketplace. You connect your own CI, and choose and operate your own GitOps, scanning and backup tools.
Ankra builds that work in.
- Applications and Ankra Pipelines test, build and scan your code with Semgrep, Checkov and Trivy in your own cluster, and publish only the image that passed the gate
- pull-request previews give each change its own environment
- stacks install Helm add-ons and manifests in dependency order, and stack profiles version them across environments
- the Security Center ranks running-workload vulnerabilities by CISA KEV and EPSS
- backups write volumes and databases to a bucket you own
Applications and previews work with GitHub repositories, and some documented pipeline stage kinds do not run yet.
Cost and AI
Both products help with cost. Cloudfleet offers cost reports with one-click optimisations. Ankra’s Cloud Cost estimates spend per namespace and stack and reprices the whole cluster on every provider Ankra can build on.
Cloudfleet’s CLI includes an MCP server that reads cluster data and handles support tickets but cannot change the cluster. Ankra’s AI analyses alerts with AI Insights, reviews pull requests and opens fix pull requests through the AI Gateway, and drafts stacks from a description. Every write waits for a person by default.
Neither vendor holds a completed SOC 2 or ISO 27001 audit yet. Cloudfleet lists both as in progress, and Ankra’s status is on its trust page.
When to choose Ankra, and when to choose Cloudfleet
Choose Ankra when you want Kubernetes on Hetzner or another European provider and also want delivery, add-ons, security ranking, backups and AI operations handled, not just the cluster.
Choose Cloudfleet when
- you want a managed control plane and to run everything on top yourself
- one cluster has to span several clouds and your own servers
- you want automatic node provisioning on Hetzner, AWS and GCP from a single cluster
Compare the work that remains
- Create the foundation. Note where the control plane runs and who upgrades it.
- Ship a release. Include build, tests, scans, a preview and a rollback, and list every tool you add.
- Add a platform component. Install an ingress controller or operator and note who keeps it current.
- Investigate a vulnerability. Find every affected workload.
- Compare total cost. Cloudfleet prices per cluster plus vCPU; Ankra prices per worker vCPU with the first 30 free. Add the tools and hours each route needs on top.
See Cloudfleet’s pricing and Ankra’s pricing.
Reviewed 9 October 2026 against Cloudfleet’s documentation, pricing and product updates to July 2026. Product scope and plans can change.