Virtual machines
Linux and Windows VMs on KVM, with a browser console and ready-made OS images.
Cloud repatriation
Leave the public cloud bill.
Keep the cloud.
Move steady workloads off AWS, Azure or Google Cloud to servers you own or rent. Your teams keep self-service VMs, Kubernetes, networks and storage, from a console, a REST API and Terraform.
Start with one server and run hybrid for as long as you need. Sign up with Google, or talk to us first.
- Servers you own or rent
- Self-service for your teams
- Console, API and Terraform
- Built and operated in the EU
Why repatriate
Public cloud is great for spikes. Steady workloads pay for it every hour.
Databases, build farms, internal tools and other always-on services run around the clock. On rented hyperscaler capacity you pay for every one of those hours, and again for data leaving the network.
On servers you own or rent, the same workloads cost what the hardware costs. CloudOps keeps the part teams like about the cloud: asking for a VM and getting it, without a ticket.
What maps where
The services you use today, on your own hardware.
Block volumes
ZFS-backed volumes you attach to VMs, with snapshots.
S3-compatible object storage
Buckets and access keys that work with the S3 tools and SDKs you already use.
Private networks and security groups
Isolated networks, firewall rules per VM, an encrypted overlay between hosts and site-to-site VPN.
Kubernetes
Kubernetes clusters on your hosts, created in the console or imported, with nodes you can add and remove.
Load balancers and databases
Load balancers in front of your VMs, and managed database instances in preview.
How a move works
Move one workload at a time.
Add your servers
Install hosts over the network with PXE, connect Linux servers you already run with a one-line import, or rent bare metal.
Recreate the environment
Define VMs, volumes, networks and security groups in Terraform or the console, next to what runs in the public cloud today.
Move the data and cut over
Copy the data across a site-to-site VPN, test, switch the traffic and shut the cloud instances down.
Keep elastic or bursty workloads where they are. Running hybrid is a normal end state, not a failure.
AI-ready, with guardrails
Your AI agents can work on it too.
CloudOps has a built-in MCP server. Connect Claude or another MCP assistant and it can read your servers, networks, storage and jobs and, if you allow it, create VMs and volumes.
Access stays off until an owner turns it on. You choose read only or safe changes, every tool call is recorded, and an agent can delete only what it created.
Questions before you move
What to know before you move.
Do my teams lose self-service?
No. They create VMs, volumes, networks and Kubernetes clusters from the console, the REST API or Terraform, and every change runs as a tracked job you can audit.
Do I have to buy hardware?
No. Use servers you already have in your racks or a colocation facility, or rent bare metal from independent operators on the CloudOps marketplace, which is in early access.
Where does my data live?
On your servers. We host the control plane in the EU and operate under the GDPR; a US control plane is available on request.
Can I start small?
Yes. Start with one server and one workload. Keep everything else in the public cloud and connect the two with site-to-site VPN while you move.
Plan your first repatriated workload.
Pick one steady service, put it on one server, and compare.