When a company runs its models, training data and business data on an outside compute platform, the first question is often not price but this: how will my data be treated? Qovane Group's answer is clear. Customer data belongs to the customer; we are the party that provides the compute.
Principle one: ownership and control stay with the customer
The data customers upload, the model weights they train and the outputs they generate all belong to them. When a service ends, customers can export and delete their data as their contract provides. We do not claim rights in that data for any reason, and we do not sell it or share it with third parties.
Principle two: never used to train our models
This is a commitment we make to every customer. Qovane Group develops its own AI software and content generation capabilities, but our model training is kept entirely separate from customer data. Customer data is never used to train any Qovane model, and it never enters our Studios production pipeline.
Principle three: isolation by default, minimal data
Technically, isolation is the default rather than an option:
- Bare-metal nodes are dedicated to a single customer and never shared.
- Networks are isolated by tenant, so customers cannot see one another.
- Access follows least privilege, with multi-factor authentication and retained access logs.
On top of that, we practise data minimisation. Our core business is compute and hosting, not processing customers' personal data, so we touch as little customer data as possible. We collect only what is needed to activate service, bill and meet compliance obligations. For customers that process personal data, we can sign a data processing addendum that sets out each party's responsibilities.
Principle four: partner facilities have no access to customer data
Qovane's capacity comes from data centers operated by signed supply partners in several regions. Each partner is bound by written security requirements. Partner facilities are responsible for power, cooling, networking and physical security, but they have no access to customer data or workloads. Customers contract with Qovane, and Qovane carries the data responsibility. Whichever region's facility a workload runs in, the customer receives the same standard of data protection; it never weakens because a job is scheduled somewhere else.
Trust is the foundation of compute
Compute can be rented by the hour; trust cannot be built by the hour. Customers will only put truly important work on our platform when they are confident their data is secure, under their control and never misused.
Protecting customer data is not an add-on at Qovane Group. It is the precondition for providing compute at all.