Scaleway review

Scaleway is a hosted European cloud and AI provider. In a software stack, it sits below application code as the infrastructure layer: teams provision compute, storage, networking and managed services rather than installing Scaleway as a library or running its cloud platform on their own servers.

The portfolio spans Instances, Dedibox and Elastic Metal bare metal, Apple Mac mini options, Object Storage, Block Storage, managed databases, Multi-AZ VPC, load balancing, Managed Kubernetes and Serverless Containers. GPU infrastructure is aimed at AI workloads, including single-node and multi-node LLM training. The homepage counts more than 100 cloud and AI products and lists four Multi-AZ cloud regions in Europe.

That breadth can reduce the number of providers a team needs for common infrastructure components. Scaleway also positions its services around European data sovereignty and open, interoperable technology. Those are architectural considerations, not a substitute for checking each service’s region coverage, data handling and portability against your requirements.

Scaleway provides an API, command line interface (CLI) and developer tools for managing cloud resources. The homepage also points to technical documentation, step-by-step tutorials, code samples, developer videos, guides, a changelog and a service status page. That gives teams reference and operational material alongside the infrastructure control surface.

The homepage does not name an API protocol, SDKs or supported programming languages. Confirm SDK support and API details in Scaleway’s documentation before selecting language-specific tooling or designing an automation layer. The named integration surfaces are the API and CLI; the homepage does not describe webhooks or plugins.

For infrastructure-as-code workflows, check the current provider and resource coverage in the docs before standardizing on a workflow. The available service range includes managed Kubernetes, databases, networking and storage, but the exact configuration options and automation coverage should be confirmed for each product. Scaleway’s changelog and status page are useful places to track service changes and incidents.

Scaleway lists a free plan, but its cloud allowance, eligible products and usage limits are not stated here. Treat it as something to verify before basing a production design on free capacity. The homepage says prices are presented at product level; compute, storage, network and managed-service costs therefore need to be checked against the services and regions you plan to use.

The published figures here are for support, not infrastructure. The Basic Support Plan is €0 per month and includes 24/7 ticket support with an eight-hour initial response time and no phone hotline. The Advanced Support Plan is €50 per month; Business is €250; Enterprise is €990. Paid plans charge the greater of the minimum or a percentage of monthly net cloud spend: 5% for Advanced above €50, 10% for Business above €250 and 20% for Enterprise above €990.

Model infrastructure and support separately. Before launch, confirm service rates, free-plan limits and whether your expected net cloud spend changes the support charge.

Scaleway is a hosted cloud platform. Its homepage promotes open standards and interoperable technologies, but that positioning does not establish that Scaleway’s platform is open source or available under an open-source licence. The vendor does not publish a platform licence on the homepage. Teams should not plan to self-host Scaleway itself.

Portability depends on the services selected. Compute, containers and databases may let teams keep application components on familiar technologies, while managed services can create dependencies on provider-specific configuration and operations. A European cloud migration write-up on Reddit describes managed Kubernetes, Postgres, Redis and object storage as portable categories, while warning that services without good replacements can make a move harder.

Before adopting managed services, confirm the export path, formats, fees and restoration process for each data store. Keep infrastructure definitions and backups in a form your team can use outside the provider, and test recovery independently.

Scaleway advertises four Multi-AZ cloud regions in Europe, Multi-AZ VPC and managed services for compute, storage, databases and containers. Those are relevant building blocks for production, but a product’s availability design is not the same as an application’s recovery plan. Check service-specific availability guarantees, quotas, backup options, restore procedures and regional coverage before depending on them.

A Reddit user described Scaleway as too technical and the interface as difficult to use. Another thread includes a customer alleging that a Dedibox server was lost and that no refund was offered; commenters stressed that customers need their own backups. These are individual reports, not evidence of an overall incident rate, but they reinforce the need to test the console and keep location- and provider-independent backups.

For AI workloads, confirm GPU availability and quotas for the specific region and configuration. The homepage states that more than 5,000 GPUs are available, but that headline is not a capacity guarantee for an individual project. Review the status page and changelog as part of ongoing operations.