Scaleway vs Hetzner: Cloud-native services vs raw compute value

Buy services, or buy cores and build the services yourself?

Every infrastructure budget is really two budgets: the invoice from the provider and the salary cost of the people who make the invoice work. Scaleway and Hetzner sit at opposite ends of that equation. Scaleway sells a catalogue of managed pieces, a container registry, a metrics and logs service, serverless runtimes, databases and clusters, so that a small team can assemble a production system without operating any of them. Hetzner sells compute, storage and networking at a price that is hard to argue with, and leaves everything above the operating system to you.

Framed that way, the comparison stops being about which provider is better and becomes a question about your team. A group with a platform engineer who enjoys running Prometheus, PostgreSQL and a Kubernetes control plane will extract more value from Hetzner than almost anywhere else. A group of application developers with no appetite for that work will spend the savings twice over in lost time.

The sections below test that principle against the specifics: what each platform contains, how it behaves, and what each one asks of you. Accounts for both are available through AllCloudShop, on the Scaleway account page and the Hetzner account page.

Three questions that settle it quickly

Who runs your database? Scaleway will, through managed PostgreSQL, MySQL or Redis-compatible caching, with backups and failover handled for you. Hetzner will not, so the answer becomes a server, an installation, a replication decision and a backup routine that somebody has to verify. The managed against self-hosted analysis is the honest accounting of what that costs.

Where do your metrics and logs go? Scaleway offers Cockpit as a managed destination, so a cluster can ship telemetry somewhere without a dedicated observability stack. On Hetzner you build that stack yourself, which is a well-trodden path but one that consumes memory, disk and attention. The monitoring guide describes the components involved.

Do you want to deploy without a server at all? Scaleway's Serverless Functions, Containers and Jobs give you a genuine option to skip machine management for suitable workloads. Hetzner has no equivalent, so every deployment lands on something you provisioned.

If your answers are Scaleway, Scaleway and yes, the decision is already made and the price difference is the cost of that convenience. If your answers are that you would rather run these things yourself, Hetzner gives you more hardware to run them on for the same money.

Platform reference

The money and the paperwork

Start with the obvious. For the same monthly spend, Hetzner puts more cores, more memory and more disk at your disposal than Scaleway does, and it is not a close contest. That advantage compounds for workloads that are simply large: batch processing, build farms, video encoding, self-hosted analytics, anything where the bill is dominated by raw capacity rather than by the number of distinct services.

Now the offset. Every managed service you would otherwise run yourself represents labour, and labour on an infrastructure team is rarely cheap. A managed Kubernetes control plane, a managed database with automated backups and a managed telemetry pipeline can easily account for a day a week of someone's attention when self-hosted, and that day has a price. The arithmetic only favours Hetzner when you have the skills already or when the workload is big enough that hardware dominates.

Governance is the second commercial axis. Scaleway's identity model separates an organisation from its projects and issues credentials to applications with policies attached, which matters when several teams, contractors or clients share one account. Hetzner scopes tokens to projects and supports team members, which is clean and sufficient for a single organisation but less expressive when access needs to be sliced finely. The IAM fundamentals article explains what to look for.

Both are European companies with European facilities, so neither creates a jurisdictional problem for an EU buyer. Scaleway leans into that positioning more actively in its marketing; Hetzner simply is a German company running German and Finnish capacity.

What the workflow looks like

A Scaleway developer typically works from a project. The scw command line tool and the Terraform provider cover every product with a consistent shape, images can be pushed to the Container Registry, a Kapsule cluster can pull them over a Private Network, secrets come from Secret Manager, and telemetry lands in Cockpit. It is a complete loop inside one vendor, which is exactly what teams building product software want.

A Hetzner developer works from an API token and a very good set of primitives. hcloud and Terraform create servers, volumes, networks, firewalls, load balancers and placement groups, and cloud-init brings each machine up configured. The pipeline pieces that Scaleway supplies as products, a registry, a metrics store, a secrets service, are chosen from the open source ecosystem and run on machines you own. Many teams prefer this, because it is portable and because nothing surprises them at renewal time.

Where the difference becomes concrete is in a new project's first fortnight. On Scaleway, an application with a database, a queue, a cluster and dashboards can be running in days. On Hetzner the same application takes longer to stand up and then costs less to run indefinitely. Neither is wrong; they pay off on different timescales. Official references are at the Scaleway documentation and the Hetzner Cloud documentation.

What sits underneath

Hetzner's infrastructure advantage comes from owning the buildings. Data centre parks built for its own use, standardised hardware bought in volume and a network engineered around its customers' traffic are what make the pricing sustainable rather than promotional. Cloud servers carry local NVMe, Cloud Volumes add attachable capacity, and Storage Box provides cheap space for backups over familiar protocols. Instances in one location share a fast internal network that teams routinely use for clustering.

Scaleway's infrastructure is more varied in character than in scale. Multiple availability zones in its main region allow a workload to survive the loss of one facility without leaving the country, which is a real architectural capability rather than a marketing line. Elastic Metal makes physical machines available by the hour, Arm instances offer an efficiency option, and the Apple silicon servers cover mobile build pipelines that almost nobody else in Europe serves.

Networking is comparable in the basics and different at the edges. Both offer private networks, firewalls, floating addresses and IPv6. Scaleway adds Public Gateways for network address translation and a global IP model that lets addresses move between resources, and its load balancer supports layer 7 rules. Hetzner's load balancer handles the common cases well and its private networks carry user-defined routes. The load balancer explainer covers what the layer distinction means in practice.

Each at its best

Scaleway

Scaleway is at its best when a small team needs a complete production environment and has no interest in operating infrastructure software. Managed Kubernetes, managed databases, a registry, secrets, observability and serverless runtimes under one identity model means the team's time goes into the product instead of into the platform. Multi-zone object storage and a multi-zone region give the resilience story a factual basis.

The hardware oddities are a genuine bonus: hourly bare metal for burst workloads and Apple silicon for iOS builds are capabilities that would otherwise require separate vendors.

Hetzner

Hetzner is at its best when the workload is large, long-lived and understood. Dedicated vCPU instances behave consistently, Arm64 machines stretch the budget further for compatible software, and the ability to move onto physical hardware keeps the growth path inside one relationship. For teams that already have their own deployment tooling, the platform adds nothing they do not need and removes a great deal of cost.

There is also value in stability. A short catalogue that changes slowly is easy to automate once and rely on for years, which suits infrastructure intended to outlast the current project.

Where each will frustrate you

Scaleway

Every region is in Europe, so a product with users in the Americas or Asia will eventually need a second provider. Services have matured at different rates, so it is worth reading the documentation for each one you plan to depend on rather than assuming uniform polish. Quotas on new accounts start conservatively and are raised on request. And the managed convenience is priced accordingly, which shows up clearly when the workload is mostly raw capacity.

Hetzner

Nothing above the operating system is provided. No managed Kubernetes, no managed database, no registry, no observability service, no serverless option and no GPUs means your team owns all of it permanently, including the parts nobody enjoys. Access control is less granular than a project-policy model. The location list outside Europe is short, and new accounts begin with modest resource ceilings that grow as the account establishes a history.

Which suits your team

Choose Scaleway if your engineers are application developers rather than infrastructure specialists, if you want Kubernetes, databases, observability and serverless from one vendor with a coherent identity model, or if multi-zone resilience inside Europe is a requirement. Product companies and startups building for a European market fit naturally. The Scaleway account listing shows the variants available.

Choose Hetzner if compute volume dominates your bill, if your team already runs its own platform software comfortably, or if you need a path from virtual machines onto physical hardware without changing vendor. Data processing, continuous integration, research computing and cost-driven software teams do best here; see the Hetzner account listing.

If the decision is still finely balanced, DigitalOcean against Scaleway compares Scaleway with a larger developer cloud, Hetzner against Linode tests the value argument against a managed platform, and OVHcloud against Scaleway covers the French pairing.

Common questions