Scaleway vs OVH
Two French providers with genuinely different product philosophies. Scaleway builds a modern developer platform; OVHcloud operates infrastructure at scale across many categories.
Scaleway
- Regions noted in source
- France and European regions
- Configurations listed
- 1
OVHcloud
- Regions noted in source
- Europe and global
- Configurations listed
- 2
Both of these companies are French, both operate their own data centres, and both market themselves to teams that want European infrastructure. That shared ground makes the differences between them easier to see, because once you remove geography from the comparison what remains is product philosophy.
OVHcloud is the larger and older business, built on dedicated servers and web hosting, now running a broad catalogue across several categories. Scaleway is smaller, newer in its current form, and has spent its energy on building a developer platform with a distinct point of view about what a cloud should feel like to use.
Philosophy before features
Scaleway behaves like a product company. The console is opinionated, the API is clean, the service list is curated rather than exhaustive, and the company has historically been willing to launch unusual hardware when it thought the idea was interesting. You get the sense of a roadmap with a thesis behind it.
OVHcloud behaves like an infrastructure operator. The priority is capacity, reliability of supply and covering the range of things customers buy, from a domain name to a hosted private cloud. Breadth wins over elegance, and that is a reasonable position for a company at its scale.
If you care most about how it feels to work with the platform day to day, Scaleway tends to win that comparison. If you care most about being able to buy whatever you need from one supplier, OVHcloud does.
The Arm question
Scaleway has a longer history of engaging with Arm based compute than most European providers, and it has been willing to put non x86 hardware in front of customers at points when that was still unusual. For teams whose workloads build and run cleanly on Arm, that interest is relevant, because power efficiency per unit of work has a way of showing up in pricing.
There is a second reason to pay attention to architecture choice, which is that it is one of the few decisions that is genuinely hard to reverse later. Provider migrations are tedious but mechanical. An architecture migration touches your build pipeline, your base images, any compiled dependency and anything that was quietly assuming x86 behaviour. Doing that work while also changing provider is not a pleasant quarter.
The caveat is the obvious one. Arm is only useful if your dependencies are available for it. Container base images, language runtimes and compiled extensions have mostly caught up, but “mostly” is doing real work in that sentence. Build your actual application for the architecture and run your test suite before you plan a migration around it.
OVHcloud’s catalogue is more conventional, with the strength being in the depth and variety of x86 hardware rather than in architectural novelty.
Bare metal and dedicated hardware
Dedicated servers are where OVHcloud’s scale is most visible. It designs and assembles its own machines, which lets it offer a very wide range of configurations and keep older generations available at lower prices. For anyone who needs physical hardware in quantity in Europe, that catalogue is one of the main reasons the company exists.
Scaleway also offers bare metal and dedicated options, with a smaller and more curated range. The experience of provisioning them is tightly integrated with the rest of its platform, which is the pattern across its whole catalogue: fewer choices, better joined up.
Deciding which model suits you is less about the vendors and more about your workload. Bare metal against virtualised compute is the place to work that out, and Vultr against OVHcloud looks at the same bare metal question from a different angle.
Sovereignty, and what the word is covering
Both providers use sovereignty as a selling point and both have a reasonable claim to it, being European companies operating European facilities under European law. The term is used loosely across the industry, though, and it is worth separating the things it can mean.
Where the data physically sits is one question. Which legal entity you contract with is another. Who operates the platform, and under which jurisdiction that operator falls, is a third. Whether the underlying software stack depends on non European vendors is a fourth. A provider can satisfy some of these and not others, and the only way to know which ones matter to you is to ask the people responsible for your compliance position.
We are not giving legal advice here. Our piece on European providers and data sovereignty lays out the questions worth taking into that conversation.
Scale, and what it buys on each side
Size is not a virtue by itself, but it has consequences and they run in both directions.
OVHcloud’s scale buys supply. When you need forty machines next quarter, a company that manufactures its own hardware and operates its own buildings is a more comfortable supplier than one that does not. Scale also buys a procurement function that knows how to answer a large customer’s questionnaire, and a support organisation with enough people in it to cover the hours.
Scaleway’s smaller size buys coherence. Fewer products means fewer inconsistencies, fewer legacy interfaces and a roadmap that can change direction without dragging a decade of commitments behind it. Smaller companies also tend to be more reachable, and the gap between filing a ticket and speaking to someone who understands the product is often shorter.
The failure mode on each side is predictable. Choosing the larger provider for a small workload means living with complexity you did not need. Choosing the smaller one for a large commitment means discovering a capacity or catalogue limit at the worst possible point. Be honest about which size of problem you actually have, now and eighteen months from now.
One more thing scale affects, which teams rarely weigh until it bites: the depth of the public record. When a large provider has an unusual failure mode, somebody has already written about it, and the search result that saves you three hours exists. With a smaller provider that corpus is thinner, and you are more often the first person to hit a given problem. That is survivable, and for some engineers it is even enjoyable, but it is a cost and it should be counted as one.
Ecosystem, tooling and exit
OVHcloud’s larger customer base means more community material, more third party integrations and more people who have already hit the problem you are about to hit. Scaleway’s ecosystem is smaller but its API design makes it straightforward to drive with standard infrastructure as code tooling.
In both cases, the practical defence against lock in is the same: keep your infrastructure defined in code, keep your data in formats you can export, and avoid building hard dependencies on any single provider’s proprietary service unless the benefit is large and conscious. The provider model in infrastructure as code covers how to structure that so switching is an exercise rather than a rewrite.
Picking one
Go with Scaleway if your team values developer experience, you want a curated platform rather than a catalogue, and the Arm or managed service direction appeals. Go with OVHcloud if you need hardware depth, want one supplier across several categories, or are buying through a procurement process that benefits from a larger vendor.
If neither is settled, look at the wider cloud service catalogue before committing, since the European field has more than two credible entries and the right answer may not be French at all.
Taking on an existing account
TrustCloudHub is independent and not affiliated with either provider. If you acquire a Scaleway account or an OVHcloud account, assume the previous holder left credentials, keys and configuration behind. Reset the password, regenerate every API token, enrol your own multi factor device, remove users you did not add and check for standing resources that are quietly billing.
Credits attached to an account are controlled by the provider, not by the seller. They expire, they commonly exclude some services, and they are only real once you have seen the balance in the provider’s own billing console after the transfer has gone through. Check there before you let the number affect your decision.