For teams that want European infrastructure, three providers dominate the conversation: OVHcloud, Hetzner and UpCloud. Unusually for cloud comparisons, they differentiate cleanly, and the decision is normally straightforward once you name your binding constraint.
The short version
| Provider | Leads on | Choose when |
|---|---|---|
| Hetzner | Price per core | Cost is the primary constraint |
| OVHcloud | Breadth and bare metal | You need range, or dedicated hardware |
| UpCloud | Storage performance | Your workload is I/O bound |
Price
Hetzner is the price leader and it is not particularly close. Because it operates its own data centres and keeps the service catalogue deliberately narrow, cost per core and per gigabyte of RAM is consistently below the alternatives. Its dedicated root servers are cheaper still for sustained workloads.
OVHcloud is competitive and offers a wider range of machine types. UpCloud prices above Hetzner, and the premium is for storage performance rather than compute.
All three are dramatically cheaper than the large platforms for straightforward virtual servers. If you are running plain compute on AWS purely out of habit, this is the comparison worth running.
Storage performance
UpCloud built its platform around consistently high IOPS, and for storage-bound workloads that is a genuine differentiator. Databases under write load, search index rebuilds and analytics scanning local data all benefit measurably.
The caveat is that most workloads are not storage bound. Stateless web servers, API workers and CPU-bound batch jobs will not notice. Measure I/O wait on your current system before paying a premium for storage you are not using.
Service breadth
OVHcloud has the widest catalogue of the three, including bare metal dedicated servers alongside public cloud instances, managed Kubernetes, object storage and databases. If you need range, it is the pick.
Hetzner covers the infrastructure essentials well: instances, load balancers, block and object storage, firewalls and managed Kubernetes. Beyond that you build it yourself.
UpCloud sits between the two, with managed databases and Kubernetes alongside its storage-focused compute.
None of the three offers anything comparable to BigQuery or a managed model training platform. If your product centres on analytics or machine learning, this is the wrong comparison to be running.
Data residency and jurisdiction
All three are European companies operating European data centres, which is the point for teams whose concern is jurisdictional rather than merely geographic. A European provider is not subject to the same non-EU legal requests as a US-headquartered one, even where the latter offers EU regions.
Residency is not compliance, though. GDPR obligations attach to how you handle personal data: lawful basis, retention, access control, processor agreements. Choosing a European provider is one input, not a conclusion.
Account creation
All three verify identity at sign-up, and all three are a common source of delay for international customers. Hetzner’s review is the strictest and most frequently referred for manual checking. OVHcloud varies by region and profile. UpCloud runs compliance checks including sanctions screening, and does not serve customers in countries subject to those measures.
A decision path
- Cheapest possible cores in Europe? Hetzner.
- Need bare metal, or the widest service range? OVHcloud.
- Database or I/O bound workload? UpCloud.
- Need managed analytics or ML? None of these; look at the large platforms.
Common questions
Are these providers suitable for production?
Yes, and a great deal of European production runs on all three. Design for failure as you would on any single provider.
Can I run across two of them?
Yes, and some teams do for redundancy. It costs operational complexity, so make it a deliberate choice.