How EU cloud providers actually compare

There are now over a dozen European cloud providers competing for your workloads. But how do they actually stack up against each other? Not in marketing speak, but in concrete features: can you get managed Kubernetes? Are there GPU instances? Is there a WAF? What about message queues?

We went through the product pages and documentation of the major EU cloud providers and mapped out exactly what each one offers across compute, storage, networking, managed services, and AI/GPU infrastructure. Every checkmark in the tables below was verified against the provider's current product catalog.

Compute

Provider Bare Metal Dedicated VPS / VMs Kubernetes GPUs Serverless
OVHcloud
Scaleway
Gcore
Open Telekom Cloud
IONOS
Hetzner
UpCloud
StackIT

Most providers offer managed Kubernetes, except Hetzner (by design). Serverless remains rare: only Scaleway, Gcore, and Open Telekom Cloud. GPU instances are broadly available across all providers listed here.

Storage

Provider Block Storage Object Storage File Storage (NFS)
OVHcloud
Scaleway
Gcore
Open Telekom Cloud
IONOS
Hetzner
UpCloud
StackIT

Block and object storage are table stakes: every provider offers both. Managed NFS is available from most, except Scaleway and Hetzner.

Networking

Provider VPCs Load Balancers Gateways VPNs WAF
OVHcloud
Scaleway
Gcore
Open Telekom Cloud
IONOS
Hetzner
UpCloud
StackIT

VPCs and load balancers are universal. Beyond that, it varies. Open Telekom Cloud has the most complete networking stack. WAFs are offered by Scaleway, Gcore, Open Telekom Cloud, and StackIT. Managed VPNs only from Open Telekom Cloud, IONOS, and UpCloud.

Managed services

Provider Databases Queues Transactional Email
OVHcloud PostgreSQL, MySQL, MongoDB, Valkey, Kafka, Cassandra, OpenSearch
Scaleway PostgreSQL, MySQL, Redis, MongoDB, ClickHouse SQS, SNS, NATS, Kafka
Gcore PostgreSQL
Open Telekom Cloud PostgreSQL, MySQL, SQL Server Kafka
IONOS PostgreSQL, MySQL, MongoDB, Redis Kafka
Hetzner
UpCloud PostgreSQL, MySQL, Valkey, OpenSearch
StackIT PostgreSQL, MongoDB, MariaDB, SQL Server, Redis, OpenSearch RabbitMQ

This is where differences become stark. Scaleway is the only provider offering databases, queues, and transactional email in one platform, with SQS/SNS protocol compatibility. OVHcloud and StackIT have the most database engines. Hetzner offers no managed services at all, by design. Message queues remain a gap: only Scaleway, StackIT (RabbitMQ), IONOS (Kafka), and Open Telekom Cloud (Kafka).

GPUs and AI

Provider GPU Models AI Inference API
OVHcloud H200, H100, A100, L40S, L4, V100S, T4
Scaleway H100, L40S, L4
Gcore H200, H100, A100, L40S
Open Telekom Cloud H100, A100, V100, T4 ModelArts (train + deploy)
IONOS H200, H100, L40S, A10, T4, RTX PRO 6000, Gaudi 2/3
Hetzner RTX 4000 Ada, RTX 6000 Ada, RTX PRO 6000 (dedicated only)
UpCloud L40S
StackIT Not disclosed (up to 8 GPUs per instance)

OVHcloud and IONOS have the widest GPU selection. IONOS uniquely offers Intel Gaudi as a non-NVIDIA option. AI inference APIs (OpenAI-compatible) are available from OVHcloud, Scaleway, Gcore, IONOS, and StackIT. Hetzner's GPUs are dedicated servers only, not elastic cloud instances.

Our picks

Best for startups and developers on a budget: Hetzner. If you are comfortable with Linux and want to keep costs as low as possible, nothing in Europe comes close. You trade managed services for raw infrastructure at prices that make even other EU providers look expensive. Pair it with external services for AI inference.

Best all-rounder with an AWS-like experience: Scaleway. If you want one provider that handles compute, Kubernetes, databases, queues, serverless, email, and AI inference without stitching together five different vendors, Scaleway is the closest thing the EU has to a full-service cloud. The developer experience and documentation are strong.

Best for GPU and AI workloads: OVHcloud or IONOS. Both offer the broadest range of GPU models. OVHcloud also has AI Endpoints for serverless inference. IONOS uniquely offers Intel Gaudi accelerators if you want a non-NVIDIA option.

Scaleway comes closest in terms of breadth. It offers managed Kubernetes, serverless functions, object storage, managed databases, message queues (with SQS/SNS protocol compatibility), transactional email, and AI inference APIs. Open Telekom Cloud also covers a wide range but with a more enterprise-oriented approach.

Most providers offer managed Kubernetes: OVHcloud, Scaleway, Gcore, Open Telekom Cloud, IONOS, UpCloud, and StackIT. Hetzner does not offer managed Kubernetes but is a popular choice for self-managed clusters using k3s or similar tools.

OVHcloud and IONOS offer the widest range of GPU models, including H200 and H100. IONOS also uniquely offers Intel Gaudi accelerators. For AI inference without managing GPUs yourself, Scaleway, Gcore, OVHcloud, IONOS, and StackIT all offer OpenAI-compatible API endpoints.

For features like advanced serverless ecosystems (Step Functions, EventBridge), proprietary databases (DynamoDB, Cosmos DB, BigQuery), or deep IAM systems, there is no direct EU equivalent today. You can self-host open source alternatives (e.g., MinIO for S3, CockroachDB for distributed SQL, Temporal for orchestration) or use a hybrid approach with a US hyperscaler sovereign cloud for specific services.