Scaleway
European cloud and AI platform for compute, storage, Kubernetes, and GPUs
About Scaleway
Scaleway is a European cloud provider, part of the iliad Group, offering 100+ cloud and AI products from Paris, Amsterdam, Warsaw, and Milan regions. It covers virtual instances, Dedibox and Elastic Metal bare metal servers, managed Kubernetes, S3-compatible object storage, managed databases, serverless, GPU instances, and hosted generative AI APIs. It positions itself as a data-sovereign alternative to US hyperscalers with per-product, pay-as-you-go pricing.
Pros & Cons
Pros
- European data sovereignty with Multi-AZ regions in France, the Netherlands, Poland, and Italy
- Egress included in instance list prices and 75 GB/month free Object Storage egress
- Free Kubernetes Kapsule control plane, plus Kosmos for multi-cloud clusters
- Very cheap entry options such as Dedibox servers from €4.99/month
- Broad AI stack: GPU instances up to H100 and B300, plus pay-per-token Generative APIs for open models
- S3-compatible storage and a Terraform provider make migration from AWS straightforward
Cons
- Smaller service catalog and global footprint than AWS, Azure, or Google Cloud (European regions only)
- Some reviewers report occasional network instability and slow Object Storage
- High-end GPU types (L40S, H100) are limited to specific availability zones
- Prices are listed in euros, and attached public IPv4 and storage are billed separately from instance prices
- Support quality is mixed in reviews; faster response requires paid support plans
Key Features
Virtual Instances
120+ VM types across Development, General Purpose, Compute Optimized, and Memory Optimized ranges, billed per hour with egress and IPv6 included in list prices.
Bare Metal (Dedibox and Elastic Metal)
Monthly dedicated Dedibox servers from €4.99/month plus hourly Elastic Metal servers and Apple Mac mini (M1 to M4) machines on demand.
Kubernetes Kapsule and Kosmos
Managed Kubernetes with a free mutualized control plane, dedicated control planes with SLA, and Kosmos for multi-cloud clusters spanning other providers.
Object, Block, and File Storage
S3-compatible Object Storage with Standard Multi-AZ, One Zone, and Glacier classes (Glacier stored 25 m underground), plus 5K/15K IOPS Block Storage and managed File Storage.
GPU Instances
NVIDIA L4, L40S, H100 PCIe, H100 SXM, and B300 instances for AI training, inference, and rendering, up to 8 GPUs per instance.
Generative APIs
Serverless API access to open models such as Mistral, Llama, Qwen, DeepSeek, gpt-oss, and Whisper, billed per million tokens with a free tier and 50% off batch requests.
Managed Inference
Dedicated deployments of AI models on reserved GPUs (L4 to H100 SXM), billed hourly once the model is deployed.
Serverless
Serverless Functions, Serverless Containers, Serverless Jobs, and Serverless SQL Database that scale to zero.
Managed Databases and Data
Managed PostgreSQL, MySQL, Redis, MongoDB, Kafka, OpenSearch, and Messaging and Queuing services.
Networking and Developer Tooling
Multi-AZ VPC, Load Balancer, Public Gateway, Site-to-Site VPN, Edge Services, Domains and DNS, plus a full API, CLI, and Terraform provider.
Pricing
- Kubernetes Kapsule mutualized control plane (up to 150 nodes)
- Generative APIs: 1 million tokens and 60 minutes of audio transcription free
- 75 GB/month free Object Storage egress
- Object Storage requests and ingress included
- Start-2-S-SSD dedicated server
- Intel C2350 2C/2T, 4 GB RAM
- 1 x 120 GB SSD, 250 Mbps
- Up to 15% off with 12-36 month commitment
- DEV1-S: 2 vCPUs, 2 GB RAM, 200 Mbps
- Billed hourly (€0.00898/hour)
- Egress and IPv6 included
- PLAY2 range from ~€10.42/month
- PRO2-XXS: 2 vCPUs, 8 GB RAM, 350 Mbps
- Billed hourly (€0.0561/hour)
- POP2 range from ~€53.65/month
- Savings Plans cut compute costs up to 25%
- Dedicated 4 GB control plane, 2 replicas
- Up to 250 nodes
- 99.5% SLA
- Larger Dedicated-8 and Dedicated-16 tiers available
- L4-1-24G: 1 NVIDIA L4, 8 vCPUs, 48 GB RAM
- ~€574.87/month
- Scales to 8 GPUs per instance
- L40S, H100, and B300 instances also available
Best For
GDPR-sensitive workloads
European companies that need data hosted by an EU provider outside US jurisdiction run apps, databases, and storage on Scaleway's Multi-AZ regions.
AI training and inference
AI teams rent L4, L40S, or H100 GPUs, or call open models through Generative APIs, without managing their own GPU fleet.
Cost-efficient hosting and bare metal
Developers and agencies host sites, game servers, and backends on low-cost Dedibox or Development instances.
Kubernetes in production
Engineering teams run containerized apps on Kapsule with a free control plane and integrated load balancers, registry, and block storage.
S3-compatible storage and archiving
Teams move backups and media from AWS S3 to Scaleway Object Storage or Glacier to cut storage and egress costs.
Similar Tools
Salesforce
The world's #1 CRM platform for sales, service, marketing, and more
Replicate
Run AI with an API
Cerebras
The world's fastest AI inference � 20x faster than GPU clouds
Vultr
High-performance cloud compute, GPU, and bare metal across 32 global data centers
Groq
Ultra-fast AI inference powered by custom LPU silicon
Deepfence ThreatMapper
Open source cloud-native application protection platform