Skills on AI 459 skills

Active theme: Light

Scaleway Cloud

Use when asked what Scaleway is, how to use its compute/Kubernetes/storage/GPU products, or how it compares to AWS/GCP/Azure — a European, sovereignty-focused cloud provider. Verify current regions/products/pricing at scaleway.com before repeating specifics from this file — captured from the site on 2026-09-05.

Scaleway is a European cloud and AI platform positioned as a sovereignty- and openness-focused alternative to the US hyperscalers (AWS/GCP/Azure), with an S3-compatible object storage API and Kubernetes-native services that make much existing cloud tooling portable to it with minimal changes.

The facts below were read from scaleway.com on 2026-09-05 — re-check the site before treating a specific product name, region count, or pricing claim here as still current.

Compute

  • Virtual Instances across General Purpose, Development, Compute Optimized, and Memory Optimized families — the standard VM-tier split seen across most cloud providers.
  • GPU Instances — named GPU models on the site include L40S, L4, H100, and B300-SXM, aimed at AI training/inference workloads.
  • Bare metal — multiple distinct offerings: Dedibox (dedicated servers, Scaleway's original product line), Elastic Metal (on-demand bare metal with cloud-like provisioning), and even Mac mini bare metal — notable since dedicated Apple-silicon bare metal isn't common among the major hyperscalers.

Containers and orchestration

  • Kubernetes Kapsule and Kosmos — managed Kubernetes offerings (Kapsule for Scaleway-hosted node pools; Kosmos described as a multi-cloud/hybrid Kubernetes option pulling in nodes from elsewhere).
  • Serverless Containers and Serverless Functions, plus a Container Registry — the standard serverless-compute-plus-registry trio also offered by the major clouds.

Storage and databases

  • Object Storage — explicitly S3-API-compatible, meaning existing AWS S3 SDKs/tools generally work against it by pointing at a different endpoint, without a Scaleway-specific client.
  • Block Storage and File Storage for attached/shared storage.
  • Managed databases: PostgreSQL, MySQL, MongoDB, Redis, plus a Serverless SQL Database offering and a Data Warehouse for ClickHouse (a managed ClickHouse-based analytics warehouse).

AI/ML

  • Generative APIs in both Serverless and Dedicated Deployment modes — i.e. a shared-hosted inference endpoint option alongside a dedicated-capacity option for a chosen model.
  • GPU clusters for training, and managed Apache Spark and Kafka for the surrounding data pipeline.

Networking

VPC, Load Balancer, a WAF (web application firewall), Edge Services, and Site-to-Site VPN — the standard networking building blocks, with the WAF/Edge Services pairing suggesting some CDN/edge capability alongside pure networking.

Regions and footprint

Per the site: 4 multi-AZ cloud regions in Europe, 10 Availability Zones across those regions, and 65+ points of presence worldwide (for edge/CDN-style reach, distinct from the smaller set of full compute regions).

Positioning versus the major hyperscalers

Scaleway's stated differentiators: European data sovereignty (data stays under European jurisdiction), openness (built on open, independent technology to avoid vendor lock-in — e.g. the S3-compatible API rather than a proprietary object-storage protocol), transparent, product-level pricing, and sustainability (renewable-powered datacenters with published Power Usage Effectiveness metrics). The site states 100+ integrated cloud and AI products in total.

Common pitfalls

  • Assuming full API/feature parity with AWS just because Object Storage is S3-compatible — the storage API surface is compatible, but that doesn't imply the rest of Scaleway's services mirror AWS's equivalents feature-for-feature.
  • Assuming the same global region footprint as a hyperscaler — Scaleway is explicitly concentrated in Europe (4 regions per the site), not a global region list; check current availability before planning a deployment outside Europe.
  • Treating marketing sustainability/PUE claims as independently audited — they're vendor-reported on the marketing site; check Scaleway's own published sustainability reports for methodology if that claim matters to a decision.
  • Confusing Kapsule and Kosmos — per the site, one is a Scaleway-native managed Kubernetes offering and the other is multi-cloud/hybrid-oriented; verify current scope/pricing differences before recommending one over the other.

Learn more

  • scaleway.com — the primary source for this skill; re-fetch before relying on a specific product, region, or pricing claim above.
  • Scaleway documentation — technical reference (CLI, API, per-product guides).

View scaleway-cloud/SKILL.md on GitHub