Scaleway expands Milan zones, adds Nordic GPU clusters

Scaleway has extended its European cloud infrastructure with new availability zones in Milan and announced GPU deployments in Nordic locations, targeting enterprise AI workloads. The expansion, unveiled at the ai-PULSE event on 13 October 2026, aims to address both compute capacity and regional resilience for AI applications and surrounding data services. While the additional zones and GPU clusters broaden options for customers, their practical impact depends on service alignment, software compatibility, and confirmed delivery timelines.

Regional expansion and resilience

The Milan cloud region now offers three geographically separate availability zones, allowing customers to distribute applications across multiple locations. This setup provides redundancy options in the event of a zone failure, though Scaleway notes that resilience still requires customer-managed replication, failover procedures, and sufficient capacity. The expansion does not automatically guarantee application uptime without these measures in place.

Separately, Scaleway has announced GPU deployments in Stockholm and Copenhagen, including an AMD Instinct MI355X cluster in Sweden and an NVIDIA B300 cluster in Denmark. The company projects a total GPU footprint of tens of thousands of processors across Paris, Copenhagen, and Stockholm, though this remains a future target rather than an installed capacity. Buyers evaluating these options must consider software compatibility, as switching between AMD and NVIDIA platforms may require framework adjustments, library updates, and performance testing. Scaleway has not provided comparative benchmarks or pricing to assist these decisions.

Service alignment and partnerships

Scaleway’s roadmap includes three availability zones each for Stockholm and Frankfurt, planned for 2027. However, the value of these zones will depend on which services—such as databases, storage, or networking—are available in each location. A regional footprint alone does not ensure seamless deployment if critical capabilities are missing.

To support AI workloads, Scaleway has partnered with Dataiku, enabling customers to deploy and manage its AI orchestration platform on Scaleway’s cloud. While this offers an additional infrastructure route for Dataiku users, the partnership does not include a migration timetable or confirmation that existing deployments can transfer without modifications. Scaleway has also added support for cached tokens and customizable rate limits to its inference platform, allowing developers to manage demand and application performance, though the financial impact will vary based on workload characteristics and charging rules.

Sovereignty and hardware dependencies

Scaleway positions its expansion as a way to develop and operate AI within Europe, providing an alternative for customers seeking regional infrastructure. However, the company’s reliance on international chip vendors—AMD and NVIDIA—means hardware independence is not part of the offering. A separate partnership with VSORA and ZML aims to develop a European inference solution using the Jotunn8 processor, but this remains a prospective project without confirmed production availability or supported software.

The company’s broader roadmap includes a Unified Computing Service for 2027, integrating CPU, GPU, and quantum resources, alongside an exploratory orbital infrastructure partnership. Neither initiative is relevant for current capacity planning, as immediate procurement decisions still hinge on confirmed availability, service coverage, pricing, and migration effort for existing applications.