Scaleway announced plans for an AMD AI supercomputer in Stockholm, Sweden, on October 13, 2026. The planned system will use AMD Instinct MI355X GPUs and become the French cloud provider’s first supercomputer deployment outside France, expanding the range of AI computing infrastructure available to European businesses.
The announcement was made during the ai-PULSE 2026 conference in Paris. Scaleway intends to provide startups, AI developers and enterprises with access to AMD-powered computing resources for artificial intelligence training and inference workloads.
The company describes the project as Europe’s first commercial AMD AI supercomputer. However, Scaleway has not disclosed the number of accelerators, total computing capacity, investment amount or expected commercial launch date. The announcement concerns a planned deployment rather than a completed installation.
Project: AMD AI supercomputer
Cloud provider: Scaleway
Location: Stockholm, Sweden
GPU architecture: AMD Instinct MI355X
Software ecosystem: AMD ROCm
Announcement date: October 13, 2026
Deployment status: Planned
GPU count: Not disclosed
Launch date: Not announced
How Scaleway’s AMD AI Supercomputer Will Be Built
The Swedish system will combine AMD Instinct MI355X accelerators with the ROCm software ecosystem. The platform is intended to support demanding AI training and inference workloads while giving customers an alternative to infrastructure based on NVIDIA GPUs.
ROCm provides software tools and libraries for applications running on AMD accelerators. For cloud customers, support for another GPU architecture can increase deployment flexibility, although actual compatibility and performance depend on individual applications and software frameworks.
According to Scaleway’s October 13 announcement, the company is working with six technology partners to deliver the computing infrastructure.
| Technology Partner | Infrastructure Responsibility |
|---|---|
| AMD | Instinct MI355X accelerators and ROCm software |
| Bull | Cluster integration, deployment and hardware maintenance |
| Giga Computing | Server hardware for the computing cluster |
| HPE Networking | High-performance Ethernet networking |
| VAST Data | High-performance data platform for AI workloads |
| Motul | Liquid-cooling fluid for high-density computing infrastructure |
The involvement of multiple suppliers reflects the infrastructure requirements of modern AI clusters. High-performance accelerators need sufficient network bandwidth, reliable data access and cooling systems capable of supporting sustained computational workloads.
Bull is responsible for bringing these components together, including installation and ongoing hardware maintenance.
Why Sweden Matters for European AI Infrastructure
The AMD AI supercomputer project represents another step in Scaleway’s expansion across Northern Europe. Sweden offers an established technology ecosystem and access to a largely low-carbon electricity mix, both relevant to the development of power-intensive computing facilities.
The deployment builds on Scaleway’s partnership with Swedish telecommunications operator Tele2, announced earlier in 2026.
Scaleway has also announced plans to open three cloud availability zones in Stockholm in 2027. However, the company has not confirmed that the same schedule applies to the AMD-powered supercomputer.
The distinction matters because cloud availability zones and dedicated AI computing clusters have different infrastructure requirements and deployment schedules.
AMD and NVIDIA Will Both Remain Part of Scaleway’s GPU Strategy
The Swedish deployment does not represent a complete shift away from NVIDIA hardware.
At the same October 13 event, Scaleway separately announced plans for a new GPU Zone in Copenhagen, Denmark, using NVIDIA HGX B300 systems.
The two projects suggest a multi-architecture approach to AI infrastructure, allowing Scaleway to develop computing services around different accelerator platforms.
For customers, the choice between AMD and NVIDIA environments can affect software compatibility, available memory, workload performance and operating costs.
The Swedish project is also distinct from AMD’s Helion AI server rack architecture, previously covered by Servers.News. Helion concerns rack-scale hardware design, while Scaleway’s announcement focuses on delivering commercial cloud computing capacity using AMD Instinct accelerators.
Computing Capacity and Launch Date Remain Undisclosed
Several important questions remain unanswered.
Scaleway has not announced the Swedish system’s GPU count, aggregate processing performance, facility power requirements, investment value or commercial availability date.
Independent reporting by Data Center Dynamics also confirms that the expected computing capacity and number of GPUs have not been provided.
Without those figures, direct comparisons with other European AI computing clusters would be premature.
The eventual capabilities of the AMD AI supercomputer will depend not only on accelerator specifications but also on networking, storage performance, system integration and the capacity made available to customers.
What Happens Next for Scaleway?
The next important milestones will be confirmation of the installation schedule, cluster size, available services and commercial pricing.
The company has not yet provided a timetable for when customers will be able to access the announced infrastructure.
For European AI developers and enterprises, the planned AMD AI supercomputer could offer another option for deploying compute-intensive applications without relying exclusively on NVIDIA-based cloud environments.
Its actual impact will become clearer once Scaleway discloses the system’s capacity, availability and operational characteristics.
