Outscale Enhances Managed Kubernetes Service with Autoscaling and Next-Gen GPU Support
Outscale has announced significant upgrades to its managed Kubernetes service, OKS, with new features including autoscaling and support for state-of-the-art GPUs. This advancement solidifies Outscale’s position in the realm of sovereign cloud infrastructures dedicated to sensitive AI workloads, combining high performance, strict isolation, and SecNumCloud certification.
As organizations increasingly adopt cloud-native architectures, managed Kubernetes platforms have become a essential foundation for application and AI projects. The focus now extends beyond simple container orchestration to encompass the ability to handle variable workloads, optimize critical resource management, and ensure compliance with stringent regulatory requirements.
In this context, Outscale — a subsidiary of Dassault Systèmes — is elevating its Kubernetes as a Service offering. Just eight months after its launch, OKS now includes autoscaling features and further enhances support for AI workloads. The clear intention is to facilitate the deployment of sovereign AI projects on a public cloud environment certified by SecNumCloud.
Autoscaling Addressing AI Workload Spikes
The inclusion of autoscaling is a pivotal component of this latest update. The platform can now automatically adjust resources based on real-time application needs and workloads, including resource-intensive AI processes. This solution addresses a well-recognized operational challenge for DevOps and AI teams who often face unpredictable workload spikes during model training, inference, and production scaling.
By automating resource allocation, OKS aims to ensure consistent performance while avoiding excessive resource consumption. The strategy is straightforward: simplify the management of cloud-native and AI environments while maintaining high availability and security levels, all at a time when cost-performance trade-offs are becoming increasingly critical.
Next-Gen GPU Support and Strict Isolation in SecNumCloud
Outscale is also emphasizing support for the latest generation of GPUs, which are integrated within a managed Kubernetes environment deployed exclusively on the ANSSI-certified public cloud, SecNumCloud. The platform accommodates multiple generations of Nvidia GPUs, catering to diverse needs ranging from training and inference to high-performance computing. Among the latest models supported are the Nvidia H100, designed for large-scale AI workloads, the widely used Nvidia A100 and A100-80 for model training and massive data processing, and the Nvidia L40, which is tailored for advanced inference and professional visualization tasks.
GPU allocation occurs at the Kubernetes node pool level, with users specifying the required model at deployment. This approach enhances operational control over critical resources, aligning with OUTSCALE’s mission to provide a managed Kubernetes environment capable of supporting sensitive AI workloads while ensuring strict isolation and sovereign compliance with SecNumCloud certification.
Each client receives a dedicated Kubernetes cluster with its own control plane, eliminating resource sharing. This architecture meets strict isolation requirements while leveraging open standards to ensure compatibility and reversibility of environments. According to Thibaud Piétri, VP of Cloud Infrastructure & Platform Services at Outscale, the combination of autoscaling and access to GPUs is aimed at “streamlining AI workflows and ensuring optimal performance in a highly secure environment qualified by SecNumCloud.”
Strengthening Sovereign AI Positioning
With these enhancements, Outscale reinforces its already strong positioning in technological sovereignty. OKS stands out as a platform capable of quickly deploying cloud-native and AI projects while maintaining comprehensive control over infrastructures and data. The objective goes beyond merely providing managed Kubernetes; it seeks to create a credible industrial foundation for strategic AI applications amid increasing reliance on non-European infrastructures.
This trajectory reflects a broader market trend where sovereignty is now evaluated not just by data hosting capabilities but also by the ability to operate complete, efficient technology stacks that align with the evolving demands of AI. By enriching OKS, Outscale aims to establish itself as a key player in this nascent yet increasingly strategic segment.



