Azure Kubernetes Service, AKS, gives you a managed container platform for modern applications. Run it on Azure Local and your containers run on your own hardware, close to your data and your machines, with the same Kubernetes tooling and Azure integration you'd get in the public cloud.
Modern applications and AI models increasingly run in containers, not virtual machines. AKS on Azure Local lets you run them locally, which matters when data can't leave the site or when you need low latency to the shop floor. It's also how you run AI models on your own GPUs, under your own governance, instead of sending sensitive data to a public service.
You add AKS to your Azure Local platform and deploy containerized applications with the tooling your team already uses. For AI, you set up Foundry Local on AKS with GPU capacity, so models run on your own hardware. Everything is managed as code and monitored centrally, so it stays something your own team can run, not a project only we understand.
Azure Kubernetes Service is Microsoft’s managed Kubernetes. On Azure Local it runs as an Arc-enabled cluster on your own hardware, with the same APIs and tooling as in the public cloud. It supports GPU nodes for AI workloads, and you deploy with the pipelines your team already uses. What that looks like in practice, from setup to deployment, is in the blogs below.
Run a real workload with your own data, in our Technology Experience Center or on your own site. Seeing it work beats any presentation.

AKS in Azure runs your Kubernetes clusters in an Azure region, on infrastructure Microsoft manages. AKS on Azure Local runs them on your own hardware, managed from the same Azure portal with the same Kubernetes APIs. You choose per application: Azure for scale, Azure Local when the data or the machines are on-site. We've run Kubernetes on this platform since the Azure Stack HCI days and have written about it in depth on our blog.

Yes. AKS on Azure Local supports GPU nodes, so you can run AI models in containers on your own GPUs. With Foundry Local on Azure Local, Microsoft provides a ready-made way to serve those models on that platform, managed through Azure Arc.

No. AKS on Azure Local runs standard Kubernetes, so kubectl, Helm and your existing CI/CD pipelines work the same way as with AKS in Azure. Developers deploy to an on-premises cluster the same way they deploy to the cloud.

It depends on how your applications are built. Modern applications that are split into services and updated often get the most from containers: faster deployments, better use of hardware and easier scaling. Older applications built for a single server usually stay on virtual machines, and Azure Local runs both side by side.


Hardware up for renewal and VMware costs climbing? Modernize instead of renewing the same setup.

Run AI on GPUs inside your own datacenter, so your data stays where you put it.