Sovereign by Architecture: Building AI Infrastructure for the EU AI Act
The EU AI Act takes effect August 2026. Compliance starts at the infrastructure layer. Learn why sovereign AI needs OpenStack, Kubernetes, and Atmosphere.
Perspectives, mises à jour et histoires de notre équipe
The EU AI Act takes effect August 2026. Compliance starts at the infrastructure layer. Learn why sovereign AI needs OpenStack, Kubernetes, and Atmosphere.
Learn how a lightweight keystoneauth1 plugin brings your existing browser-based MFA and SSO to the OpenStack CLI, with no changes to any client tools.
Hyperscaler AI looks fast but hides long-term lock-in and rising costs. See how OpenStack and Kubernetes deliver GPU infrastructure you actually control.
Kubernetes is everywhere in 2026—but where you run it matters. Explore infrastructure trade-offs between hyperscalers and OpenStack.
Kubernetes won. We all know that by now. Whether you're at a startup or an enterprise, chances are you're running containers orchestrated by K8s. But here's the thing nobody talks about enough: where you run Kubernetes matters just as much as how you run it.
In 2026, Kubernetes is everywhere. But "everywhere" doesn't mean "equal underneath."
"Kubernetes is portable!" we told ourselves. "Write once, run anywhere!" And technically, that's true. Your manifests will work on EKS, GKE, AKS, or a bare metal cluster running in your basement.
But portability of workloads doesn't mean portability of operations. It doesn't mean portability of costs. And it definitely doesn't mean freedom from vendor decisions that affect your entire infrastructure strategy.
When you deploy a Kubernetes cluster, you're making a choice about the infrastructure layer beneath it. That choice has ripple effects:
Hyperscaler-managed Kubernetes (EKS, GKE, AKS): You get convenience. Managed control planes, seamless integration with cloud-native services, auto-scaling that actually works, and someone else's pager going off at 3am when etcd has issues.
But you also get:
OpenStack + Kubernetes: You get control. Full ownership of the infrastructure stack, transparent costs, the ability to optimize at every layer, and the freedom to run workloads without worrying about surprise bills.
But you also get:
Not every organization needs to think deeply about this. If you're running a few microservices with modest traffic, just use a managed service and move on with your life.
But infrastructure choice becomes critical when:
You're operating at scale. Once you're running hundreds of nodes, egress fees and premium managed service costs add up fast. A 2TB/month data transfer bill becomes 20TB, then 200TB. Suddenly that "convenient" managed service is costing you six figures annually just in bandwidth.
You have regulatory requirements. Some industries need to know exactly where data lives and who has access. "Trust us, it's in the cloud" doesn't cut it when auditors come knocking.
You're building platform services. If you're a SaaS provider offering Kubernetes to your customers, running K8s on someone else's managed K8s gets expensive and architecturally weird fast.
You need predictable costs. Hyperscaler bills are notoriously difficult to forecast. OpenStack + Kubernetes on your own hardware or bare metal cloud gives you cost predictability that matters for long-term planning.
This is how 2025 at VEXXHOST and cloud computing world went by. Here's what we're seeing in 2026: smart teams aren't picking one approach exclusively. They're running:
This isn't "multi-cloud" in the buzzword sense. It's pragmatic infrastructure design that matches the tool to the job.
If you're making infrastructure decisions in 2026, ask yourself:
Kubernetes everywhere is real. But Kubernetes on equal footing? Not even close. The infrastructure beneath your clusters shapes your costs, your capabilities, and your constraints more than most teams realize until it's too late.
In 2026, the smartest infrastructure choice isn't "cloud" or "on-prem" or "OpenStack" or "hyperscaler." It's understanding the trade-offs clearly enough to make the right call for your specific situation—and being willing to use different approaches for different workloads.
Because Kubernetes might be everywhere, but where you run it still matters. A lot.
We've been running production Kubernetes on OpenStack since before it was cool. VEXXHOST can help you build infrastructure that scales without the hyperscaler tax. Reach out to discuss your needs.
Choose from Atmosphere Cloud, Hosted, or On-Premise.
Simplify your cloud operations with our intuitive dashboard.
Run it yourself, tap our expert support, or opt for full remote operations.
Leverage Terraform, Ansible or APIs directly powered by OpenStack & Kubernetes