Kupe Cloud is in closed alpha: Kubernetes without the assembly

Kupe Cloud is in alpha.

Alpha is a real platform rather than a prototype: you can spin up a cluster, deploy to it through GitOps, manage it through the console, and inspect its monitoring dashboards. We’re super proud of where we are, the foundations we’ve built and how the platform is shaping up.

What alpha is actually for

We want to see if our platform works for other people: does Kupe take away enough of the tedious platform work to be genuinely useful? We built Kupe as the platform we always wished we had, alpha is where we get some honest truths, where we’ve oversteered, where we’re unclear, and what still feels heavy.

The target we’ve set ourselves is to go from zero to a running cluster with healthy workloads, monitoring and alerting in about fifteen minutes, no matter how much Kubernetes you’ve done before. Where we are feels pretty close to this, but alpha is how we find out how close we really are.

The bits we’re most proud of

Getting a running Kubernetes cluster isn’t the hard part. The hard part is everything around the cluster, the GitOps, the secrets, the metrics, the logs, the ingress, the dashboards, the alerts. We’ve poured our effort into making all of this ready to go from day one. Some of the highlights:

• GitOps is wired before you arrive. Create a cluster from the console, the CLI, or straight from Terraform. By the time it’s up, GitOps delivery is ready to use. Connect your repository, select your cluster, and let the platform keep your deployments in sync.
• It really is Kubernetes underneath. We aren’t abstracting to the point where you lose control: your existing tools, pipelines and workflows all work, and you get full Kubernetes API access to your clusters.
• High availability when a cluster needs it. Control planes run a single replica by default. Ask for high availability when you create the cluster, from the console, the CLI, or Terraform, and you get three control-plane replicas with resilient cluster-state storage instead.
• A working CLI. A few kupe cluster commands and you’re talking to your cluster with kubectl. It covers the basics rather than every corner of the platform, but the day-to-day path is there.
• Infrastructure as code, as you’d expect. We love Terraform, and wouldn’t have been able to move forward without a working provider. Ours is published on both the Terraform and OpenTofu registries. It doesn’t cover everything just yet, but it handles the core stuff.
• A console that doesn’t hide the cluster. See every workload, drill into a pod, tail its logs, and open a shell. Workload metrics and costs are part of the same view, so you can inspect your Kubernetes resources in one place.
• Secrets that aren’t weird. Managed secrets sync into your clusters as ordinary Kubernetes Secret objects. You manage them in one place, and your workloads read them without a bespoke SDK or sidecar.
• Observability from the first minute. Monitoring, logging, alerting, and dashboards are ready when you create a cluster, with a dedicated monitoring workspace for your tenant.

We try to keep everything simple and stay as close to native Kubernetes as possible. We don’t want to remove control or capabilities. We want to absorb the complexity so the common paths feel obvious. This balance is what we’re most proud of, not a single technology, or capability, but bringing the right pieces together in a way that makes building in the Kubernetes ecosystem make sense, no matter the team size or Kubernetes expertise.

What’s missing, and we know it

We’ve made some great progress, but there are still a few areas we know aren’t where we want them yet.

• Dedicated nodes. Right now your cluster shares host infrastructure with others. That’s genuinely fine for testing and for a lot of real workloads, but if you need hard isolation or reserved, predictable capacity, that option isn’t there yet.
• Backup tooling. Backup cover for our own platform services isn’t complete yet, and we’re still filling it in. What runs inside your cluster stays yours to back up, and that won’t change, but we want to make doing it far easier than it is today.
• The docs are catching up. We’ve put real effort into getting our docs into good shape, but they’re still catching up with our engineering. If you spot anything wrong in them, please let us know!

Please see our Alpha Service Notice for the full breakdown.

Try it out for yourself

We welcome you to sign up and give it a go. Alpha started invite only; signup is now open, and Kupe Cloud alpha is open covers what the closed months found.

It’s completely free during alpha. No card, no charges, nothing to cancel. We’re not going to start charging anyone while we’re still working out what needs fixing, and when billing does arrive you’ll get plenty of notice.

We’ve got a getting started section in our documentation which walks you through the first steps, but feel free to skip it and use the Kupe platform in any way that makes sense to you.

We’d really appreciate any feedback, big or small, good or bad, please send it to support@kupe.cloud.

Thank you, genuinely, for helping us learn what Kupe needs to become.