Managed Kubernetes

What it is

Managed Kubernetes clusters run a conformant control plane that we operate, upgrade and patch for you. You bring your manifests; we schedule pods onto GPU-aware node groups across Atlas regions. Kubeconfig is short-lived and rotated via the API.

When to use it

Existing Helm charts, operators, or in-house tooling that targets Kubernetes.
Teams that want a control plane abstraction familiar to all their engineers.
Mixed GPU + CPU workloads with custom scheduling logic.

Quickstart

asc k8s create prod --region apac --version 1.30 --nodes h100=2 --nodes cpu-xl=4
asc k8s kubeconfig prod > ~/.kube/asc-prod
kubectl --kubeconfig ~/.kube/asc-prod apply -f my-app.yaml

cluster = client.k8s.create(name='prod', region='apac', k8s_version='1.30', node_groups=[{'gpu': 'h100', 'count': 2}, {'cpu_sku': 'cpu-xl', 'count': 4}])
kubeconfig = client.k8s.kubeconfig(cluster.id, ttl_seconds=900)

const cluster = await client.k8s.create({ name: 'prod', region: 'apac', k8sVersion: '1.30', nodeGroups: [...] });
const kubeconfig = await client.k8s.kubeconfig(cluster.id, { ttlSeconds: 900 });

Limits & quotas

Limit	Default	Burst	Notes
Clusters per project per region	10	—
Nodes per cluster	1,000	5,000	Lift via sales
Supported versions	Current + previous minor	—	Upgrades on a 6-month cadence

Pricing

See pricing. Pay-as-you-go, billed monthly via Stripe.

API surface

POST /v1/atlas/k8s
POST /v1/atlas/k8s/{id}/kubeconfig
POST /v1/atlas/k8s/{id}/upgrade

Required scope(s): k8s:read, k8s:write. See Scopes.

Security

All access is authenticated and scoped. See Auth & scopes and Network controls.