Two directions from the cluster
Palette and Ankra both put an agent in your clusters and a reusable blueprint at the centre. They point in different directions.
Palette manages Kubernetes downward. A Cluster Profile layers the operating system, distribution, networking, storage and add-ons, and Palette holds each cluster to it. That reaches bare metal through MAAS, an immutable edge OS and data-centre platforms such as vSphere, Nutanix and CloudStack. Spectro Cloud raised a Series D of more than $100 million in July 2026, and plans to use it to expand PaletteAI, its product for running AI infrastructure.
Ankra builds the cluster on the provider you choose and then works upward. It turns repositories into running services, keeps add-ons current, ranks vulnerabilities, estimates cost, backs up data and operates it all with AI.
If the operating system and the hardware are part of what you deliver, Palette fits. If your team delivers software and wants the platform work done, Ankra covers more of it.
Where the clusters come from
Palette’s reach below Kubernetes is its strength. It provisions public clouds, data-centre platforms and edge sites with an immutable OS, air-gapped bundles and zero-downtime updates, and Spectro has tested one Palette instance against 10,000 simulated edge clusters with 36,000 nodes. Palette is offered as multi-tenant SaaS in the US or the EU, as dedicated SaaS or self-hosted.
Ankra builds kubeadm or k3s clusters on Hetzner, OVHcloud, UpCloud, DigitalOcean, AWS EC2 and Proxmox VE, with a private network and bastion. It creates six provider-run services that are generally available (EKS, GKE, AKS, DOKS, OVHcloud MKS and UpCloud UKS), plus Scaleway Kapsule and Ankra Cloud Kubernetes on Ankra’s own EU cloud, both in closed beta. It also imports any conformant cluster, including ones Palette made.
On clusters it builds, Ankra handles upgrades, node groups and power schedules and, on the five public clouds, node autoscaling and rolling OS security patching, one host at a time. It does not manage operating system images or bare metal. Ankra’s control plane is a hosted service in the EU, and Enterprise customers can run Ankra self-hosted, including air-gapped, as an add-on on the Enterprise plan.
Profiles and stacks
A Palette profile is a versioned object in Palette. Teams manage it as code through the Terraform and Crossplane providers, and Palette can deploy Argo CD as a pack for application GitOps.
An Ankra stack is a set of Helm charts and manifests with dependency edges, so the database comes up before the service that needs it. With a GitHub or Bitbucket Cloud repository connected, stacks live in your repository, and commits you push are applied by Ankra’s own deployment engine. A sync that would remove a large share of a cluster is held for confirmation. Stack profiles version a working setup and carry it to each new environment with its own inputs.
Shipping your own code
Palette’s App Mode models applications with services on virtual clusters. It has no CI pipeline with tests and scanners, and no image builds.
Ankra treats delivery as part of the platform. Applications analyse a GitHub repository, open a setup pull request with build files, and deploy every successful build. If the repository already builds its image in GitHub Actions, GitLab CI, Bitbucket Pipelines or CircleCI, Ankra deploys that image instead. Ankra Pipelines run tests, rootless builds and Semgrep, Checkov and Trivy scans in your own cluster, and publish only the image that passed the gate. Verified deploy provenance can refuse images your pipeline did not sign.
Some limits apply. Several documented pipeline stage kinds do not run yet. Pull-request previews work for Applications on GitHub, and Ankra’s own docs say not to rely on them yet while an open bug is fixed.
Security, cost and recovery
Palette is built for regulated buyers. It runs CIS, penetration, conformance and SBOM scans, and reports each vulnerability’s severity and fixed version. Its VerteX edition uses validated FIPS 140-3 cryptographic modules, and since October 2026 VerteX multi-tenant SaaS holds FedRAMP Moderate certification. Self-hosted and air-gapped installs are not covered by it.
Ankra’s Security Center ranks vulnerabilities in running workloads by CISA’s Known Exploited Vulnerabilities catalogue and FIRST’s EPSS scores, so the exploited ones come first. It offers optional SBOMs and produces compliance evidence for Pod Security Standards, and for CIS and NSA/CISA where those inputs are enabled. Ankra’s security and compliance material is on its trust page.
On cost, both products go below the cluster total. Palette estimates cluster cost and breaks it down per namespace from actual usage. Ankra’s Cloud Cost estimates spend per namespace and stack from list prices. It also reprices a whole cluster on every provider Ankra can build on, and it tracks budgets through the CLI, API and AI (the portal screen is in closed beta).
Palette runs Velero backups to S3, GCS, Azure Blob or S3-compatible storage. Ankra’s backups write volumes and PostgreSQL or MySQL dumps to a bucket you own and restore them in place. CloudNativePG and Percona databases are captured, but restoring them in place is not available yet.
Two meanings of AI
PaletteAI is about running AI workloads, with GPU-backed clusters, pre-validated AI stacks and private model endpoints. To operate Palette with AI, Spectro offers an MCP server that is read-only by default, with write tools behind a startup flag, and a Claude Code plugin with diagnostic skills. Both connect your own AI client.
Ankra’s AI is built into the platform. It drafts stacks from a description, analyses firing alerts with AI Insights, and reviews pull requests on GitHub, GitLab and Bitbucket Cloud through the AI Gateway, opening fix pull requests on GitHub. In-product writes wait for a person by default, under autonomy controls you set. Ankra also has an MCP server for your own tools, with read and write scopes. Its writes are audited, and your MCP client decides how to confirm them.
When to choose Ankra, and when to choose Spectro Cloud
Choose Ankra when your team ships software on Kubernetes and wants clusters, delivery, CVE ranking, cost comparison and backups handled in one product, on European providers such as Hetzner, OVHcloud and UpCloud, on Proxmox VE or on the big clouds.
Choose Spectro Cloud when
- you manage bare metal, operating system images and the distribution as one declarative object
- you run large edge estates where the device is part of the deliverable
- you need FIPS 140-3, a FedRAMP Moderate SaaS or US public-sector procurement routes
- you run vSphere, Nutanix, CloudStack or MAAS in your data centres
- you are moving virtual machines off VMware, or building GPU infrastructure with PaletteAI
Compare the work that remains
- Create the foundation. Record the work outside the product, including the operating system.
- Ship a release. Include build, tests, scans and a rollback.
- Upgrade an add-on. Note who checks dependencies and how a failed upgrade is undone.
- Investigate a vulnerability. Find every affected workload and release the fix.
- Compare total cost. Include subscriptions, infrastructure and operating effort.
Palette’s free allotment of 25 kCh a month and Ankra’s free 30 worker vCPU are similar in size. Beyond that, Spectro’s own site says to contact sales, and an AWS Marketplace listing shows annual tiers from $8,500 for up to 300 kCh. Ankra’s pricing is €0.01 per worker vCPU-hour on Startup, about €7.20 a month, and your provider bills the servers.
Reviewed 9 November 2026 against Spectro Cloud’s documentation and pricing. Product scope and plans can change.