Metal3: Infrastructure Automation Ledger 2026 Scorecard
Metal3 turns bare metal into Kubernetes objects reconciled by an operator on top of Ironic; CNCF Incubating since 2025, with no product of its own.
Where Metal3 put its effort, 2020 to 2026
Each square is one capability area in one year. The brighter it is, the more the area gained in the Infrastructure Automation Ledger that year; 1, 2 and 3 mark that year's biggest pushes.
| ’20 | ’21 | ’22 | ’23 | ’24 | ’25 | ’26 | |
|---|---|---|---|---|---|---|---|
| Provisioning breadth | |||||||
| Configuration management depth | |||||||
| State, planning and safety | 1 | ||||||
| Language and authoring | |||||||
| Orchestration and self-service | 2 | ||||||
| Scale and control plane | |||||||
| Security and compliance | |||||||
| Bare metal and day-2 operations | 3 | ||||||
| Ecosystem and integrations | 1 | 1 | |||||
| AI assistance |
Biggest pushes overall: Ecosystem and integrations (+1), State, planning and safety (+0.5), Orchestration and self-service (+0.5). Already strong in 2019: Bare metal and day-2 operations, Scale and control plane, Language and authoring.
Top focus each year: 2020 Ecosystem and integrations; 2025 Ecosystem and integrations; 2026 State, planning and safety. 4 of 7 years show no gain in any area.
Part of the Infrastructure Automation Ledger (edition 2026-10-16). How the scores work: methodology.
Snapshot
- Type: community project (CNCF Incubating)
- Owner: Cloud Native Computing Foundation; started by Red Hat in 2019 and led by Ericsson and Red Hat (secondary)
- Engine: OpenStack Ironic
- Licence: Apache-2.0; free; no product or paid support of its own; vendors embed it (not verified)
- Latest release: baremetal-operator 0.14.2 (2026-10-15); cluster-api-provider-metal3 1.14.1 (2026-09-23)
- Support lifecycle: Two most recent minors supported (BMO 0.14 and 0.13; CAPM3 1.14 and 1.13); no LTS
- Next milestone: none published
- Targets: Bare metal only, as Kubernetes BareMetalHost objects, through IPMI, Redfish, Dell iDRAC 8+ and HPE iLO 5+ (Fujitsu iRMC removed in 0.13, iLO 4 and iLO 5 vendor drivers in 0.12); builds Kubernetes clusters through Cluster API; no VM, cloud, vSphere, Proxmox VE or Nutanix integration
- Tier: Host and bare-metal provisioning
- Archetype: Dependable builder
- Evidence grade: B. Release feeds, the version support table, the supported hardware page, the licence and the CNCF project page were read from primary sources. Origin, adopters and contributor counts come from secondary summaries. Releases from 2021 to 2024 were not researched, so capability before 2026 is held flat on inferred values; the BMO 0.13 month is inferred from the CAPM3 1.13.0 date.
Scope of the scores: Metal3 has no plan or dry-run mode. It holds desired state as BareMetalHost objects that an operator reconciles continuously, with inspection data and status in the object, which is how it passed inclusion test 4. It needs a Kubernetes management cluster. Scored with those limits.
Scorecard
| Dimension | Score | Direction |
|---|---|---|
| Capability today | 34.3 / 100 | higher is better |
| Momentum since end of 2022 | +3.8 points (2019 to 2022: +0.8) | higher is better |
| Credibility | 15.5 / 25 | higher is better |
| Sustainability risk | 7.5 / 25 | lower is better |
| Technical lock-in | 8.0 / 25 | lower is better |
| Integrator fit | 7.5 / 20 | higher is better |
| AI leverage | 2.0 / 20 | higher is better |
| Ledger Index | 43.9 / 100 | balanced weights; see the methodology |
Capability by domain, 2015 to today
Scores 0 to 5 against fixed anchors; total is weighted to 100.
| Domain (weight) | 2015 | 2019 | 2022 | Today |
|---|---|---|---|---|
| Provisioning breadth (15) | n/a | 0.5 | 0.5 | 0.5 |
| Configuration management depth (12) | n/a | 1 | 1 | 1 |
| State, planning and safety (12) | n/a | 1.5 | 1.5 | 2 |
| Language and authoring (10) | n/a | 2 | 2 | 2 |
| Orchestration and self-service (10) | n/a | 1.5 | 1.5 | 2 |
| Scale and control plane (8) | n/a | 2.5 | 2.5 | 2.5 |
| Security and compliance (10) | n/a | 2 | 2 | 2 |
| Bare metal and day-2 operations (8) | n/a | 3 | 3 | 3.5 |
| Ecosystem and integrations (8) | n/a | 1.5 | 2 | 2.5 |
| AI assistance (7) | n/a | 0 | 0 | 0 |
| Total / 100 | n/a | 29.7 | 30.5 | 34.3 |
Year by year: 2019: 29.7 · 2020: 30.5 · 2021: 30.5 · 2022: 30.5 · 2023: 30.5 · 2024: 30.5 · 2025: 31.3 · 2026: 34.3
What moved the score
| Year | Domain | Change | Points | Trigger |
|---|---|---|---|---|
| 2020 | Ecosystem and integrations | 1.5 → 2 | +0.8 | CNCF Sandbox (2020-09) |
| 2025 | Ecosystem and integrations | 2 → 2.5 | +0.8 | CNCF Incubating (2025-08) |
| 2026 | Bare metal and day-2 operations | 3 → 3.5 | +0.8 | BMO 0.14.0: fast Redfish inspection mode, BareMetalSwitch CRD; firmware settings and RAID where the BMC supports them |
| 2026 | Orchestration and self-service | 1.5 → 2 | +1.0 | BMO 0.14.0: HostClaim controller |
| 2026 | State, planning and safety | 1.5 → 2 | +1.2 | BMO 0.14.0: URL validation webhooks, provisioning retry limit |
Direction, 2023 to 2026
Each release item carries one theme and one driver (V vision, C customer need, M market window, U upstream, P portfolio).
| Date | Item | Theme | Driver |
|---|---|---|---|
| 2025-08 | CNCF Incubating (accepted 2025-08-14, announced 2025-08-27) | Ecosystem | C |
| 2026-02 | OCI image deployment to bare metal with a custom IPA hardware manager (project blog) | Bare metal | C |
| 2026-05 | CAPM3 1.13.0; BMO 0.13 in the same release train removes the Fujitsu iRMC driver (iLO 4 and iLO 5 drivers went in 0.12) | Bare metal | P |
| 2026-06 | Copilot review Prow plugin for the project's own pull requests | AI | V |
| 2026-09 | BMO 0.14.0: fast Redfish inspection mode, provisioning retry limit, BareMetalSwitch CRD | Bare metal | C |
| 2026-09 | BMO 0.14.0: HostClaim controller, URL validation webhooks, BareMetalHost taints deprecated | Orchestration | C |
| 2026-09 | CAPM3 1.14.0: v1beta2 API | Language | U |
| 2026-10 | BMO 0.14.2 and 0.13.6 patch releases | Packaging | C |
- Centre of gravity: Bare metal (3), Ecosystem (1), AI (1)
- Blind spots (no items): Provisioning breadth, Configuration management, State and planning, Scale, Security, Licensing
- Driver mix: Vision 13%, Customer need 63%, Market window 0%, Upstream 13%, Portfolio 13%
- Idea-to-need index: 0.17 (0 = demand-pulled, 1 = thesis-pushed)
Credibility: 15.5 / 25
| Criterion | Score | Basis |
|---|---|---|
| Cadence | 3.5 | Two to three minors a year (CAPM3 1.13.0 on 2026-05-18, 1.14.0 on 2026-09-09) with regular patch releases; no published schedule |
| Roadmap transparency | 3.5 | Public documentation, version support table and open tracker; no dated roadmap |
| Say-do | 3 | Few commitments to check; the v1beta2 API landed in 1.14 |
| Velocity | 3 | Several features per BMO minor |
| Lifecycle stability | 2.5 | Two minors supported; the operator is still 0.x; BMC vendor drivers removed in 0.12 and 0.13 |
Sustainability risk: 7.5 / 25 (lower is better)
| Criterion | Score | Basis |
|---|---|---|
| Maintainer concentration | 2.5 | Led by Ericsson and Red Hat; 57 organisations contributed during the sandbox phase (secondary) |
| Backer health | 1.5 | CNCF Incubating since 2025-08; adopters named include SUSE, Ericsson, Red Hat, Fujitsu and Ikea (secondary) |
| Licence-change risk | 0.5 | Apache-2.0 under the CNCF |
| Abandonment signs | 1 | Steady releases through 2026; BMO 0.14.2 on 2026-10-15 |
| Security response | 2 | Process not read and no advisory record collected |
Technical lock-in: 8.0 / 25 (lower is better)
| Criterion | Score | Basis |
|---|---|---|
| Formats | 1 | Kubernetes YAML |
| Export path | 2 | Plain YAML; the Ironic underneath can be driven standalone, but no exporter exists |
| Stack coupling | 2.5 | Needs a Kubernetes management cluster and Ironic |
| Hardware / cloud coupling | 0.5 | Servers with IPMI or Redfish; offline capable |
| Skills and tooling coupling | 2 | Kubernetes and Cluster API skills |
Integrator fit: 7.5 / 20
| Criterion | Score | Basis |
|---|---|---|
| Partner programme openness | 1 | None; vendors embed the project |
| Multi-tenancy and self-service | 1.5 | Kubernetes namespaces and RBAC only |
| API and infrastructure-as-code | 3.5 | Kubernetes API and Cluster API provider |
| Skills and certification | 1.5 | None found beyond general Kubernetes training |
AI leverage: 2.0 / 20
How the product uses AI, scored per year like capability. Only GA counts.
| Criterion | 2019 | 2022 | Today | Latest step |
|---|---|---|---|---|
| AI code and plan generation | 0 | 0 | 0 | 2019: None |
| AI platform stack | 0 | 0 | 0 | 2019: None |
| AI-assisted operations | 0 | 0 | 0 | 2019: None |
| Agent openness | 2 | 2 | 2 | 2019: Kubernetes API and custom resources; no MCP server found |
Year by year: 2019: 2 · 2020: 2 · 2021: 2 · 2022: 2 · 2023: 2 · 2024: 2 · 2025: 2 · 2026: 2
AI say-do, 2023 to 2026: 0 AI items announced; 0 GA, 0 still preview, 0 pending, 0 dropped. No AI product items; the Copilot review plugin serves the project's own contributors.
Metal3 has no AI features. Agents can drive it through the Kubernetes API like any operator, and the only AI in the project is a code review plugin for its own pull requests.
Reading
Metal3 is the Kubernetes-native way to run Ironic, and its scores follow from that: strong bare-metal depth inherited from the engine, a continuous reconciliation model instead of a plan, and nothing outside bare metal. CNCF incubation in 2025 and a steady 2026 release train keep sustainability risk moderate, although the operator is still 0.x and each release has retired a BMC vendor driver. Integrator fit is the weakest in the group because nobody sells or certifies it directly. Check that your servers are on the regularly tested list (iDRAC 8+, iLO 5+, generic Redfish) before planning around it.
Open questions and items to re-score
- Release dates and features for 2021 to 2024 were not researched; early scores are inferred.
- The incubation date differs between the CNCF project page (2025-08-14) and the blog (2025-08-27).
- Which vendors support Metal3 commercially was not verified.
- Air-gapped operation and BMO upgrade rules were not read.
- The BMO 0.13 release month is inferred from CAPM3 1.13.0 (2026-05-18).
Related reading
More from Infrastructure Automation Ledger 2026
- Overview: all platforms side by side
- Infrastructure Automation Ledger Methodology: How Every Score Is Built
- Ansible AAP: Infrastructure Automation Ledger 2026 Scorecard
- Ansible: Infrastructure Automation Ledger 2026 Scorecard
- Atlantis: Infrastructure Automation Ledger 2026 Scorecard
- Bicep: Infrastructure Automation Ledger 2026 Scorecard
- CFEngine: Infrastructure Automation Ledger 2026 Scorecard
- Chef: Infrastructure Automation Ledger 2026 Scorecard
- Cinc: Infrastructure Automation Ledger 2026 Scorecard
- CloudFormation: Infrastructure Automation Ledger Scorecard
- Crossplane: Infrastructure Automation Ledger 2026 Scorecard
- Digger: Infrastructure Automation Ledger 2026 Scorecard
- env zero: Infrastructure Automation Ledger 2026 Scorecard
- Foreman: Infrastructure Automation Ledger 2026 Scorecard
- Harness: Infrastructure Automation Ledger 2026 Scorecard
- Ironic: Infrastructure Automation Ledger 2026 Scorecard
- Juju: Infrastructure Automation Ledger 2026 Scorecard
- MAAS: Infrastructure Automation Ledger 2026 Scorecard
- OpenTofu: Infrastructure Automation Ledger 2026 Scorecard
- OpenVox: Infrastructure Automation Ledger 2026 Scorecard
- Pulumi: Infrastructure Automation Ledger 2026 Scorecard
- Puppet: Infrastructure Automation Ledger 2026 Scorecard
- Salt: Infrastructure Automation Ledger 2026 Scorecard
- Satellite: Infrastructure Automation Ledger 2026 Scorecard
- Scalr: Infrastructure Automation Ledger 2026 Scorecard
- Semaphore: Infrastructure Automation Ledger 2026 Scorecard
- Spacelift: Infrastructure Automation Ledger 2026 Scorecard
- Terraform: Infrastructure Automation Ledger 2026 Scorecard
- Terragrunt: Infrastructure Automation Ledger 2026 Scorecard
- Tinkerbell: Infrastructure Automation Ledger 2026 Scorecard