MAAS: Infrastructure Automation Ledger 2026 Scorecard

Part of the Infrastructure Automation Ledger (edition 2026-10-16). How the scores work: methodology.

Snapshot

Type: product
Owner: Canonical Ltd (UK, privately held)
Licence: AGPL-3.0; support per managed machine per year: free self-support, Essential US$30, Standard US$50, Advanced US$100; controllers need Ubuntu Pro (US$500 per machine per year); HA, RBAC and SLAs listed as paid-tier features
Latest release: 3.7 (December 2025); 3.7.3 (2026-08-11); 3.8.0-rc1 (2026-08-31) with no GA by 2026-10-15
Support lifecycle: Four most recent versions supported: 3.7 and 3.6 fully, 3.5 and 3.4 bug fixes only; no LTS; no end dates published; deb packages tied to the matching Ubuntu release
Next milestone: 3.8 (release candidate since 2026-08-31, no GA date)
Targets: Bare metal over IPMI and Redfish, including NVIDIA BlueField-3 DPUs; VMs on LXD and libvirt hosts; Proxmox VE as a power target only; can image ESXi onto bare metal; no public cloud, vSphere, Nutanix, Hyper-V or XCP-ng API; Windows, RHEL and ESXi images are Packer templates (mostly Beta), not shipped images; Terraform provider from Canonical (2.10.0, 2026-09-09)
Tier: Host and bare-metal provisioning
Archetype: Dependable builder
Evidence grade: B. Release notes for 3.4 to 3.7, the supported versions page, the licence file, the price list and the Terraform provider feed were read from Canonical pages and GitHub. Scale claims are vendor case studies. Releases before 2.8 (2020) were not found, so 2015 to 2020 capability is inferred and held flat where no dated trigger exists.

Scope of the scores: MAAS has no plan or dry-run mode of its own. Desired state is held as machine records and the Terraform provider from Canonical supplies plan and refresh for MAAS objects; drift after deployment is not detected. Scored with that limit, which keeps State, planning and safety low.

Scorecard

Dimension	Score	Direction
Capability today	41.7 / 100	higher is better
Momentum since end of 2022	+5.4 points (2015 to 2022: +1.5)	higher is better
Credibility	15.0 / 25	higher is better
Commercial risk	9.0 / 25	lower is better
Technical lock-in	8.0 / 25	lower is better
Integrator fit	11.0 / 20	higher is better
AI leverage	2.0 / 20	higher is better
Ledger Index	47.6 / 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)	1	1	1.5	1.5
Configuration management depth (12)	1	1	1	1
State, planning and safety (12)	1.5	1.5	1.5	2
Language and authoring (10)	2	2	2	2
Orchestration and self-service (10)	2.5	2.5	2.5	2.5
Scale and control plane (8)	2.5	2.5	2.5	3.5
Security and compliance (10)	2	2	2	2.5
Bare metal and day-2 operations (8)	3.5	3.5	3.5	4
Ecosystem and integrations (8)	2	2	2	2.5
AI assistance (7)	0	0	0	0
Total / 100	34.8	34.8	36.3	41.7

Year by year: 2015: 34.8 · 2016: 34.8 · 2017: 34.8 · 2018: 34.8 · 2019: 34.8 · 2020: 34.8 · 2021: 36.3 · 2022: 36.3 · 2023: 38.1 · 2024: 40.5 · 2025: 40.5 · 2026: 41.7

What moved the score

Year	Domain	Change	Points	Trigger
2021	Provisioning breadth	1 → 1.5	+1.5	3.0: LXD projects as VM hosts, Proxmox VE power driver
2023	Scale and control plane	2.5 → 3	+0.8	3.3: machine list usable above 10,000 machines
2023	Security and compliance	2 → 2.5	+1.0	3.3: HashiCorp Vault integration for secrets
2024	Bare metal and day-2 operations	3.5 → 4	+0.8	3.5: ephemeral in-memory deployments, soft power off, release scripts; Packer templates for SLES and RHEL 9 in 3.4
2024	Ecosystem and integrations	2 → 2.5	+0.8	3.5: Prometheus metrics; Terraform provider maintained by Canonical
2024	Scale and control plane	3 → 3.5	+0.8	3.5: Temporal-based process model, Go agent, boot resources moved to region controllers
2026	State, planning and safety	1.5 → 2	+1.2	terraform-provider-maas 2.7 to 2.10 (Canonical): plan and refresh for MAAS objects

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
2023-02	3.3: HashiCorp Vault integration for secrets	Security	C
2023-02	3.3: machine list usable above 10,000 machines; native Ubuntu 22.04 package	Scale	C
2024-01	3.4: redesigned UI, session timeout	Orchestration	C
2024-01	3.4: Packer templates for SLES, openSUSE and RHEL 9 images	Bare metal	C
2024-07	3.5: Temporal-based process model and Go agent; boot resources moved to region controllers (one way)	Scale	V
2024-07	3.5: ephemeral in-memory deployments, soft power off, release scripts	Bare metal	C
2024-07	3.5: Prometheus metrics and consolidated logs	Ecosystem	C
2025-04	3.6: Ubuntu 24.04 base, PostgreSQL 16; PostgreSQL 14 and DHCP snippets deprecated	Packaging	U
2025-04	3.6: reserved IPs, kernel crash dumps, Grafana dashboard	Bare metal	C
2025-12	3.7: NVIDIA BlueField-3 DPU commissioning and deployment	Bare metal	M
2025-12	3.7: own recursive DNS resolver in rack-only mode; OpenStack Nova power driver removed	Provisioning breadth	P
2026-08	3.8.0-rc1; no GA by 2026-10-15	Packaging	C
2026-09	terraform-provider-maas 2.10.0 (Canonical)	State and planning	C

Centre of gravity: Bare metal (4), Scale (2), Packaging (2)
Blind spots (no items): Configuration management, Language, AI, Licensing
Driver mix: Vision 8%, Customer need 69%, Market window 8%, Upstream 8%, Portfolio 8%
Idea-to-need index: 0.09 (0 = demand-pulled, 1 = thesis-pushed)

Credibility: 15.0 / 25

Criterion	Score	Basis
Cadence	3	Stated six-month rhythm; actual gaps of 6, 9 and 8 months between 3.4, 3.5, 3.6 and 3.7; dates not published in advance
Roadmap transparency	3	Public repository, release notes and supported versions table; no dated roadmap
Say-do	3	Little announced in advance; deprecations (DHCP snippets, PostgreSQL 14) are tracked; 3.8 has sat at rc1 since 2026-08-31
Velocity	3	Architecture work (Temporal, Go agent) and hardware support in each release
Lifecycle stability	3	Four versions supported, two of them fully; no LTS; deb packages tied to the Ubuntu release; the 3.5 boot resource migration cannot be reversed

Commercial risk: 9.0 / 25 (lower is better)

Criterion	Score	Basis
Price and licence volatility	1.5	Per-machine tiers (US$30, 50 and 100 per year) and Ubuntu Pro; no change found, but no price history was found either
Purchase constraints	2	Free self-support for any number of machines; HA, RBAC and SLAs listed as paid-tier features; controllers need Ubuntu Pro for support
Channel and access	1.5	Open source, snaps and images; support direct from Canonical
Owner stability	1.5	Privately held single vendor; no ownership change found
Cost of staying supported	2.5	Only the four latest versions are supported, about two years; Ubuntu Pro required for controllers

Technical lock-in: 8.0 / 25 (lower is better)

Criterion	Score	Basis
Formats	1	Standard images with curtin and cloud-init; open Packer templates
Export path	2	REST API, CLI and Terraform provider; no migration tooling to another bare-metal manager
Stack coupling	2.5	Ubuntu centred: debs match an Ubuntu release, controllers under Ubuntu Pro; one-way boot resource migration
Hardware / cloud coupling	1	Any server with a BMC; offline image mirror supported
Skills and tooling coupling	1.5	Ubuntu, cloud-init and curtin skills are common

Integrator fit: 11.0 / 20

Criterion	Score	Basis
Partner programme openness	2.5	Open source and freely deployable; Canonical partner terms not researched
Multi-tenancy and self-service	2	Resource pools and RBAC (paid tiers); no requester portal
API and infrastructure-as-code	4	Full REST API, CLI and a Canonical-maintained Terraform provider
Skills and certification	2.5	Ubuntu training; no MAAS-specific certification found

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	2015: None
AI platform stack	0	0	0	2015: None
AI-assisted operations	0	0	0	2015: None
Agent openness	2	2	2	2015: REST API, CLI and Terraform provider; no MCP server found

Year by year: 2015: 2 · 2016: 2 · 2017: 2 · 2018: 2 · 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 items announced; BlueField-3 DPU support serves AI hosts but is not AI assistance.

MAAS has no AI features and no official agent interface beyond its REST API. Its AI relevance is on the hardware side: 3.7 commissions and deploys NVIDIA BlueField-3 DPUs.

Reading

MAAS does one job, bare-metal commissioning and deployment, and does it with a clear support price list, which is rare in this group. Its capability total is low because the Ledger scores the whole automation stack and MAAS has no configuration engine, no plan mode and no cloud or hypervisor API beyond LXD and libvirt; the Terraform provider from Canonical is the only declarative path. Credibility is middling: the stated six-month rhythm has slipped to eight or nine, and 3.8 has been a release candidate since August 2026. The one thing to check is which features your estate needs from the paid tiers, because HA and RBAC are listed there.

Open questions and items to re-score

3.8 contents and GA date are unpublished; the release candidate has been out since 2026-08-31.
Releases before 2.8 (2020) were not found; 2015 to 2020 scores are inferred.
Price history since 2023 was not found; the current list is undated.
Exact gating of HA and RBAC between the open source code and the paid tiers was not verified beyond the vendor page.
Windows, RHEL and ESXi deployments rely on Packer templates marked Beta or Alpha, not shipped images.
No AI or MCP plans found; re-check at 3.8.

	’21	’23	’24	’26
Provisioning breadth	1
Configuration management depth
State, planning and safety				1
Language and authoring
Orchestration and self-service
Scale and control plane		2	1
Security and compliance		1
Bare metal and day-2 operations			2
Ecosystem and integrations			3
AI assistance