Infrastructure as code tools describe servers, networks and cloud services in files that are reviewed, versioned and applied by a program. The category now splits into provisioning engines, configuration managers and platforms that run them.
This guide ranks twelve on first-ninety-day effort, the bill once state, policy and runners are added, how work leaves, and whether you can run without a partner.
Vendors can pay for visibility on this page. It never changes what an entry
says about a product, including the criticism, and we earn nothing when you click through to a
vendor. How that works.
In short
What infrastructure as code tools software does
Infrastructure as code tools let teams declare cloud resources, servers and network settings in version-controlled files, then plan and apply those definitions automatically instead of changing environments by hand in consoles.
In this order: setup effort, what it really costs, how your data comes back out, whether
you can leave, and who each infrastructure as code tool is built for. One rule sits on top of those tests:
Places 1 to 3 go to challengers, vendors outside the best-known brands in a category, wherever the category has them; the established names follow from place 4 in their ranked order. Why those five, and why there is no
score out of ten, is on the how we work page.
Pay as you go with no minimum fees; registry extensions billed per handler operation
—
AWS-only teams that want provisioning without adding another vendor
Multi-cloud teams wanting one language across providers
Country is where the vendor is headquartered or contracts from, which is a
different question from where your data is hosted. Where the two tell different stories, the
entry says so.
Linux Foundation fork of Terraform under the MPL 2.0 licence
Ranked #1 of 12 in Best Infrastructure as Code Tools in 2026.
Open sourcePricing on requestElsewhere
OpenTofu answers the independence test more cleanly than anything else in this list: MPL 2.0 code, foundation stewardship and a public registry of 3,900+ providers and 23,600+ modules. The first ninety days are short for existing Terraform users, since configuration carries over.
The real price is the surrounding stack. Remote state, policy checks and plan approval come from a CI system or a platform such as Spacelift or env zero, so budget engineering hours rather than licence fees.
What stands out
Open source
Terraform fork
Community-run
Where it costs you
The FAQ names Terraform 1.5.x as the newest state-file version it promises to read
The project site shows no hosted service, so state storage, approvals and run scheduling need other tools
Right for
Teams that want a Terraform-style workflow without a commercial licence or vendor
Wrong for
Buyers who want a vendor-run service with support and SLAs
Not stated by the vendorOpen source under MPL 2.0; no vendor price list
General-purpose languages for infrastructure, with published cloud tiers
Ranked #2 of 12 in Best Infrastructure as Code Tools in 2026.
Free tierOpen sourcePublished pricingNorth America
The ladder is published, which most vendors here avoid. Free covers one user; Essentials starts at 40 dollars a month with 500 managed resources and unlimited users; Pro starts at 400 dollars with 2,000 resources; Enterprise starts at 2,000 dollars with 4,750.
Beyond the included credits, managed resources are metered per hour, and secrets, workflow minutes and AI tokens are separate meters. Model your resource count first, because the credit system makes the invoice depend on it.
What stands out
Open source SDK
Multi-language
Credit-based tiers
Where it costs you
The free plan allows a single member and one concurrent stack update
SAML/SSO and audit logs begin at the Pro tier, and self-hosting only at Enterprise
Right for
Developers who want loops, tests and types in a normal language
Wrong for
Operators who want a purely declarative file format
United StatesFree plan for one user; Essentials, Pro and Enterprise tiers published
Runs OpenTofu, Terraform, Pulumi and Kubernetes code from one pipeline
Ranked #3 of 12 in Best Infrastructure as Code Tools in 2026.
Free tierSelf-hostablePublished pricingElsewhere
Because it sits above the IaC engine, Spacelift lowers exit risk: your code stays in OpenTofu, Terraform or Pulumi form. The price is built from private workers.
Starter Plus lists one, Business three, Enterprise from five and Enterprise Plus from twelve, with a policy engine, drift detection and cost estimation included from Starter Plus. Self-hosted and air-gapped operation needs the top tier. Ask what an extra worker costs before assuming the published figure holds.
What stands out
Orchestration layer
Private workers
Self-hosted tier
Where it costs you
The free plan is limited to two users and one public worker
Business, Enterprise and the self-hosted tier are quoted, not published
Right for
Teams already on OpenTofu or Terraform that need approvals and policy
Wrong for
Individuals or small teams that only need a local CLI
Not stated by the vendorFree plan for two users; Starter Plus published, higher tiers quoted
Environment-based IaC automation with a usage-capped free tier
Ranked #4 of 12 in Best Infrastructure as Code Tools in 2026.
Free tierSelf-hostablePublished pricingNorth America
Each environment maps to one Terraform workspace, and native support covers Terraform and Terragrunt. That makes the metering easy to understand and awkward to forecast: a team with many short-lived preview environments consumes runs fast.
The free plan is a real trial of drift detection and cost estimation within its caps, and a runner can be self-hosted. Count destroy and drift runs too, since the free tier meters them, not only applies.
What stands out
Per-apply pricing
Self-hosted runner
Terragrunt support
Where it costs you
The free plan stops at 250 runs, 30 active environments and 20 minutes per run
Paid plans are quoted per apply or environment with no public price
Right for
Platform teams that want self-service environments with time-to-live controls
Wrong for
Buyers who need to see the paid price before a sales call
United StatesFree plan up to 250 runs a month; paid plans quoted per organisation
Declarative provisioning tool under IBM with per-resource cloud billing
Ranked #5 of 12 in Best Infrastructure as Code Tools in 2026.
Self-hostablePublished pricingNorth America
HashiCorp is now part of IBM, and the pricing page carries IBM branding. HCP Terraform lists Essentials from 0.10 dollars per resource a month, Standard from 0.47 and Premium from 0.99, with a 500 dollar trial credit.
The docs also describe free organisations capped at 500 managed resources. An EU service locality is offered by quote. The cost to check is the resource count: a large estate multiplies that per-resource rate quickly, so estimate before committing.
What stands out
Source-available BUSL
IBM-owned
Per-resource billing
Where it costs you
The licence is source-available and carries a competitive-use restriction
Per-resource billing grows with every managed object, and Enterprise has no public price
Right for
Organisations already standardised on Terraform modules and HashiCorp tooling
Wrong for
Teams that need an OSI open-source licence or a vendor-neutral owner
United StatesHCP Terraform billed per managed resource, rates published; self-hosted Enterprise quoted
Agentless YAML automation, with Red Hat support sold per quote
Ranked #6 of 12 in Best Infrastructure as Code Tools in 2026.
Open sourcePricing on requestNorth America
The community core costs nothing to adopt and needs only SSH access, which keeps the first ninety days to writing playbooks. The platform is where money enters: self-managed on your own infrastructure, a managed service run by Red Hat, or a managed application in a public cloud, each sized by quote.
With the Red Hat site as the only price source, plan a procurement conversation. Playbooks are plain YAML, so leaving costs a rewrite of logic, not an export.
What stands out
Agentless
GPL v3 core
Self-managed or managed
Where it costs you
The pricing page shows no amounts, only a quote based on sizing and subscription
Automation Platform is offered self-managed, as a managed service or via cloud marketplaces, each sized by quote
Right for
Teams configuring servers and network gear who want no agent to install
Wrong for
Teams that want a recorded plan and stored state before every change
United StatesOpen source core; Automation Platform quoted, Standard or Premium support
Open-source wrapper that orchestrates OpenTofu and Terraform stacks
Ranked #7 of 12 in Best Infrastructure as Code Tools in 2026.
Open sourcePublished pricingNorth America
The site names its aims as segmenting infrastructure into units with their own state, ordering updates through a run queue and reusing configuration through includes and stacks.
The MIT licence and a wrapper design keep independence high, since your modules stay ordinary. The trade is complexity: two tools in every pipeline, and a wrapper vocabulary new hires must learn. Test it on one layered environment before adopting it across a monorepo.
What stands out
Open source
Wrapper tool
DRY configuration
Where it costs you
It needs OpenTofu or Terraform underneath, so it adds a layer without replacing one
The hosted pipeline is free only up to 25 infrastructure units
Right for
Teams with many environments repeating the same OpenTofu or Terraform code
Wrong for
Small projects with a single state and few modules
United StatesMIT-licensed tool; optional Gruntwork pipeline free up to 25 units
Kubernetes-based control planes for cloud resources, a CNCF graduated project
Ranked #8 of 12 in Best Infrastructure as Code Tools in 2026.
Open sourcePricing on requestElsewhere
The model differs from the file-and-apply tools above it: resources are Kubernetes objects reconciled continuously, packaged as providers, functions and configurations.
Independence is strong, with an Apache 2.0 licence and CNCF governance, but the learning curve is real: compositions, composite resource definitions and operations are concepts to master. Upbound describes its product as built on Crossplane and offers demos. Expect the first ninety days to go on designing APIs, not provisioning.
What stands out
CNCF graduated
Kubernetes-native
Apache 2.0
Where it costs you
It runs inside Kubernetes, so a cluster must exist before any cloud resource is managed
The project site shows no pricing, and the commercial edition belongs to Upbound
Right for
Platform teams building self-service APIs on top of existing Kubernetes clusters
Wrong for
Small teams without Kubernetes experience or a cluster to spare
Not stated by the vendorOpen source under Apache 2.0; commercial platform from Upbound
Event-driven configuration and remote execution framework from the Salt Project
Ranked #9 of 12 in Best Infrastructure as Code Tools in 2026.
Open sourcePricing on requestElsewhere
Salt's documentation highlights an event system carried over ZeroMQ, a master that publishes jobs, minions that run them, and states written in YAML. That design suits speed across many machines.
The unknown is commercial support: the project site names Broadcom in its copyright but offers no pricing page, so ask before assuming a supported edition exists. Exit is straightforward on licence grounds, though state files and pillar data would need rewriting for another tool.
What stands out
Open source
Master and minion
YAML states
Where it costs you
The project site lists no plans, prices or support tiers
The documentation is organised around machines, so cloud-resource provisioning is not its stated focus
Right for
Teams managing large fleets of servers who want fast remote execution
Wrong for
Buyers needing a published support offer or cloud provisioning
Not stated by the vendorOpen source; the project site shows no vendor price list
Agent-based configuration management from Perforce with a small-fleet free tier
Ranked #10 of 12 in Best Infrastructure as Code Tools in 2026.
Free tierOpen sourcePublished pricingNorth America
The plan structure is simple to read and hard to price: Core Developer is free under 25 nodes, Commercial is custom from 25 nodes, and both higher platforms are custom. Standard support is included, with 24x7 premium available.
The compare table lists continuous CIS Benchmark enforcement as Enterprise material, so a compliance buyer should budget above Core. Perforce is based in Minneapolis, a fact to state in procurement files.
What stands out
Perforce-owned
Puppet Forge
Node-based licensing
Where it costs you
Every price above 24 nodes is a quote, and Core's CIS enforcement is limited to the commercial version
The web interface and role-based access control sit in Enterprise, not Core
Right for
Teams enforcing server configuration at scale with audit and compliance needs
Wrong for
Teams provisioning cloud accounts rather than configuring operating systems
United StatesCore free below 25 nodes; commercial and Enterprise tiers quoted
Per-node configuration automation from Progress, in SaaS or on-prem
Ranked #11 of 12 in Best Infrastructure as Code Tools in 2026.
Self-hostablePublished pricingNorth America
Chef's price page is the plainest in the configuration-management group: 59 dollars per node per year for Business, 189 for Enterprise, and a quote above. A 500-node fleet on Enterprise would therefore cost about 94,500 dollars a year, so size carefully.
Deployment is flexible, covering SaaS, on-prem, hybrid and air-gapped setups. The repository for Chef carries Apache 2.0, but the commercial platform is Progress's product. Ask which edition the licence price actually covers.
What stands out
Progress-owned
Per-node billing
SaaS or on-prem
Where it costs you
Billing is per node, so cost scales with fleet size
The trial is by request, not self-service, and Enterprise Plus has no public price
Right for
Enterprises managing compliance and jobs across many nodes who can budget per node
Wrong for
Teams that want a self-service trial or a free entry plan
United StatesPer node per year, Business and Enterprise published; Enterprise Plus quoted
Template-based provisioning service for AWS and third-party resources via registry
Ranked #12 of 12 in Best Infrastructure as Code Tools in 2026.
Pricing on requestNorth America
Setup is minimal for an existing AWS customer: create a template, create a stack, and CloudFormation provisions the resources in order. The cost is mostly invisible until registry extensions enter, since third-party and private resource types bill per create, update, delete, read or list call.
The vendor is the cloud provider itself, so independence means living with AWS terms. Rank it last here only because buying and exit reward tools that outlive one cloud.
What stands out
AWS-native
Template stacks
No extra servers
Where it costs you
A template is capped at 500 resources, which forces nested stacks on large estates
Templates are AWS-specific, so moving to another tool means rewriting them
Right for
AWS-only teams that want provisioning without adding another vendor
Wrong for
Multi-cloud teams wanting one language across providers
United StatesPay as you go with no minimum fees; registry extensions billed per handler operation
How to choose infrastructure as code tools software
Infrastructure as code tools let teams declare cloud resources, servers and network settings in version-controlled files, then plan and apply those definitions automatically instead of changing environments by hand in consoles. The differences that matter are rarely in the feature list, so this is
the order we would work through them.
01
Decide whether you need a published price
7 of the 12 tools here publish what they cost; the other 5 quote per organisation. The ones you can compare without a sales call: Pulumi, Spacelift, env0, Terraform, Terragrunt, Puppet, Chef.
02
Decide how much the jurisdiction matters
These 12 vendors are established in 2 countries across 2 regions (North America 8, Elsewhere 4). That decides whose disclosure law applies to what the vendor holds, wherever the servers are.
03
Consider whether you want the source
7 of these are open source: OpenTofu, Pulumi, Ansible, Terragrunt, Crossplane, SaltStack, Puppet. Hosting one yourself trades a subscription for maintenance.
Provisioning engines and configuration managers do different jobs
OpenTofu, Terraform and Pulumi create and change cloud resources and keep a record of what they made. Ansible, Puppet, Chef and SaltStack configure machines that already exist. Ansible's documentation describes agentless playbooks over SSH that restore a declared state, while Puppet's pricing page describes agent-based and agentless setups.
Many teams need one of each, and buying a configuration manager to provision a cloud account is the commonest category error. Decide first whether your pain is creating infrastructure or keeping servers consistent, because that choice removes half this list before price enters. Crossplane belongs to a third group, where Kubernetes itself reconciles cloud resources, so it only suits teams already running clusters.
List what you create (accounts, networks, databases) and what you configure (operating systems, packages)
Check whether the tool keeps a state record or only re-runs declared steps
Confirm whether the tool needs an agent or a cluster on every target
What OpenTofu, Terraform and Pulumi cost once state is hosted
The engines are cheap or free to run yourself, and the bills start with the hosted layer. Terraform's HCP service charges per managed resource, from 0.10 dollars a month on Essentials to 0.99 on Premium, with Terraform Enterprise quoted. Pulumi publishes credit-based tiers with a one-user free plan and 500 managed resources on Essentials.
OpenTofu has no vendor price list, which moves state storage and approvals to your own CI or to a platform such as Spacelift. Compare by multiplying your real resource count by each metered rate, then add the engineering hours a self-built pipeline needs. Licence differs too: Terraform is under the Business Source License, OpenTofu under MPL 2.0.
Count managed resources in production today and in twelve months
Read the licence text, not the label, for competitive-use clauses
Price the CI pipeline hours you would otherwise save
Orchestration platforms meter workers, runs or environments
Spacelift and env0 do not replace the engines, they run them. Spacelift's plans scale by private workers, from one on Starter Plus to five or more on Enterprise, and the free plan allows two users. env zero, the current name of env0, meters runs and environments: 250 runs a month and 30 active environments on the free plan, then quotes per successful apply or environment.
A team with many preview environments and short deployments pays very differently under each model. Terragrunt sits at another layer, wrapping OpenTofu or Terraform to organise units and run order, and Gruntwork's pipeline is free up to 25 infrastructure units. Ask each vendor for a quote built from your own deployment counts.
Count weekly applies, destroys and drift scans separately
Ask what an extra worker or environment costs
Check whether self-hosted runners need the top tier
Leaving matters more than starting
Exit is the least-tested part of an IaC decision. OpenTofu states that it reads Terraform state files only up to version 1.5.x, so a move from newer Terraform releases is not guaranteed. CloudFormation templates are an AWS format with a 500-resource cap per template, so leaving means rewriting.
Pulumi and Terragrunt keep your logic in languages and modules you own. Chef bills per node per year, which makes a shrinking fleet the only way to lower the cost. Before signing, test an export or a dry migration on one small environment, and keep modules in plain repositories your team controls.
Run a trial migration of one module into a second tool
Confirm which state formats the target tool reads
Keep a copy of modules and state outside any vendor account
What goes wrong most often when buying infrastructure as code tools software
Buying a configuration manager such as Puppet or Chef to provision cloud accounts, which the pricing pages do not describe.
Assuming Terraform and OpenTofu will stay interchangeable, when OpenTofu promises state compatibility only up to 1.5.x.
Comparing a free plan's caps with a paid tier's published price without counting runs, resources and workers.
Adopting Crossplane or Terragrunt as a first tool, when each assumes a cluster or an existing OpenTofu or Terraform setup.
07
Frequently asked questions
8 answers
What is the best infrastructure as code tools in 2026?
OpenTofu leads our ranking of 12. The only entry here with no vendor to contract with: the project sits under the Linux Foundation's stewardship and the licence is MPL 2.0, so you can run, fork and leave without a partner.
It reads existing Terraform configuration, and its FAQ promises state-file compatibility only up to Terraform 1.5.x. State storage, approvals and scheduling are for you to assemble elsewhere.
Which infrastructure as code tools publish their pricing?
7 of the 12, with the pricing model each one publishes:
Pulumi: Free plan for one user; Essentials, Pro and Enterprise tiers published.
Spacelift: Free plan for two users; Starter Plus published, higher tiers quoted.
env0: Free plan up to 250 runs a month; paid plans quoted per organisation.
Which infrastructure as code tools can you host yourself?
Spacelift, env0, Terraform, Chef. The other 8 are hosted by the vendor only.
What should you use instead of OpenTofu?
Pulumi and Spacelift are the next two on this page. Pulumi is for Developers who want loops, tests and types in a normal language; Spacelift is for teams already on OpenTofu or Terraform that need approvals and policy.
Who should not buy OpenTofu?
Buyers who want a vendor-run service with support and SLAs. The FAQ names Terraform 1.5.x as the newest state-file version it promises to read.
If your infrastructure as code tools product belongs among these 12, tell us what it does and who it is for. Inclusion is an editorial call; what a listing is and is not is set out under software advice.
