Infrastructure as Code in 2026: Terraform, OpenTofu, and a Fractured Ecosystem

Infrastructure as Code stopped being a two-tool conversation (Terraform, or a cloud-native alternative) once HashiCorp changed Terraform's license from the permissive MPL 2.0 to the more restrictive Business Source License 1.1 in August 2023 (Pulumi Docs). That single decision triggered a community fork — OpenTofu, now governed by the Linux Foundation — and by 2026 it has genuinely fractured what used to be a default choice into a real three-way decision between Terraform, OpenTofu, and Pulumi (dasroot.net).

The licensing backdrop, and why it still matters in 2026

This isn't old news that's settled. A January 2026 survey found 56% of IaC users still consider the BUSL license shift disruptive to their tooling decisions, and it's cited directly as a factor pushing renewed interest toward Pulumi as well as toward OpenTofu (tech-insider.org). The acquisition dynamics compounded the concern: IBM completed its $6.4 billion acquisition of HashiCorp in February 2025, and since then HashiCorp has raised Terraform Cloud pricing at an average of 18% year-over-year — a hard number for enterprises deeply integrated with proprietary HCP features and therefore locked into the increases (Clanker Cloud).

OpenTofu: adoption is real but still early

OpenTofu offers what matters most for teams that want to stay on HCL without the licensing risk: an OSI-approved MPL 2.0 license, multi-vendor governance under the Linux Foundation, full provider compatibility with the existing Terraform ecosystem, and — notably — CLI features the closed Terraform binary doesn't have, including native state encryption and OCI-registry-based module distribution (DEV Community, tech-insider.org).

The adoption numbers, though, show this is still an early-majority story rather than a done migration: as of April 2026, OpenTofu sits at roughly 12% global adoption, with a further 27% of teams actively evaluating it (tech-insider.org). For teams already on Terraform who want to keep using HCL, the emerging consensus is that OpenTofu is the lower-risk default going forward — true open source, a compatible state format, and a faster feature-release cadence than Terraform's closed binary (Encore, tech-insider.org).

Where HCP Terraform still has an edge

HashiCorp isn't standing still on differentiation. Terraform Stacks, generally available since September 2025, orchestrate multiple infrastructure components as a single lifecycle unit — but this capability is exclusive to HCP Terraform, not available through the open CLI (Clanker Cloud). Project Infragraph, in private beta since December 2025, connects Terraform state directly to IBM watsonx, Ansible, OpenShift, and IBM's observability stack — a clear signal HashiCorp is leaning into IBM's enterprise ecosystem as its differentiator rather than competing purely on open-source feature parity (Clanker Cloud).

Pulumi: general-purpose languages instead of HCL

Pulumi occupies a distinct lane entirely — instead of a domain-specific language like HCL, it lets you define infrastructure in TypeScript, Python, Go, or other general-purpose languages you likely already use. Benchmark comparisons in 2026 cite Pulumi supporting roughly 4,800 providers versus Terraform's 1,800 (tech-insider.org) — though provider count isn't the same as provider maturity, so this number should be read as breadth, not necessarily depth, per provider.

The pattern that's emerged: engineering-led startups where application developers own infrastructure frequently choose Pulumi specifically so infra code lives in the same language and tooling as their application code, avoiding a context switch to HCL (env zero).

Three-way comparison

PR-driven and multi-tool orchestration platforms

Underneath the core tool choice, a separate layer of tooling has matured for managing IaC workflows regardless of which underlying tool you pick. Atlantis remains the open-source standard for PR-driven Terraform/OpenTofu workflows — it runs as a server receiving webhooks from GitHub or GitLab, executes terraform plan on every pull request, and posts the plan output as a PR comment for review before merge (Clanker Cloud). Spacelift has emerged as the most prominent commercial alternative, notable for supporting Terraform, OpenTofu, Pulumi, Ansible, and CloudFormation within a single platform — useful if your organization is running more than one IaC tool across teams rather than a single standardized choice (Clanker Cloud).

Typical Atlantis PR workflow:
1. Engineer opens PR with .tf changes
2. Atlantis webhook triggers `terraform plan`
3. Plan output posted as PR comment
4. Reviewer approves based on actual plan diff
5. `terraform apply` runs on merge (or via comment command)

Decision framework

1. Already on Terraform, no HCP-exclusive dependency, want lowest migration friction: OpenTofu is the practical default — state format compatibility means the migration is largely mechanical, and you eliminate the license/pricing risk entirely.
2. Deep, load-bearing dependency on HCP-only features (Stacks, Infragraph, IBM ecosystem integration): staying on HCP Terraform may be the pragmatic choice short-term, but treat the pricing trajectory as a cost to budget for, not a fixed number.
3. Greenfield project, application developers will own infrastructure, team already fluent in TypeScript/Python/Go: Pulumi removes the DSL context-switch and lets infra code go through the same review, testing, and type-checking tooling as application code.
4. Multi-tool environment across teams: evaluate a platform like Spacelift rather than trying to force organizational convergence on a single IaC tool before you're ready.

Sources: Pulumi Docs: OpenTofu vs Terraform, dasroot.net: Terraform vs OpenTofu vs Pulumi 2026 Comparison, tech-insider.org: OpenTofu vs Terraform vs Pulumi 2026, tech-insider.org: Pulumi vs Terraform 2026, Clanker Cloud: Terraform Latest Trends 2026, DEV Community: OpenTofu vs Terraform in 2026, Encore: OpenTofu vs Terraform in 2026, env zero: Pulumi vs Terraform vs OpenTofu Decision Framework