The Relicensing and Its Stated Logic
On 10 August 2023, HashiCorp announced that Terraform and several of its other products — including Vault, Consul, and Nomad — would move from the Mozilla Public Licence 2.0 to the Business Source Licence, known as BUSL. The MPL-2.0 is an OSI-approved weak copyleft licence; the BUSL is not OSI-approved and restricts commercial use for a defined period before converting to a more permissive licence. HashiCorp set the conversion term at four years, meaning each new Terraform release would become MPL-2.0 only in 2027 or later.
The company's stated rationale was competitive: managed-service providers were, in its view, wrapping Terraform in commercial products without contributing back to the project. HashiCorp did not name specific vendors publicly, though the infrastructure management space had grown considerably more crowded since Mitchell Hashimoto and Armon Dadgar founded the company in San Francisco in 2012. The change was effective immediately for new code; existing MPL-2.0 releases remained under that licence.
The BUSL originated at MariaDB, which promoted it as an instrument for commercial open-source companies looking to close the gap between permissive licensing and full proprietary control. HashiCorp was not the first to reach for it — MariaDB had experimented with it earlier — but the scale and visibility of Terraform made this the highest-profile BUSL deployment to date. Terraform is embedded in the infrastructure workflows of thousands of organisations; its provider ecosystem runs to hundreds of integrations.
What the move foreclosed was straightforward: no competing managed-service provider could legally offer Terraform-as-a-service using post-August 2023 code without a commercial agreement with HashiCorp. What it did not address was the question of community contributors who had submitted code under the expectation of an open-source licence. A Contributor Licence Agreement had required contributors to grant HashiCorp broad rights over their submissions — the CLA being the legal instrument that makes unilateral relicensing possible. Without that CLA, HashiCorp could not have acted at all.
The Fork and the Foundation
The community response arrived quickly. Within days of the announcement, a coalition of companies — including Gruntwork, Spacelift, env0, Harness, and others with commercial interests in the Terraform ecosystem — began coordinating a fork. The OpenTofu project was publicly announced in September 2023, presented as a community-governed, truly open-source continuation of the last MPL-2.0 Terraform codebase.
- CHRONOLOGYAS RECORDED
- 10 August 2023HashiCorp announces Terraform relicensed from MPL-2.0 to BUSL
- September 2023OpenTofu fork publicly announced; joins the Linux Foundation; community interest in the fork grows
- January 2024OpenTofu 1.6.0 ships as first stable release; declared generally available
- April 2024IBM announces acquisition of HashiCorp; closes February 2025
- April 2024 OpenTofu 1.7introduces features not present in HashiCorp's Terraform at time of release
The fork joined the Linux Foundation in September 2023, giving it a neutral governance home — an important signal that this was not simply one commercial rival appropriating another's work. The Linux Foundation had played a similar role in the Redis relicensing dispute, when it shepherded the Valkey fork after Redis Ltd abandoned the BSD licence in March 2024. The OpenTofu Technical Steering Committee was established with representatives from multiple contributing organisations, rather than a single controlling vendor.
By January 2024, the project had shipped OpenTofu 1.6.0 — the first stable release under the new name — and declared general availability. The release was built from the MPL-2.0 codebase and carried that licence forward. Subsequent releases added features that diverged from HashiCorp's own Terraform roadmap: 1.7 introduced early-evaluation syntax and provider-defined functions that had no equivalent in Terraform at the time. The fork was not merely preserving; it was developing.
The Open Source Initiative weighed in to confirm what most practitioners already understood: the BUSL does not meet the Open Source Definition, and Terraform was no longer, in any meaningful sense, open source. Voices including Bruce Perens and Simon Phipps had long argued that source-available licences marketed with open-source adjacency caused governance confusion; the Terraform episode added another data point. The OSI's position carried particular weight because HashiCorp had, in its announcement, described the BUSL as allowing its code to remain "open" — a framing the community contested sharply.
Acquisition and Its Implications
The story acquired a further dimension in April 2024, when IBM announced it would acquire HashiCorp for approximately $6.4 billion, a deal that closed in February 2025. IBM's portfolio already included Red Hat, a company whose entire commercial model rests on open-source software and enterprise subscriptions. The combination made the relicensing look, in retrospect, like preparation for acquisition rather than simply a reaction to competitive pressure — though HashiCorp did not characterise it that way, and the timing may reflect coincidence as much as strategy.
For OpenTofu, the IBM acquisition changed the governance picture. Questions that were previously bilateral — HashiCorp's commercial interests versus community expectations — became questions about a much larger corporation's infrastructure product roadmap. IBM has not moved Terraform back to an OSI-approved licence. HashiCorp's enterprise product, Terraform Cloud and its successor branding, continues under terms that diverge from what the open-source community maintains in OpenTofu.
The OpenTofu project, for its part, has had to manage a more immediate problem: codebase divergence. As HashiCorp's Terraform and OpenTofu each ship new features, the distance between them grows. Providers written against Terraform's evolving API surface may work differently against OpenTofu's. The project has stated a commitment to compatibility, but compatibility is not identity, and organisations choosing between the two are making a longer-term bet than a simple licence swap would imply.
What the Case Demonstrates
The Terraform episode is, structurally, one of the cleaner examples of how open-source relicensing and community forking interact. The MPL-2.0 baseline gave the community a legitimate, unambiguous starting point for the fork. The CLA regime gave HashiCorp the authority to relicense. The Linux Foundation gave OpenTofu the institutional credibility to attract contributors without subordinating them to any single commercial interest. All three mechanisms worked as designed — and they produced two diverging ecosystems where there had been one.
The broader pattern is not unique to Terraform. Elastic moved Elasticsearch to the SSPL in January 2021 before eventually returning to the AGPL in 2024; the AWS-backed OpenSearch fork has continued in parallel. MongoDB authored the SSPL in 2018 partly to prevent exactly the kind of managed-service competition HashiCorp cited. Each of these cases involves a profitable infrastructure company concluding that permissive licensing is subsidising competitors rather than building community. Each has produced a fork or an alternative, and in no case has the fork fully displaced the original.
What distinguishes the Terraform case is the speed and organisation of the community response, and the scale of the project being forked. Terraform's provider ecosystem — hundreds of integrations, millions of lines of community-contributed code — is not trivially replicated. OpenTofu's continued development suggests the fork is viable. Whether viability translates into dominance, or whether both tools coexist indefinitely serving different constituencies, remains the open question that August 2023 set in motion.