The Fork That Actually Stuck
Two and a half years ago, HashiCorp’s pivot to the Business Source License felt like corporate theater. The kind of move that generates angry Hacker News threads and gets forgotten by October. I was skeptical. Most forks die in obscurity, maintained by three people with strong opinions and limited bandwidth. Yet here we are in early 2026, and OpenTofu has stopped being a protest project and started being a legitimate alternative that enterprises are actually choosing.
The Linux Foundation takeover gave OpenTofu something forks usually lack: institutional credibility and realistic funding. More importantly, it gave the project a separation from the original maintainers’ vision that freed them to make different technical decisions. When HashiCorp’s acquisition by IBM completed in mid-2024 for $6.4 billion, uncertainty rippled through enterprise infrastructure teams. Not because IBM is inherently hostile to open source, but because the incentive structure changed. Terraform is no longer the flagship. It’s a component in IBM’s portfolio. That matters.
The Substance Beneath the Licensing Drama
Here’s what surprised me most about tracking OpenTofu’s evolution: the team didn’t just mirror Terraform’s feature set and call it done. Late 2025’s 1.9 release introduced provider-defined functions and state encryption that Terraform’s current releases still don’t have. This isn’t vaporware. This is OpenTofu official documentation and changelog showing architectural decisions that prioritize different customer segments than HashiCorp’s newer roadmap apparently does.
State encryption deserves a paragraph of its own because it’s the kind of feature that makes infrastructure teams sleep better at night. Encrypted state files that don’t require external vaults for basic confidentiality protection. It’s table stakes for regulated industries, and the fact that OpenTofu shipped this while Terraform hasn’t says a lot about whose priorities are being served. The provider-defined functions feature is equally telling: it extends what’s possible without requiring framework-level changes, giving provider maintainers more flexibility.
The registry numbers back this up too. OpenTofu hit 2,000 mirrored providers by early 2026. That’s not theoretical capacity. That’s actual ecosystem depth. The project’s GitHub repository crossed 23,000 stars, and for several consecutive months, the commit frequency exceeded HashiCorp’s Terraform. When was the last time you saw a fork win on velocity metrics against the original?
Where Enterprise Teams Actually Are
Adoption data gets squirrelly because vendors have incentives to spin it favorably, but the Spacelift survey from 2025 showed something concrete: 34 percent of infrastructure teams actively evaluating OpenTofu completed a full migration. Another 28 percent reported partial migration in progress. That’s not overwhelming, but it’s also not fringe. Those are real teams running real workloads on production infrastructure, betting their deployment reliability on this fork.
What’s driving the movement? Licensing uncertainty tops the list. Pulumi’s 2025 infrastructure report found that 41 percent of platform engineers cited IaC tool licensing as a top-three risk factor in their toolchain planning. When your infrastructure automation layer has licensing ambiguity hanging over it, you start making contingency plans. You test alternatives. And if the alternative works, you migrate.
The migration itself matters less than what it signals. Every team that moves from Terraform to OpenTofu is effectively saying: we’d rather maintain compatibility with an open-source governance model than bet on a commercial vendor’s long-term commitment to openness. That’s not a technical preference. That’s an institutional risk calculation.
The Honest Assessment: It’s Not About Being Better
I need to be direct here because I’ve seen too many infrastructure teams make tool decisions based on marketing framing rather than actual requirements. OpenTofu isn’t objectively better than Terraform. Both deploy infrastructure effectively. Both have extensive provider ecosystems. Both solve the core problem of managing cloud resources declaratively.
The decision matrix changed. What used to be a clear winner on maturity and ecosystem depth is now a choice between a commercial product with license uncertainty and an open-source project with governance transparency. For many teams, especially those in regulated industries or with strong open-source preferences, that’s enough to tip the scales. For others, Terraform’s deeper enterprise integration with HashiCorp’s broader product suite still makes sense.
The real insight is this: in 2024, choosing Terraform was the obvious default. In 2026, it requires justification. You have to articulate why you’re comfortable with the licensing trajectory and IBM’s long-term vision. That’s not a flaw in Terraform. It’s just how the board shifted when the vendor structure changed.
What Actually Matters in Your Decision
If you’re evaluating this decision for your infrastructure, ask the questions that matter: What’s your regulatory environment? Do you need state encryption out of the box? How dependent are you on HashiCorp’s commercial product integrations versus pure Terraform functionality? Do you want your core infrastructure automation layer governed by a publicly accountable foundation or guided by a corporate roadmap?
The fork has matured. It’s not a protest anymore. It’s a viable alternative with different trade-offs. Some organizations will land on Terraform for good reasons. Others will choose OpenTofu. The uncomfortable truth is that the fork succeeded not by being technically superior, but by offering a governance model that felt less risky to decision-makers who’ve watched enough enterprise vendor dynamics to develop healthy skepticism.
I’d genuinely like to know where you’ve landed on this. Are you actively using OpenTofu? Still on Terraform and planning to stay? Stuck in the middle of a migration? The infrastructure automation space is finally interesting again, and it’s because the boring question of licensing actually changed the technical landscape.
