Every few weeks the same question surfaces in Slack communities, on Reddit, in evaluation docs: "Should I be worried about SQLMesh? Is it still maintained? What happens if it goes away?"
It's a fair question, and it has a timeline behind it. In September 2025, Fivetran acquired Tobiko Data, the company behind SQLMesh. A month later, Fivetran announced plans to merge with dbt Labs — the company behind SQLMesh's most direct competitor. The community math wrote itself: why would the combined company keep two transformation frameworks? Fivetran says the overwhelming majority of its customers already use dbt. Surely SQLMesh gets quietly starved, and everyone gets migrated to the framework with the bigger install base.
Then, in March 2026, Fivetran contributed SQLMesh to the Linux Foundation.
That should end the conversation. But it's worth spelling out why it ends the conversation, because the reasoning matters more than the headline.
What the Linux Foundation donation actually means
A Linux Foundation project is not a corporate side project with a nicer logo. The trademark, the repository, and the governance no longer belong to any single vendor. SQLMesh is now developed under vendor-neutral, community-driven governance, with an initial member roster — Benzinga, CloudKitchens, Harness, Infinite Lambda, Jump AI, and Minerva — whose production workloads depend on the project continuing to exist. No merger, acquisition, pivot, or product strategy meeting at any one company can unwind that. Fivetran couldn't kill SQLMesh now if it wanted to, and donating it to a neutral foundation is a fairly expensive way of signaling that it doesn't want to.
This is the same home that hosts Kubernetes, PyTorch, and Node.js. Projects that outgrew, outlived, or simply out-neutraled their corporate parents. The pattern is well established: putting a project under foundation governance is how you make it bigger than any one company's roadmap.
The part that was true even before the donation
Here's the thing, though: even before March 2026, the doomsday scenario never quite held up, because it misunderstands what open source is.
SQLMesh is Apache-2.0 licensed. That license is irrevocable for the code you already have. There is no mechanism — NONE — by which a vendor, an acquirer, or a merged mega-vendor can reach into your infrastructure and take the software back. The worst case for an open source project is not "it goes away." Going away is not a thing source code does. The worst case is "the original maintainers stop working on it," and the remedy for that is built into the license: the code forks, and whoever still needs it keeps it alive.
Compare that to the actual worst case for proprietary software, which every
data team has lived through: the vendor sunsets the product, doubles the
price at renewal, or gets acquired and "aligns the portfolio." Your
recourse is a migration project. With open source, your recourse is git clone.
Fork it, verify it, PR it back
When you hit something in SQLMesh that needs to change — a bug, a missing capability, behavior that doesn't fit your workloads — the workflow is boring and repeatable:
- Fork the repository. You now have the entire project, history and all. Nobody's permission required.
- Make the change and verify it works — against your real workloads, not a toy example. Run the test suite. Run your pipelines against your fork.
- Open a PR back to the main project. Now the fix belongs to everyone, the community reviews it, and you're not carrying a patch forever.
That loop is the entire insurance policy. If upstream is healthy, your PR lands and you delete your fork. If upstream ever went quiet — which, under Linux Foundation governance with member companies staking their infrastructure on it, is now about as likely as Kubernetes going quiet — the same loop keeps working. Your fork is the project until the community reconverges. That is not a disaster recovery plan bolted onto open source; it is how open source works on a normal Tuesday.
Judge projects by their exits, not their owners
The instinct to ask "but what if the company behind it disappears?" comes from a world of proprietary vendors, where the software and the company are the same thing and one dies with the other. Open source severs that link. The right question for any tool you build on is not "will this vendor exist in ten years?" — nobody knows that about any vendor — but "if the worst happened, what's my exit?"
For SQLMesh the answer is: an Apache-2.0 license, a public repository, a vendor-neutral foundation holding the keys, member companies invested in its future, and a fork button. That's a better continuity story than almost any product you're paying for.
SQLMesh is here. It's community-governed. It's forever available. Build on it.