Yesterday, MinIO’s GitHub repository was formally archived and marked read-only. It had been a long time coming. But it still stings a little.
We used MinIO at VM Farms for years. It was the default answer when a customer needed self-hosted object storage: S3-compatible, simple to deploy, fast, and genuinely well-built software. By the time of its 2022 funding round, MinIO had recorded 762 million Docker pulls and had become one of the most widely deployed pieces of cloud-native storage infrastructure in the world. It felt like infrastructure you could depend on.
We moved away from it a while back. But I feel something watching it end this way.
What MinIO Was
MinIO is an object storage system that speaks the S3 API. It meant you could build your application against Amazon S3 and swap in MinIO for development, testing, or production without changing your code. For a lot of teams, that was genuinely valuable. It became the default self-hosted S3 alternative.
Founded in 2014, licensed under Apache 2.0, built by people who clearly cared about the software. It had all the hallmarks of a project that was going to matter. And for a long time, it did.
How It Fell Apart
There is a thorough account of what happened in a piece published in February by Reading.sh: “How MinIO Went from Open Source Darling to Cautionary Tale.” I recommend reading it. Here is the condensed version.
In May 2021, MinIO silently changed its license in code, then published a formal explanation in October 2021. The post reframed the move as principled: “copyleft licenses protect the freedoms of all users by preventing proprietary derivatives and forks.” In practice, it created compliance headaches for commercial users who had built on the Apache 2.0 version.
Eight months after the license change, MinIO closed a $103 million Series B at a $1 billion valuation, led by Intel Capital with participation from SoftBank Vision Fund 2.
Then came the enforcement. In July 2022, MinIO publicly accused Nutanix of violating its license in the Nutanix Objects product. In March 2023, Weka received the same treatment. The AGPL was being used as a sales tool: pay up or face public accusations.
Then came the feature stripping. In June 2025, MinIO removed the full admin console from the community edition, leaving users with a bare-bones object browser and no web-based management unless they subscribed to the commercial AIStor product. In October 2025, MinIO stopped publishing Docker images and pre-built binaries for the community edition.
On February 12, 2026, the README was updated to declare: “THIS REPOSITORY IS NO LONGER MAINTAINED.” On April 25, 2026, the repository was formally archived and locked read-only.
What’s Different About This
What’s notable is how far it went compared to other open source license changes. Redis, Elasticsearch, HashiCorp, Terraform: all changed their licenses and drew criticism. But they maintained functional community editions. They gave their users something to work with.
MinIO eliminated the free alternative entirely, directing everyone toward AIStor at pricing that analysis estimates starts around $96,000 per year for enterprise deployments (the official pricing page now uses “Request Pricing” rather than listing figures). For the homelab users and small teams who adopted MinIO because it was free, there is no migration path. There is a cliff.
The VC Equation
Here is what I keep thinking about.
The roughly 60,000 GitHub stars, over a billion Docker pulls, and billion-dollar valuation that made MinIO attractive to SoftBank Vision Fund 2 are the same numbers that created pressure to monetize community goodwill. The scale of adoption that made the software valuable was also the thing that made it a target for extraction.
SoftBank Vision Fund 2 does not invest in projects. It invests in companies with paths to enormous revenue. A billion-dollar valuation carries expectations that community downloads cannot satisfy on their own. The license change, the enforcement actions, the feature stripping: each step makes more sense when you understand the cap table.
This is not unique to MinIO. The pattern is documented well enough at this point. The question is never whether a VC-backed open source company will try to monetize its community. The question is how far they are willing to go.
MinIO went further than anyone had before.
What We Did
We moved away from MinIO before things reached this point. When the signals appeared, the license change and the enforcement actions, we made the decision that building customer infrastructure on a project with that trajectory was not something we could recommend in good conscience.
Most of the tech world is built on open source. That is not a problem. It is genuinely remarkable, and most open source maintainers deserve far more credit than they get. But “open source” is not a single thing.
An Apache 2.0 project maintained by a distributed community of contributors is a fundamentally different dependency than one controlled by a single company with growth-stage investors. The license on the tin does not tell you who makes the decisions, or what pressures they are under.
Where People Are Going
The community has been rebuilding. SeaweedFS has been the most common destination for teams migrating away: production-ready, more complex to set up, but consistently solid in practice. Garage is a lighter Rust-based option designed for geo-distributed setups. RustFS is a ground-up Rust rewrite built to be what MinIO used to be.
MinIO’s AGPL licensing made a direct fork difficult, unlike Redis (which produced Valkey) or Terraform (which produced OpenTofu). The community is rebuilding rather than forking. That is its own kind of commentary on where things ended up.
The Lesson
What happened to MinIO is sad. Not because the company made inexplicable decisions: the decisions are completely explicable when you understand the incentives. But because the software was genuinely good and a lot of people built things on it in good faith.
The lesson is not that you should avoid VC-backed open source. The lesson is that dependency is a risk that deserves honest thought before you build. Who controls the project? What are their incentives? What happens to your infrastructure if they change direction?
Those are questions worth asking before you build, not after the repository is archived.