AWS Acquires DuckLabs, the Company Behind DuckDB — and DuckDB Stays Open Source

TL;DR: AWS signed a definitive agreement to acquire DuckLabs, the Amsterdam company behind DuckDB. Closing expected in early September. DuckDB itself stays MIT-licensed under the independent DuckDB Foundation; founders Hannes Mühleisen and Mark Raasveldt keep leading the project. For builders using DuckDB today, nothing changes — same binaries, same license, same governance. The pattern here is the version of an open-source acquisition that's rare in practice.

What changed, in two pieces

Amazon announced today that AWS has signed a definitive agreement to acquire DuckLabs. Two entities to keep straight, because the acquisition only touches one of them:

DuckLabs is a private company. Five years old, bootstrapped, fully owned by its founders and the development team. It employs the core engineers who maintain DuckDB, DuckLake, Quack, and adjacent projects, and runs a services business around those tools.

DuckDB is the open-source database. MIT-licensed. Governed by the nonprofit DuckDB Foundation, which is independent of both DuckLabs and Amazon.

AWS is acquiring DuckLabs — the people and the service business. AWS is not acquiring DuckDB. The foundation keeps stewardship. The license stays the same.

Why this is the acquisition pattern open source actually wanted

This is the version of an open-source acquisition that's rare in practice: the maintainers get long-term funding and reach, the project keeps its governance, and the license stays MIT. The DuckLabs founders will continue leading the team. Mühleisen and Raasveldt will keep driving the technical direction of DuckDB.

The DuckDB Foundation is the load-bearing piece. Open-source projects that get acquired without an independent foundation tend to drift — maintainers get pulled into the acquirer's roadmap, decisions get made on quarterly planning cycles, and the things that mattered to the original community quietly stop happening. The foundation structure here means DuckDB has a check on that drift. AWS gets a seat through the team it hired, not through a license change.

It's also the version AWS has been signaling it wanted. Andy Warfield, AWS VP and distinguished engineer, framed the deal in terms most hyperscaler PR would never use: "the idea of building a tool that lets people work with data, with confidence." That's an open-source-developer framing, not a cloud-platform framing.

What builders should actually take away

If you're running DuckDB in production today, nothing changes. The same binaries ship, under the same license, on the same release cadence. The risk profile for using DuckDB in 2027 is not higher than it was yesterday — if anything, it's lower, because the team behind the project now has a hyperscaler's engineering budget behind the work.

If you're choosing between managed analytics services and self-hosted DuckDB for a new project, the calculus is the same as last week: DuckDB for embedded analytical workloads and data pipelines, managed services for the cases where you need cross-region replication and a control plane. This announcement doesn't change that.

If you're building in the open-source data ecosystem, this is the deal to study. The DuckLabs founders bootstrapped for five years, kept the license MIT, and put the foundation structure in place before any acquisition conversation. That's the play. Bootstrapping + foundation + clean acquisition beats relicensing drama every time.

What to watch

Three signals over the next six months:

1. Release cadence. If DuckDB major releases keep shipping on the same quarterly-ish timeline the project has historically hit, the acquisition is working. If releases slow or the roadmap gets pulled toward AWS-specific features, the foundation needs to flex.

2. Foundation independence. Any structural change to the DuckDB Foundation — board composition, governance modifications, anything that reduces the foundation's independence from AWS — is the leading indicator of future drift.

3. AWS service integration. AWS will almost certainly build deeper integration between DuckDB and services like Athena, Redshift, and SageMaker. The question is whether those integrations stay as optional layers on top of upstream DuckDB, or whether AWS starts maintaining a private fork with AWS-only features. The former is healthy. The latter is a sign the acquirer is treating the project as IP rather than community infrastructure.

The foundation structure makes all three of those observable. That's why it matters.