The AWS–DuckLabs Deal: Everything You Need to Know

Sep 2

On August 26, 2026, AWS announced it had signed a definitive agreement to acquire DuckLabs, the Amsterdam-based company behind DuckDB, one of the fastest-growing pieces of open-source data infrastructure of the last few years. The deal set off immediate debate across the data community. Here's a clear-eyed rundown of what's actually happening.

What is DuckDB, and why does this matter?

DuckDB is an "in-process" analytical database: it runs directly inside another application (a Python script, a notebook, a command-line tool) rather than requiring a separate database server to install and manage. That design makes it fast to start, simple to operate, and easy to embed into other software. It has become a go-to engine for data engineers building lightweight, SQL-based pipelines, and adoption has exploded, with DuckLabs citing more than three million downloads a day at its peak.

What did AWS actually buy?

This is the part that trips a lot of people up, so it's worth being precise.

AWS acquired DuckLabs, the roughly 30-person commercial company co-founded by DuckDB's creators, Hannes Mühleisen and Mark Raasveldt, back in 2020. DuckLabs had operated for over five years without venture funding, sustaining itself through support and development contracts, and it employs much of the core engineering team that builds DuckDB.

AWS did not acquire:

The DuckDB open-source project itself. The code remains MIT-licensed and is governed by the independent, nonprofit DuckDB Foundation, which holds the project's intellectual property and trademarks. Peter Boncz, a DuckDB Foundation board member, said publicly that the Foundation will continue to hold all of DuckDB's open-source IP regardless of the acquisition.
MotherDuck, the venture-backed startup (roughly $100 million raised) that sells a hosted, cloud version of DuckDB. MotherDuck was co-founded in partnership with DuckLabs and has a development-services agreement with it, but it's a legally separate company with its own investors.

As part of the arrangement, the DuckDB Foundation plans to introduce a technical advisory board so that companies and contributors building on DuckDB have a formal channel into the project's roadmap, independent of any single company, including AWS.

Why would AWS want this?

A few things line up:

AWS and DuckLabs were already working closely together. The two organizations had collaborated since roughly 2024, including on DuckDB support for Amazon S3 Tables and SageMaker Lakehouse. This deal formalizes and deepens a relationship that already existed, rather than coming out of nowhere.

Timing lines up with DuckDB's technical roadmap. Just nine days before the acquisition was announced, the DuckDB team published a preview of DuckDB 2.0, centered on a new client-server protocol called Quack that lets one DuckDB instance serve data to multiple other processes over a network, something DuckDB has never natively supported before. That capability is a natural fit for AWS's cloud services, and the short gap between the 2.0 preview and the acquisition announcement has led several observers to see the deal as tied directly to that roadmap.

It strengthens AWS's push to make S3 a place customers analyze data, not just store it. AWS has been building toward a world where data sits in open formats in object storage (like S3) and gets queried directly, rather than being loaded into a separate proprietary warehouse first. That shift puts competitive pressure on companies like Snowflake and Databricks, which sell the compute and governance layer that sits on top of stored data, even though both remain major AWS customers too.

AI agents may be changing what kind of database is needed. AWS has described how AI agents explore data differently than humans do: they run many small, exploratory queries in quick succession while figuring out what they're actually looking for, without getting tired or slowing down. DuckDB's ability to spin up instantly inside another process, with no server to manage, fits that bursty, disposable usage pattern well. This has been cited by AWS and multiple industry analysts as part of the strategic logic behind the deal.

There's also a straightforward talent and reputation angle. DuckLabs' founders have built a technically respected project with a deeply devoted developer community. Acquiring the team gives AWS both deep expertise and a genuine reservoir of goodwill in a developer population that hasn't always trusted large cloud vendors with open-source projects.

Will DuckDB stay open source?

AWS and DuckLabs have both been explicit that DuckDB, along with related open-source projects like DuckLake and Quack, will remain free and MIT-licensed, governed by the independent DuckDB Foundation rather than by AWS directly.

That said, not everyone is fully reassured. Some commentators have pointed out that "open governance" doesn't erase the reality that most of DuckDB's core committers now draw a paycheck from AWS, which creates real influence over the roadmap even if the license itself never changes, a dynamic some have called "open-source capture." DuckLabs' founders have said one motivation for the deal was that scaling their own sales and support organization was starting to pull engineering attention away from the technical work and open-source community that made DuckDB successful in the first place; joining AWS was framed as a way to keep that from happening.

There's a useful historical comparison here: AWS built Redshift using database technology it licensed from a company called ParAccel years ago, and that deal didn't end up undermining PostgreSQL or the broader open-source database ecosystem. It's not a perfect parallel, since this time AWS is hiring the maintainers directly rather than just licensing their technology, but it's part of why some in the community are cautiously optimistic rather than alarmed.

What’s Next For MotherDuck?

MotherDuck's CEO, Jordan Tigani, has addressed this announcement candidly and from both sides. In DuckLabs' own announcement, he said Amazon's backing would add momentum and strengthen the ecosystem. On MotherDuck's blog the same day, he described the move as classic Amazon playbook: waiting for an open-source project to reach sufficient scale and then turning it into a first-party service. At the same time, Tigani has argued the deal is probably good for DuckDB overall, since AWS now has a direct financial incentive to keep the project healthy: every workload that runs against data sitting in S3 is compute Amazon gets paid for, whether or not AWS ever sells a "DuckDB service" of its own.

According to GeekWire, MotherDuck is now moving into a line of business it had deliberately stayed out of before: selling enterprise support contracts for DuckDB itself, a market it had left to DuckLabs up to this point. Tigani told the outlet he has the founders' "explicit blessing" to do so, since Mühleisen and Raasveldt are no longer running an independent support business themselves. Separately, MotherDuck also picked up a small infrastructure company, Tower, just days before the acquisition was announced, strengthening its own engineering bench.

The bottom line

AWS bought DuckLabs, the commercial team and company behind DuckDB, not the open-source project, its trademark, or MotherDuck.
DuckDB, DuckLake, and Quack all stay MIT-licensed, governed by the independent, nonprofit DuckDB Foundation, which is adding a technical advisory board as part of this transition.
The deal formalizes a partnership AWS and DuckLabs already had, and lines up closely with the timing of DuckDB 2.0's new client-server capabilities.
It fits AWS's broader strategy of making S3 a place where data gets analyzed directly, and AWS has explicitly pointed to AI agents' bursty, exploratory query patterns as part of the rationale.
Open governance will be put to the test once most of a project's core maintainers work for one company; this is the main live tension to watch.
MotherDuck will expand into enterprise services and continue to develop its own independent cloud platform for managed DuckDB.

Deals like this are a good reminder that choosing and implementing the right data warehouse is never a one-and-done decision. If you're trying to figure out which warehouse can effectively answer the data questions that keep coming up in your team, need help standing it up, migrating onto it, or wiring it into an AI-driven workflow, we're a NYC-based data consultancy that can help, end to end. Get in touch to talk through your architecture.

Contact Database Tycoon