DuckDB is an in-process analytical database. You query large local or cloud data files without standing up a separate database server first. Five years without a venture round, and DuckLabs, the Amsterdam company that employs its core team, still built something Amazon wanted enough to buy. Amazon Web Services signed a definitive agreement to acquire DuckLabs on August 26. DuckLabs says the deal should close in early September; Amazon says only that it expects to close shortly, subject to customary closing conditions (AboutAmazon, 2026; DuckLabs, 2026). Financial terms were not disclosed.

Hannes Mühleisen and Mark Raasveldt, who created DuckDB and co-founded DuckLabs, will keep leading the team and its technical direction. They stay in Amsterdam. The project started at Centrum Wiskunde & Informatica, the Dutch research institute where Python was created, and DuckLabs grew to more than thirty people, funded entirely through support and development contracts (AWS Big Data Blog, 2026).

Amazon Buys the Company. A Foundation Keeps the Code.

Amazon says it is not acquiring DuckDB. The database, along with DuckLake and Quack, stays free and MIT-licensed under the nonprofit DuckDB Foundation. Peter Boncz, the foundation's board member from Centrum Wiskunde & Informatica, said the foundation holds the open-source IP and will keep doing so (DuckLabs, 2026).

DuckLabs says the Foundation plans to add a technical advisory board so outside contributors get input on direction, and the team wants to open the extension stack so software signed by other developers and organizations can run inside DuckDB (DuckLabs, 2026). Both are plans, not completed changes. They exist because the people who write DuckDB's next release will be AWS employees once the deal closes.

AWS Wants the Queries Redshift Was Too Slow For

Mai-Lan Tomsen Bukovec, the AWS vice president who leads its cloud data services, frames DuckDB as the engine for everyday queries. In AWS's telling, more than 90 percent of data queries in the world run against a terabyte or less, below the scale Redshift, Athena, and EMR were built to serve (AWS Big Data Blog, 2026). DuckDB runs inside the application itself instead of talking to a separate server, which is why those queries return fast. Bukovec says the same pattern fits AI agents, which tend to poke around in small chunks of data before they commit to a larger query.

AWS is already using DuckDB internally. Amazon Quick's dashboard engine has processed more than 2.5 billion queries with DuckDB since October 2025, cutting average latency by 30 percent, and David Feng of the Allen Institute told AWS that queries against terabytes of neurophysiology data stored in S3 now return in under a second instead of minutes (AWS Big Data Blog, 2026). About nine days before the acquisition, Mühleisen and Raasveldt published DuckDB 2.0 highlights. Quack, the protocol that lets one DuckDB instance serve data to others over a network, was already public. AWS is buying a team already building the networked mode its own analytics stack can use.

Jordan Tigani thanked Amazon in DuckLabs' press release. The same day, he told MotherDuck's own readers to expect Amazon's usual playbook.

MotherDuck Said Two Things the Same Day

MotherDuck sells a hosted DuckDB service. Its CEO, Jordan Tigani, says three of its engineers rank among DuckDB's top ten external contributors (MotherDuck, 2026). On DuckLabs' announcement, he said Amazon's backing would add momentum and strengthen the ecosystem (DuckLabs, 2026).

On his own blog that same day, Tigani described a different read: "That's Amazon's playbook, after all: wait until an open source project gets big enough, then launch it as a service" (MotherDuck, 2026). He also wrote that the deal is being done in a way that's good for DuckDB's long-term prospects, since Amazon profits when more compute runs against the database. MotherDuck is also changing its product line. The company is launching enterprise support it had avoided for years to stay out of DuckLabs' way, with what Tigani called the founders' explicit blessing now that they are joining Amazon (MotherDuck, 2026). Fivetran's George Fraser was bluntly positive, calling DuckDB the center of a vendor-neutral open data stack and Amazon the right home for it (DuckLabs, 2026).

What Happens Next

Watch whether the extension stack opens to outside signers within a defined timeline, or stays a stated intention past the close.

Watch whether AWS ships DuckDB integrations for Google Cloud Storage and Azure Blob Storage at anywhere near the pace it ships them for S3.

About Shashi Bellamkonda

Loading bio…

LinkedIn Profile

Get new posts by email:

Subscribe

Powered by