What AWS Buying DuckLabs, Creators of DuckDB, Really Says About Data Lakehouses

Amazon has acquired DuckLabs, the Amsterdam-based team behind the popular open source project DuckDB. DuckDB is an analytics engine that sits directly atop object storage and is able to process data without migration. DuckDB allows for straightforward human interaction with commonly used languages like SQL, Python and Go, but is also optimized for AI agents. AWS's own announcement notes: “because agents behave a lot like people when interacting with data. They poke. They experiment. They run exploratory analysis on small data sets before figuring out what they really want to do. DuckDB ends up being naturally optimized for AI agents to use.”

DuckDB has been able to carve out a place in the data lakehouse stack by being able to function at an enterprise level on both large and small scale. Deployment flexibility has added another feather to the project's cap, as it can be deployed on both private hardware and the public clouds making it uniquely able to deal with the ongoing question of data sovereignty. These characteristics are the heart of the software defined lakehouse, and DuckDB’s acquisition by AWS is only adding to the center of gravity defining these as critical elements of the modern stack. Data Lakehouses are this era’s winning ticket.

Why Did DuckLabs Sell 

To be clear AWS has not acquired the DuckDB open-source project itself. The project will continue to be overseen by the DuckDB foundation under their MIT license. But AWS has brought both the brain power and to some extent the inertia of the project into its fold.

Apart from the obvious, DuckLabs founders Hannes Mühleisen and Mark Raasveldt have said the reason they sold their company is that they feared "...that DuckDB’s growth would eventually outpace our ability to support it. That our small company could become a bottleneck for the project…" This fear is perhaps validated, as at the time of the acquisition, DuckDB had passed 40,000 GitHub stars and was seeing more than 50 million monthly PyPI downloads, more than double the year before.

Ryan Boyd, co-founder of MotherDuck (along with Jordan Tigani) the cloud data warehouse built on top of DuckDB, said the DuckLabs founders sold largely because they wanted to focus on engineering rather than build a sales organization. He said the founders "wanted to spend more time doing engineering," rather than staff up a consulting sales team.

All in all this paints a picture of a group of founders involved in a project whose growth has largely been lead by its popularity among developers looking forward to a future where they can still focus on the most important (and in my view interesting) aspects of the project while still growing. A win/win.

What is the Short Term Impact

MotherDuck’s Boyd and Tigani are doing more than commenting on the acquisition and have moved quickly to respond, by acquiring Tower.dev, a company supporting infrastructure behind its agent-oriented product line, in the same week as the AWS-DuckLabs announcement.

MotherDuck is now moving into enterprise support for DuckDB, a market it previously avoided out of respect for DuckLabs' business model. MotherDuck said in a  statement, “We welcome the competition” and have "the explicit blessing from Hannes and Mark that this won't be stepping on their toes.”

Boyd expects the acquisition to shift DuckDB's roadmap toward priorities that matter for server side and enterprise workloads, including reliability at scale, observability, memory management, and support for lakehouse formats like Iceberg and DuckLake.

Of course, acquisitions such as this one will always be subject to the underlying worry of open-source capture, where a company can leave a license untouched while quietly gaining outsized influence over what the project becomes next. We will have to see what the future holds for this project.

Long Term Impact of DuckLab Acquisition on Data Lakehouses

But, for those of us who don’t work at DuckLabs, AWS or related business, what is the impact of this acquisition on the lakehouse ecosystem as a whole? 

It’s no longer possible to deny the momentum of data lakehouses. With first in class innovators like AWS throwing their weight in the argument, object storage as primary storage has become a given principle. 

DuckDB, like some other analytics engines trying to slough off the stolid legacy of traditional databases, has often been accused of just being a desktop tool for development and testing and incapable of running at large scale in production environments. AWS’s acquisition proves how wrong those assumptions are.

This shift was already underway before the acquisition, when AWS agreed to sponsor DuckDB's Apache Iceberg extension specifically to make it simpler to build applications on top of Amazon S3 Tables tying the query engine's development directly to adoption of object storage.

AWS has shown by its repeated investment in people, projects and ideas that Iceberg, analytics engines and object storage are winning combinations. That is a fact that anybody building data infrastructure should recognize.

Try it Out Yourself

DuckDB and MinIO AIStor pair well together for exactly the reasons this piece has been making the case for. DuckDB lets analysts, engineers, and data scientists query data directly where it lives, skipping the extract and load steps that slow other workflows down. 

If you’re exploring using Apache Iceberg, AIStor Tables embedded catalog collapses your required stack into just two components: your quacking fast query engine and high-performant object storage.