⏱️ Lectura: 10 min
DuckLabs, the company behind the DuckDB database engine, announced on August 26, 2026 that it will join Amazon Web Services (AWS) in early September. The announcement is signed by the founders themselves, Mark Raasveldt and Hannes Mühleisen.
📑 En este artículo
What changes is the corporate structure behind the project. What doesn’t change, according to DuckLabs, is that DuckDB open source will remain free under the MIT license, along with DuckLake and Quack, with the DuckDB Foundation responsible for its governance.
TL;DR
- DuckLabs, the company behind DuckDB, will join AWS in early September 2026.
- DuckDB, DuckLake, and Quack will remain free and open source under the MIT license.
- The DuckDB Foundation, a nonprofit entity, continues overseeing the project’s governance.
- The team of more than 30 people will stay together in Amsterdam after joining AWS.
- DuckDB logs more than a million daily downloads, according to DuckLabs’ announcement.
- DuckLabs and AWS had already been collaborating for over a year before announcing the deal.
- Mark Raasveldt and Hannes Mühleisen founded DuckLabs in 2021, just over five years ago.
- Andy Warfield (VP at AWS) called the DuckLabs team one of the most technical he’s ever worked with.
Introduction
For anyone who doesn’t closely follow the analytical database ecosystem, DuckDB is today one of the most widely used open source projects for in-memory data analysis, comparable in popularity to what SQLite represents for embedded transactional databases. It runs inside a single process, with no separate server, and lets you run SQL queries over CSV, Parquet, or JSON files directly from a notebook, a Python script, or the terminal.
Many data teams in Latin America already use DuckDB to replace processes that used to require Spark or a full data warehouse, simply because it runs on a laptop and reads Parquet files with no prior configuration. That’s why DuckLabs’ announcement isn’t technical news in the sense of a new release: it’s business structure news with direct consequences for thousands of developers and companies that depend on DuckDB in production.
What happened
DuckLabs, the company founded by DuckDB’s creators, will be acquired by AWS. According to the official announcement, the change takes effect in early September 2026, and the team of more than 30 people will stay together in Amsterdam.
The announcement details that DuckLabs and AWS had already been collaborating for over a year before the announcement. That prior relationship, according to the founders, gave them confidence to take the step: they understood how each team worked and what could be built together by combining DuckLabs’ technical expertise with AWS’s infrastructure and customer reach.
Andy Warfield, Distinguished Engineer and Vice President at AWS, called the DuckLabs team “one of the most technically deep, humble, and high-velocity teams” he has worked with, according to the quote included in the announcement itself. For the data community in the region, the immediate question isn’t whether DuckDB will keep working (it will), but whether the pace of new features accelerates or slows down under AWS’s umbrella.
Context and history
DuckLabs was founded just over five years ago, in 2021, to give Raasveldt and Mühleisen a stable base from which to keep developing the project. At that point the engine was already gaining traction: the first commercial contracts were arriving, along with interest from venture capital funds.
DuckDB started as a research project at the Centrum Wiskunde & Informatica (CWI) in Amsterdam, the same institute where Python was also born. That academic origin partly explains why the project prioritized engine correctness and performance over enterprise sales-oriented features for years.
The founders chose a different path from other infrastructure startups: keeping DuckLabs bootstrapped, with no outside investment, owned by the founders and the development team. That decision, according to the announcement, gave them the freedom to build patiently and prioritize technology over rapid growth.
The result: a team that grew from a handful of people to more than 30 in Amsterdam, while DuckDB racked up more than a million daily downloads. But that same growth created a problem: the founders acknowledge they feared the company would become a bottleneck for the project itself, and that scaling DuckLabs into a much larger sales and support organization would have distracted the team from the technical work and the open source community that made DuckDB’s success possible.
Technical details: DuckDB open source and the rest of the Duck Stack
The announcement doesn’t include code changes or a new DuckDB release. What matters on the technical side is which components of what DuckLabs calls the “Duck Stack” fall under which legal regime, and who maintains them going forward.
| Component | What it is | License | Who maintains it |
|---|---|---|---|
| DuckDB | Embedded analytical database engine (OLAP) | MIT | DuckDB Foundation, with contributions from the DuckLabs team within AWS |
| DuckLake | Lakehouse format with a metadata catalog stored in a SQL database | MIT | DuckDB Foundation |
| Quack | Companion project within the Duck Stack mentioned in the announcement | MIT | DuckDB Foundation |
The central point, emphasized repeatedly in the announcement, is that no open source component changes license. DuckDB open source keeps being distributed under MIT, the permissive license that allows using, modifying, and redistributing the code with no commercial restrictions, which was already the case before the deal with AWS. The MIT license also lets any company, including AWS competitors like Google Cloud or Snowflake, keep bundling DuckDB inside their own services without paying royalties, as long as they keep the original copyright notice.
For a developer already using DuckDB in production, this means that the binary, the Python library, the Node.js driver, or the R extension installed yesterday still work exactly the same today. Nothing in the build pipeline or the release process changes simply because of the announcement.
How to start trying DuckDB
If you’ve never used DuckDB, installing it takes less than a minute on any operating system.
# macOS (Homebrew)
brew install duckdb
# Linux (official script)
curl https://install.duckdb.org | sh
# Windows (winget)
winget install -e --id DuckDB.cliOnce installed, open the CLI and run your first query:
duckdb
D SELECT 'Hello, DuckDB' AS message;
┌───────────────┐
│ message │
├───────────────┤
│ Hello, DuckDB │
└───────────────┘That first block confirms the binary runs. The second, more realistic example shows why the AWS deal matters in practice: DuckDB can read Parquet files directly from an S3 bucket using the httpfs extension, without moving the data to an intermediate server.
INSTALL httpfs;
LOAD httpfs;
SELECT product_category, COUNT(*) AS orders
FROM read_parquet('s3://latam-sales/orders/2026/*.parquet')
GROUP BY product_category
ORDER BY orders DESC
LIMIT 10;That query groups orders by category directly on files hosted in S3, with no need to spin up a cluster or load the data into a traditional database before analyzing it. To confirm the extension is active before running a query against remote data, you can check it like this:
SELECT extension_name, loaded
FROM duckdb_extensions()
WHERE extension_name = 'httpfs';If the loaded column returns true, the extension is ready and the query against S3 should run without connection errors.
💡 Tip: DuckDB doesn’t need a server running in the background: each query runs inside the process that invokes it, so you can try it on your laptop with no additional infrastructure.
Impact and analysis
For AWS, the interest is direct: DuckDB is already used inside products and workflows of S3 customers, as Warfield himself acknowledges in the announcement. Bringing in the team that builds it gives AWS control over the roadmap of a piece of infrastructure that already runs, directly or indirectly, inside its data ecosystem.
flowchart TD
A["DuckLabs (Amsterdam)"] --> B["Joins AWS"]
A --> C["DuckDB Foundation"]
C --> D["DuckDB (MIT)"]
C --> E["DuckLake (MIT)"]
C --> F["Quack (MIT)"]
B --> G["AWS infrastructure and customer reach"]The diagram sums up the separation DuckLabs is trying to communicate: the company, now part of AWS, keeps contributing code, but the legal governance of the open source project stays with the nonprofit foundation, not with AWS directly.
That separation isn’t new in the industry, and it doesn’t guarantee anything on its own. Elasticsearch changed its license after AWS launched a managed fork of the project, and Redis did something similar in March 2024, when it moved from the BSD license to the RSAL and SSPL licenses, which led part of the community to create the Valkey fork under the Linux Foundation. Neither of those two cases involved an independent foundation controlling the license: in both, the founding company itself unilaterally decided to restrict the terms.
⚠️ Heads up: the commitment to keep DuckDB open source under MIT being genuine today doesn’t remove the structural risk that, in the future, the incentives of whoever funds the team contributing the most code to the project could change.
That said, there’s a real difference from the previous cases: the existence of the DuckDB Foundation as a separate, nonprofit entity dedicated specifically to the project’s governance. As long as that structure holds and the project keeps receiving contributions from outside DuckLabs, a unilateral license change would be far harder to pull off than in a project controlled by a single company.
What’s next
The deal takes effect in early September 2026, according to the announcement. From there, what to watch is whether AWS announces concrete integrations between DuckDB, DuckLake, and its existing data services (S3, Athena, Glue), and whether the release cadence and the pace of external contributions to the GitHub repository stay steady.
For companies that currently rely on commercial support from DuckLabs, the point to watch closely is the terms of existing contracts and whether AWS absorbs them, renews them, or replaces them with its own enterprise support model. Companies in the region that use DuckDB embedded inside their own products don’t need to take any immediate action: the engine remains the same MIT binary they can audit, fork, or compile themselves if they see fit.
📖 Summary on Telegram: View summary
Try it yourself: run curl https://install.duckdb.org | sh today and test the query against your own Parquet file to see the engine in action before the transition to AWS wraps up.
Frequently Asked Questions
Does DuckDB stop being free after the AWS acquisition?
No. According to the official announcement, DuckDB remains free software under the MIT license, with no changes to its distribution model.
What happens to DuckLake and Quack?
Both remain part of the open source Duck Stack, under the same MIT license and the same DuckDB Foundation governance model.
When does the deal take effect?
DuckLabs states that the transition will be completed in early September 2026.
Is the DuckLabs team moving to the United States?
No. The announcement specifies that the team, more than 30 people, will stay together in Amsterdam.
Did DuckLabs already have a relationship with AWS before this announcement?
Yes. Both companies had been collaborating for over a year before the acquisition was made public.
Who controls DuckDB’s license going forward?
The DuckDB Foundation, a nonprofit entity independent of both AWS and DuckLabs, according to the announcement itself.
References
- DuckLabs: DuckLabs to Join AWS, Projects to Remain Open Source: official announcement with the deal details and the Andy Warfield quote.
- DuckDB repository on GitHub: the project’s source code and MIT license.
- Official DuckDB website: documentation, installation, and extensions like httpfs.
📱 Enjoying this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day. @programacion
Imagen destacada: Foto de Domaintechnik en Unsplash
0 Comments