Starting today, you can run MotherDuck in two Microsoft Azure regions: East US (Virginia) and UK South (London).
For a while now, we've heard the same request from teams that run on Microsoft: MotherDuck, but in Azure. Today we're shipping it. Get started on Azure →
It’s the same MotherDuck you know: an AI data team built on DuckDB. Flights bring your data in, Dives are data apps built by agents, and Guides document what your team knows so it can be shared. And now, it runs alongside the data you already have in Microsoft Azure.
Why Azure
We kept hearing “if you’re an Azure shop, a vendor on another cloud is a hard sell”. Your enterprise agreements, your identity provider and your security reviews are all set up around Microsoft. Bringing in an AWS-only service means a new vendor review and an exception to policy, and that's often where the conversation ends.
We built MotherDuck on Microsoft Azure to take that blocker away. We have deployed into two regions to start, and if your data is there too, your queries stay within Azure, with no round trip out and back.
MotherDuck is your AI data team
A data team has one job: enable the business to make good decisions. The faster they can see what's happening, the better. But this only works when questions and answers iterate quickly back and forth, like changing filters on a spreadsheet. Too often, this process involves hand-offs and questions get lost in an ever-growing backlog. Each new business question adds pressure to the data team, which is already overloaded.
To solve this problem, MotherDuck serves as your AI data team. Simply pose a question, and receive immediate insight, over and over again. Anyone in your organization can ask questions, in the MotherDuck web app or through their own agent subscription. It's like asking a question to the most senior analyst in your company, every time.
Every person, agent, and customer gets an isolated Duckling: its own DuckDB instance for running queries. Ducklings scale to zero while idle, so you pay only for what you use. Storage sits apart from compute, so adding data simply scales without worry of migration or acquiring more hardware.
The rest of the work happens inside MotherDuck, too:
- Flights bring data in from its source. They're Python jobs written by your team (or its agents), and MotherDuck schedules and runs them.
- Guides hold context, information that the model couldn’t possibly know in advance. It’s shared with people and agents so everyone works from the same context.
- Dives are interactive data apps built by agents, shared with your team or embedded in your own product.
"On top of compute-heavy warehouses, agent-driven queries would create a serious cost problem. MotherDuck's architecture is an excellent fit for the kind of exploration and unpredictable queries that agents need to be useful." -- Pablo Ferrari, Director of Data Engineering.
Read the Together AI case study.
What's available on Azure
Pulse, Standard, Jumbo and Mega Ducklings are available on Azure today. Find the details are on our pricing page.
The availability target and support response targets are the same as on AWS. Our support policy remains unchanged as we onboard new regions.
MotherDuck on Azure also works with many common Microsoft tools:
Ask from Copilot
Connect the MotherDuck MCP server to GitHub Copilot in VS Code or to an agent in Copilot Studio. Ask a question in plain English, and MotherDuck writes the SQL, runs it on your data and sends back the answer. Setup takes a few clicks: connect Copilot to MotherDuck.
Query OneLake without copying it (preview)
Attach a Microsoft Fabric Lakehouse as a MotherDuck database through OneLake's Iceberg REST catalog. Query its tables with DuckDB SQL, join them with your MotherDuck data, and write new tables back that show up in Fabric. Nothing gets exported, so nothing goes stale: OneLake integration docs.
Bring it into Excel (experimental)
MotherDuck for Excel lets you import a table, refresh it with one click and write changes back, all from a pane inside your workbook. In our demo, 50,000 rows land in a sheet in under 2 seconds. It is an early preview, so try it on data you can re-create: install MotherDuck for Excel.
What's next
We open Azure regions where customers ask for them. More regions will follow.
If you need MotherDuck in an Azure region we don't run in yet, join the waitlist on the Azure page. That's how we decide where to go next.
Want to try it today?
Sign in, pick an Azure region, and point a query at a Parquet file in your own Blob container:
Copy code
CREATE SECRET azure_blob IN MOTHERDUCK (
TYPE azure,
CONNECTION_STRING '<your-storage-connection-string>'
);
SELECT *
FROM 'az://<container>/<path>/*.parquet'
LIMIT 10;
Table of contents
FAQS
East US (Virginia) and UK South (London). Need another region? Tell us on the Azure page.
Sign up for a new account and pick East US or UK South when you create your organization. You’ll need a new account even if you already use MotherDuck, because an existing organization can’t move regions.
Compute costs 10% more than the equivalent AWS region. Storage costs the same. The pricing page lists every rate.
Yes: the same DuckDB SQL, Ducklings, Dives, Flights, Guides and MCP server. Pulse, Standard, Jumbo and Mega Ducklings are available today.
In Azure Blob Storage, separate from compute.
All except for prompt_jev(): requests from every region are processed by TypeSafe in the United States. It’s on by default, and admins can turn it off on the AI Features settings page. Details are on the cloud regions page.
Yes. From an AWS region you can query Parquet and CSV files in Azure Blob Storage today. Azure bills you for the egress.
