Serverless DuckDB
Run DuckDB SQL analytics in a serverless cloud data warehouse. Dual execution runs query stages on your laptop, in the cloud, or across both, and zero-copy shares let teammates query the same data.
A serverless DuckDB data warehouse for data teams
Run DuckDB SQL locally and in the cloud, with isolated compute for every user and built-in collaboration.
Why MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuckWhy MotherDuck
Isolated compute for every analyst
Each user gets a dedicated serverless runtime, so analysts can work without competing for shared warehouse resources. Develop locally and then ship to the cloud with the same database.
Each user gets their own compute instance “duckling”
MotherDuck’s per-user tenancy model enables teams to configure each org member individually. By default, each user in an Organization, or team, has their own dedicated, serverless isolated runtime. CPU visibility is provided per-user for simple, straightforward cost attribution and usage auditing.
Push work down to the client
MotherDuck turns users’ laptops into local, easy-to-use execution nodes. DuckDB’s unique portability enables teams to work locally and in the cloud with the same database. Since queries process closer to where data is stored, local compute is saved for analytical processing.
A faster SQL and Python workflow
Use your favorite agent & the DuckDB python library to rapidly build and iterate on python notebooks. Simply use SQL interchangably with data frames and offload larger queries to MotherDuck.
AI-enabled UI and workflows
As the primary work surface for many analysts, our mission is to make MotherDuck’s UI your preferred place to quickly hone in on the data that matters. Stay focused on getting answers quickly and editing complex SQL queries auto-magically with FixIt. Alternatively, use the Column Explorer’s automated sparklines and summary stats to hone your analysis. Once added to your data toolkit, we think you’ll take to it like a duck to water.
Versatile SQL and Python Support
MotherDuck builds on DuckDB’s portable nature and allows you to integrate it directly into your data analysis workflows. With comprehensive support for enhanced and traditional SQL, users do not have to compromise on query readability, even in a Python environment. You can even run Python scripts and Jupyter notebooks or transfer data between DuckDB and pandas dataframes. For additional flexibility, DuckDB also has a dataframe-style API, giving SQL and Python users something to quack about.
Backed by the modern duck stack
MotherDuck complements and integrates with your existing data stack. Thanks to DuckDB’s momentum, our flock of partners is ready to ensure your analytics workflows take flight with local and remote queries. Many of our partners already support DuckDB, and some even use it in their core product as a cache or batch processing engine. For DuckDB users, getting started is as easy as executing ‘.open md:’ in the CLI.
MotherDuck UI
See it in action!
Query, Upload, share
Join MotherDuck Co-founder, Ryan Boyd, for a tour of the MotherDuck UI and see how you can:
- Query data from the public_data share
- Get a bird’s eye view of your data with the Column Explorer, and see how result sets are stored in DuckDB running in the browser to enable faster filtering and pivoting
- Upload data from the local machine to MotherDuck
- Write basic analytic queries and use FixIt to catch SQL syntax errors
- Share uploaded data with colleagues
Analytics beyond a laptop
MotherDuck keeps DuckDB's efficient, embeddable engine while adding cloud compute and storage. Teams use it for datasets from a few CSVs to many terabytes.
DuckDB's columnar engine
DuckDB uses columnar, vectorized execution for analytical workloads. MotherDuck extends that engine to managed cloud compute and adds differential storage and zero-copy sharing to add even more scale.
Run queries across local and cloud compute
Dual execution routes work to your laptop, the cloud, or both, then returns one result. Use local compute when it is available and scale beyond local memory when you need to.
Features
MCP for AI agents
Connect Claude, ChatGPT, Cursor, or any MCP-compatible agent to MotherDuck. Natural-language questions come back as accurate, traceable SQL, running in sandboxed compute.
Share business context
Guides give AI agents shared business context on every query, so natural-language answers stay accurate for the people who know the business.
Live data visualizations
Dives are live, interactive visualizations built by an AI agent or with SQL — filter and drill down on live data, share them, or embed them in your app.
Scheduled data pipelines
Flights are scheduled Python data pipelines that run inside MotherDuck. Describe a source and your agent writes, deploys, and schedules the ingestion job.
Isolated compute
Hypertenancy gives each user, customer, or AI agent an isolated DuckDB instance — high concurrency with no shared clusters or noisy neighbors, plus up to 16 read-scaling replicas.
Architecture
Managed DuckDB-in-the-cloud
Ecosystem
Modern Duck Stack
