MotherDuck
Serverless cloud data warehouse built on DuckDB, with isolated compute per user and agent and an MCP server.
MotherDuck is a serverless cloud data warehouse that runs on DuckDB. MotherDuck gives every user, dashboard and AI agent session a separate compute instance, so heavy queries do not slow other work. MotherDuck is queried with SQL, Python, a command line tool or natural language through its MCP server, and it is billed by usage.
Best for Data teams and developers who want SQL analytics without managing infrastructure, especially for customer-facing analytics and workloads driven by AI agents.
- Platforms
- Web, API, MCP
- Pricing model
- Freemium
- Free plan
- Yes
- Free trial
- Yes
- API
- Yes
- Made by
- MotherDuck
- Added
Last verified: . Information taken from motherduck.com.
What is MotherDuck?
MotherDuck is a cloud data warehouse that packages the DuckDB engine as a managed, serverless service. Its architecture, which the company calls hypertenancy, starts an isolated DuckDB instance named a Duckling for each person, dashboard or agent session, and those instances come in five sizes from Pulse to Giga. Storage is shared and works with Parquet, Apache Iceberg, DuckLake and cloud object storage. Around the engine sit agent-oriented tools: an MCP server offering read and write access, Guides that supply schema context to agents, Dives for agent-built visualisations that can be embedded in an application, and Flights, a Python runtime for scheduled pipelines. Integrations include dbt, Fivetran, Airflow, Tableau, Looker and Hex. A free Lite plan exists, and the Business plan adds production features on top of metered compute and storage.
Key Features
Per-User Compute
Starts a dedicated DuckDB instance for each user, dashboard or agent session so one workload cannot starve another.
Instance Sizes
Lets you match compute to the job with five instance types, Pulse, Standard, Jumbo, Mega and Giga.
MCP Server
Gives MCP-capable assistants such as Claude Code, Cursor and ChatGPT scoped read and write access to query data in natural language.
Dives
Publishes interactive visualisations built by agents, for internal use or embedding in a product, without a separate BI tool.
Flights
Runs Python pipelines on MotherDuck compute for scheduled ELT, enrichment or AI classification jobs.
Open Storage Formats
Reads and writes Parquet and Apache Iceberg tables and offers a managed DuckLake.
Screenshots
Use Cases
- Customer-Facing Analytics
- Serve dashboards inside your application, with each end user querying on separate compute.
- Internal Data Warehouse and BI
- Centralise company data and connect BI tools such as Tableau, Looker or Hex.
- Agent-Driven Data Exploration
- Let a coding agent run many exploratory queries in parallel without affecting dashboards.
- Offloading Analytics from Postgres
- Move analytical queries that overload a transactional database to a columnar engine.
- Scheduled Data Pipelines
- Run dbt projects and ingestion jobs on a timetable with Flights.
Things to know
- Usage is billed on top of the plan
- Storage is charged per GB per month and compute per second or per query, depending on the instance type.
- Lite is limited to Pulse instances
- The free plan cannot use the larger Standard, Jumbo, Mega or Giga instances.
- AI functions have their own rate
- The SQL Assistant is billed per call and other AI functions by tokens, priced in AI Units.
- Compliance options sit in Enterprise
- A HIPAA BAA, PrivateLink connectivity and IP allowlisting are listed only for the Enterprise plan.
FAQs
What is MotherDuck?
MotherDuck is a serverless cloud data warehouse built on the DuckDB engine. It handles compute and storage so teams can run SQL analytics without operating infrastructure.
What is MotherDuck used for?
MotherDuck is used for internal data warehousing and BI, for analytics embedded in customer-facing applications, and for data work carried out by AI agents.
Is MotherDuck free?
MotherDuck has a Lite plan with no platform fee. It includes 10 GB of storage, 10 hours of Pulse compute per month and up to three internal active users.
How much does MotherDuck cost?
MotherDuck charges a monthly platform fee per organisation on its Business plan, plus metered storage and compute, and offers a 7-day free trial. Enterprise pricing is custom; see the official pricing page for current rates.
Does MotherDuck have an MCP server?
Yes. MotherDuck provides an MCP server so MCP-capable agents can read and write data and answer questions asked in natural language.
Which tools does MotherDuck integrate with?
MotherDuck lists integrations including dbt, Fivetran, dltHub, Apache Airflow, Tableau, Looker, Omni and Hex, and client support for Python, TypeScript, Java, Rust and other languages.
What is a Duckling in MotherDuck?
A Duckling is MotherDuck's name for the isolated DuckDB compute instance it creates for each user, dashboard or agent session.
Who it's for
- Data engineers
- Analytics engineers
- Application developers
- Data teams
- Startups
- DuckDB users
Comments
No comments yet. If you have used MotherDuck, tell other readers how it went.
Alternatives to MotherDuck
| Tool | Free plan | Starting price | API |
|---|---|---|---|
 DatabricksUnified data and AI platform for data engineering, warehousing, analytics, governance and building AI agents. |
Yes | — | — |
 EnterpretCustomer intelligence platform that structures feedback from support, sales and market channels for teams and AI tools. |
— | — | — |
 FernGenerates API documentation sites, SDKs in multiple languages and a CLI from one API specification. |
Yes | — | Yes |
 GongRevenue AI platform that records customer interactions and turns them into forecasting, coaching and sales automation. |
— | — | — |
 AI Graph MakerWeb tool that turns typed descriptions or uploaded data into charts, flowcharts, mind maps and diagrams. |
Yes | — | — |
Listed in the same categories as MotherDuck. A starting price appears only where we checked it on the maker’s pricing page; a dash means not checked or not published. All Data Analysis & Analytics tools