MotherDuck acquires Tower to own the infrastructure behind its AI data pipelines

MotherDuck, a serverless data warehouse company, has announced its first acquisition: Tower, a data infrastructure startup founded by ex-Snowflake engineers whose technology was already powering MotherDuck's AI-powered data pipeline features.

The move reflects a broader strategic shift in how data platforms are approaching artificial intelligence. Rather than relying on third-party tools for critical functions, MotherDuck is bringing Tower in-house to deepen its control over the execution layer that allows AI agents to not just query data but autonomously create, deploy, and schedule data pipelines.

The problem Tower solves#

Tower, founded in late 2024, tackles what its creators call data engineering's "last mile." Once an AI model or developer writes pipeline code, that code still needs to be packaged, deployed to the right infrastructure, have credentials wired up, and be kept running reliably in production—tedious work that Tower's managed Python runtime handles automatically.

The startup's Tower Control tool takes this further, letting users describe the data pipeline they want in plain language. An AI agent can then generate the code, deploy it, and run it, moving from natural language prompt straight to production without manual setup.

Why MotherDuck needed ownership#

MotherDuck, built on the open-source database DuckDB, launched Flights in June—a feature that gives AI agents a general-purpose Python runtime for creating and scheduling data pipelines. Tower's technology was the infrastructure underpinning Flights from the start.

MotherDuck founder Jordan Tigani explains the strategic logic: "We became their largest customer almost overnight, and our teams have been shipping together ever since." The company initially sought Tower as a third-party recommendation for customers, but when AI began solving data problems in new ways, the need shifted. "Claude can solve connectors for data movement, but what it can't do is the sandboxing and scheduling," Tigani notes.

Tower provided exactly that—a safe execution environment for AI-generated code. But as Tower became central to MotherDuck's product, keeping it at arm's length created a problem. Customers would inevitably hold MotherDuck responsible for any failures, security issues, or performance problems in the runtime, even though MotherDuck didn't own it.

Tigani articulates the underlying principle: "You can rent a feature, but you can't rent a foundation." Once a product has dependencies that define its reliability and behavior, those dependencies must be owned.