Fivetran and dbt Labs: what the merger means for data teams

Fivetran, the company behind one of the most popular tools for loading data into warehouses, and dbt Labs, the company behind dbt, completed their merger in June 2026, eight months after first announcing the all-stock deal in October 2025. The combined company operates as Fivetran + dbt Labs, with George Fraser as CEO and dbt's creator Tristan Handy as President. Since then, the first big result has shipped: dbt v2. Here is what the merger means in practice.

Why it matters

In a typical modern data stack, Fivetran moves data from sources such as Salesforce, Stripe or a production database into the warehouse (the "E" and "L" of ELT), and dbt transforms it into clean, tested models (the "T"). Many teams already used both. Bringing them together means one company now covers most of the path from raw source data to the tables your dashboards use.

What was announced

• dbt Core v2.0 under Apache 2.0: the fast Rust engine previously called Fusion was opened up as the base of dbt Core 2.0. It is now available as dbt-oss, alongside the full dbt distribution.
• dbt State: a caching layer that decides, on every run, which models really need rebuilding and skips or reuses the rest. The company claims infrastructure savings of 30% or more; it is now generally available on Snowflake, BigQuery, Databricks and Redshift, with usage-based pricing.
• dbt Wizard: an AI assistant for writing, refactoring and debugging models, grounded in your project's lineage, tests and metrics.
• Agents Schema: an open standard that stores metric definitions, semantic models, lineage and documentation as plain SQL tables in one schema of your warehouse, as shared context for AI agents.

The combined company says its tools will keep working across any cloud, warehouse and BI tool, with business logic that stays portable.

What stays the same

• Your dbt project is still SQL and YAML in Git. Nothing about the merger forces you to use Fivetran.
• Other loaders still work. dbt runs on top of whatever is in your warehouse, whether it was loaded by Fivetran, Airbyte, a Python script or a native connector.
• Open source dbt continues, and is now more capable, since the new engine's core is under the Apache 2.0 licence.

What to watch

• Pricing and bundling: expect more combined offers. Compare carefully against using the tools separately, or against alternatives, especially for smaller teams.
• The line between open source and paid: dbt now ships as dbt-oss and the full dbt. Check which features your team relies on, such as state-aware runs or the AI assistant, and which distribution includes them.
• AI features: the focus on "data for AI agents" means more AI tooling in your workflow. As always, its quality depends on well-named, documented and tested models.

What it means for analytics engineers

The core skills do not change: SQL, data modelling, testing and documentation are still what make data trustworthy. If anything, they matter more, because both the new performance features and the AI assistants work best on clean, well-documented projects. A good next step is learning dbt v2's static analysis and column lineage, which will become everyday tools.

New to the role? Read What does an analytics engineer do? and build your first dbt project. Preparing for interviews? Try the interview prep.

Sources: dbt Labs' announcement "Fivetran + dbt Labs Complete Merger to Create the Data Infrastructure for Trusted AI Agents" (June 2026), dbt's v2 documentation and release coverage from September 2026.

Alessandro Ecclesie AgazziWritten by Alessandro Ecclesie Agazzi, analytics engineer in London. Updated 13 October 2026. About me