In October 2025, the "Modern Data Stack" finally stopped being a collection of a dozen different tabs and officially became a platform. The announcement that Fivetran and dbt Labs signed a definitive agreement to merge wasn't exactly a shock to anyone who’s spent time in a Slack channel for data engineers. Honestly, we’ve been joking about "Fiv-dbt" for years. But seeing it actually happen in an all-stock deal that creates a combined company approaching $600 million in annual recurring revenue? That hits different.
It signals the end of an era. The days of choosing "best-of-breed" tools for every tiny slice of your pipeline are fading. Now, it’s about who can give you the most reliable, least annoying way to get data from a messy API into a clean, governed table without needing a team of ten people to babysit the "glue code."
Why the Fivetran dbt labs merge is a massive pivot
For the longest time, Fivetran was the "E" and "L" (extract and load), and dbt was the "T" (transform). They were the peanut butter and jelly of data engineering. You’d use Fivetran to dump raw data into Snowflake or BigQuery, then immediately trigger a dbt run to make sense of it.
But things got complicated lately. Fivetran started building its own "Quickstart" transformations, and dbt Labs was pushing harder into dbt Cloud, trying to find ways to monetize a product that everyone was already using for free via dbt Core. By joining forces, they’ve basically decided to own the entire "middle" of the data stack.
George Fraser, Fivetran’s CEO, is taking the wheel as CEO of the combined entity. Tristan Handy, the founder of dbt Labs, is moving into a role as President. It’s a "refounding" moment, as Fraser put it. Basically, they realized that if they didn't merge, they'd spend the next five years stepping on each other's toes while Snowflake and Databricks tried to eat their lunch by building native ingestion and transformation tools.
The Elephant in the Room: Open Source
You've probably seen the hand-wringing on Hacker News about this. People are worried. dbt Core is the backbone of thousands of data teams, and Fivetran is a proprietary, usage-based SaaS company. The cultures are... well, they're different.
🔗 Read more: Edward Snowden Joe Rogan: What Really Happened During That Three-Hour Marathon
Tristan Handy has been very public about the fact that dbt Core will remain open source. They’ve reaffirmed the commitment to the current license (Apache 2.0 for Core, though the newer "dbt Fusion" engine is under the more restrictive ELv2).
Here is the reality:
- Fivetran needs dbt’s community to stay relevant.
- The community needs Fivetran’s engineering muscle to keep the tool from stagnating.
- If they kill the open-source spirit, the community will fork the project faster than you can say "SQLMesh."
What this means for your daily workflow
If you're a data engineer or an analyst, things aren't going to break tomorrow. But the roadmap for 2026 looks a lot more unified. We’re moving toward something the industry is calling "Open Data Infrastructure." Think about the way schema changes currently break your pipelines. Fivetran detects a new column in Salesforce, it adds it to your warehouse, but then your dbt model fails because it wasn't expecting it. In a post-merge world, that feedback loop becomes a lot shorter. The ingestion layer can "talk" to the transformation layer in real-time.
Consolidation is the new black
Fivetran didn't just stop at dbt. They've been on a bit of a shopping spree.
- Census: They bought the leader in Reverse ETL (activation).
- Tobiko (SQLMesh): They snapped up a major emerging competitor to dbt.
- dbt Labs: The crown jewel of the transformation layer.
Basically, they’ve assembled an "Infinity Gauntlet" of data tools. You move it with Fivetran, you model it with dbt, and you push it back into your CRM with Census. One bill, one support team, one interface. For a CFO, this is a dream. For a developer who likes tinkering with a dozen different open-source tools, it’s a bit of a bittersweet moment.
Is "Vendor Lock-in" actually a threat?
It’s the question everyone asks. "If I put all my eggs in the Fivetran/dbt basket, am I stuck?"
Sorta. But not in the way people think. The "lock-in" here is more about convenience than technical impossibility. Because the whole stack is built on SQL and open formats like Apache Iceberg, you can still take your code and go elsewhere if you really want to. The "lock-in" is that the integrated experience will probably be so much smoother than the manual alternative that you won't want to leave.
The real risk is pricing. Fivetran is known for its "Managed Information Schema" and MAR-based pricing, which can get expensive if you aren't careful. dbt Cloud has also been ratcheting up its per-seat and per-run costs. When they're one company, expect the "bundle" to be the only way to get a decent deal.
Practical next steps for data teams
Don't panic and start migrating to a different stack. That would be a waste of energy. Instead, start thinking about your architecture in a more "platform-agnostic" way.
First, audit your dependencies. Look at how tightly your dbt models are coupled to specific Fivetran naming conventions. If you’re using Fivetran’s pre-built dbt packages, you’re already in the ecosystem—and that’s fine, just be aware of it.
Second, keep an eye on dbt Fusion. This is the new engine that was announced around the time of the merge. It’s designed for high-performance, state-aware orchestration. If you're a heavy dbt Cloud user, this is where you'll see the most "synergy" from the Fivetran merger, especially regarding how it handles data freshness.
Third, don't ignore the competitors. Tools like Paradime or Coalesce are leaning into being the "neutral" alternative. If you feel like the Fivetran-dbt ecosystem is getting too restrictive or too pricey, these are the exits.
💡 You might also like: Show of Hands Strands: Why This Polling Method Changes How We See Data
The Fivetran dbt labs merge is basically an admission that "modern" data engineering has matured. We’re moving past the wild west of 2021. It’s less about the novelty of the tools and more about the reliability of the output. If this merger means fewer 3:00 AM PagerDuty alerts because an ingestion-to-transformation link snapped, most of us will probably take that deal—even if it means our "best-of-breed" dreams have to change a little.
Actionable Roadmap for 2026:
- Review your contracts: If you have separate Fivetran and dbt Cloud contracts, start talking to your reps about "unified platform" pricing as they come up for renewal. There is likely a volume discount hidden in there.
- Standardize on SQL: Avoid using proprietary features that only work in one specific vendor's cloud. Stick to standard SQL and dbt's core macros.
- Explore Iceberg: The combined company is betting big on Apache Iceberg as the storage layer. If you haven't looked into how dbt handles Iceberg tables yet, now is the time to start.