Exec Summary
Fivetran and dbt Labs have announced a landmark all-stock merger, creating a unified leader in open data management with nearly $600 million in annual recurring revenue. The combination integrates Fivetran’s automated data movement with dbt’s transformation and metadata capabilities to simplify enterprise data management and power AI-ready analytics at scale. Coinciding with dbt’s Coalesce 2025 conference, the company also introduced major innovations, including the dbt Fusion engine for cost optimization and the dbt Agents suite, embedding agentic AI into analytics workflows. Together, these moves signal a new phase in data platform convergence, metadata, movement, and model management unify to deliver governed, intelligent, and scalable data systems for the AI era.
Setting the Standard for Open Data Infrastructure
The merger combines Fivetran’s industry-leading automated data movement platform with dbt’s open data transformation capabilities, creating a unified foundation for open data management infrastructure. This combined approach unifies data movement, transformation, metadata, and activation, which are four critical pillars defining how data flows and operationalizing analytics and AI.
Both companies have championed openness and interoperability as core principles. The merged entity commits to preserving dbt Core as an open-source project under its existing license, ensuring the vibrant community of over 100,000 practitioners remains central to the platform’s evolution. Built on open standards such as SQL and Apache Iceberg, the unified data management infrastructure aims to reduce engineering complexity and eliminate vendor lock-in, while remaining adaptable to future compute engines, catalogs, BI tools, and AI models.
Fivetran CEO George Fraser called the deal a “refounding moment” for the combined company and the broader data ecosystem, emphasizing that AI transformation requires a trusted, open foundation. Handy, soon to be president of the combined company, echoed this sentiment, noting that the merger accelerates dbt’s mission to create infrastructure that “acts as an abstraction layer across the entire ecosystem,” enabling practitioners to work seamlessly across environments.
dbt Labs Coalesce 2025: AI-Driven Innovation at the Core
The merger announcement coincides with dbt Labs’ Coalesce 2025 conference in Las Vegas, where the company unveiled major updates to the dbt platform that further its role at the intersection of metadata, AI, and governance. Central to these announcements is the dbt Fusion engine, which delivers cost optimization and performance improvements and a new class of AI-native capabilities designed to make analytics development faster, more intelligent, and more governed.
Fusion is now in Preview for key cloud data platforms, like BigQuery, Databricks, Snowflake, and Redshift. It introduces state-aware orchestration, allowing data teams to automatically reduce compute costs by up to 10%, with further optimizations driving 15%+ total data platform savings. We will dig in here more. By running only models that have changed and dynamically determining the most efficient execution paths, dbt is redefining how teams can align performance with cost efficiency.
Fusion also extends dbt’s role in enabling open table formats, allowing dbt-powered pipelines to create and manage Apache Iceberg tables natively across multiple clouds. This is going to be an interesting point, managed versus unmanaged iceberg. This supports cross-platform portability, an essential differentiator as data teams balance hybrid, multicloud, and AI workloads.
The most forward-looking innovation at Coalesce 2025 is the introduction of dbt Agents, a suite of intelligent assistants that embed agentic AI directly into the analytics development lifecycle (ADLC). Built into the dbt cloud platform and powered by the remote dbt MCP server, these agents aim to integrate AI into the core workflow of developers, analysts, and data engineers.
- Developer Agent: Explains logic, validates models, and refactors code from VS Code or dbt Studio prompts.
- Discovery Agent: Finds trusted datasets and definitions for faster, safer exploration.
- Observability Agent: Monitors jobs, identifies root causes, and proposes automated fixes.
- Analyst Agent: Embedded in dbt Insights, answers questions about models, metrics, and lineage.
These AI-driven assistants operate on structured, governed context, a key differentiator in an era where AI often struggles with hallucination and accuracy. Being at the metadata and model layer gives their agents a potential advantage. The MCP server, now generally available, exposes this context securely to AI model providers and IDEs such as OpenAI, Anthropic, and Cursor, ensuring data lineage, quality, and governance remain intact even as AI systems interact with enterprise data.
Our ANGLE
This week will be pivotal for both dbt Labs and Fivetran, as the dbt Coalesce 2025 conference brings together one of the most active communities in data engineering just as these two companies redefine their shared future. The merger represents more than a combination of two crucial data management companies; it aligns data movement and modeling under one open, AI-ready foundation.
From my perspective, this is a wise and necessary consolidation for the data ecosystem. Customers of dbt and Fivetran have long sought tighter integration between ingestion, transformation, and metadata management. As dbt expands its capabilities into governance, metadata, and model management, and Fivetran continues to lead in automated data movement, the merger closes critical gaps that have long existed between ingestion and transformation layers.
The move also underscores the growing strategic importance of metadata in the AI era. Metadata and data lineage are now the connective tissue between data quality, governance, and explainable AI. In this context, dbt’s push into agentic AI, via dbt Agents and the Fusion engine, makes sense: it could bring intelligence and automation to the analytics development lifecycle while maintaining context and trust. The dbt MCP server connecting to OpenAI, Anthropic, and Cursor also signals that dbt intends to be part of the broader agentic AI ecosystem, not just a standalone data modeling tool. This is important as they compete in a world where every data and storage vendor pulls together data platforms, in reality, and not just marketing.
It’s worth noting that this merger follows Salesforce’s intent to acquire Informatica, another signal that the data infrastructure layer is consolidating and its importance to AI-driven architectures. The combined Fivetran and dbt Labs entity has the potential to challenge traditional integration vendors while staying true to the open principles that fueled their communities. Many companies want to build a true, open data platform with the best-of-breed parts. The competition and coopetition with the full-stack providers, like Salesforce/Informatica, Databricks, Snowflake, AWS, Google, Oracle, Microsoft, and other data management startups, is about to heat up.
Ultimately, this merger represents a key next phase in the evolution of data platforms: one where data movement, transformation, and metadata converge to deliver governed, AI-ready data pipelines. This could mean less fragmentation, lower operational costs, and faster innovation for enterprises. It’s a pivotal moment for the ecosystem that could reshape how modern data stacks evolve in the age of agentic AI.
In this episode from the Future of Data Platforms, Frederic and I discuss dbt’s role a bit in the future. Ties in nicely to the news of today and this week.
