dbt Becoming the Control Layer
Tristan Handy, CEO of dbt Labs, on dbt’s multi-cloud tailwinds
The bragging phase ended once companies had to run these stacks at scale. In 2020 and 2021, stitching together Fivetran, Snowflake, dbt, and BI tools looked like technical sophistication. A year or two later, it looked like too many contracts, too many brittle handoffs, and too much engineering time spent keeping pipelines green instead of shipping analysis. That is the opening for dbt to move from a transformation tool into the control layer that makes the stack feel like one product.
-
The warehouse stayed durable, but the surrounding tool sprawl became the problem. Snowflake plus dbt still made architectural sense, yet enterprises increasingly wanted fewer moving parts and tighter workflows, not more best of breed pieces.
-
Downmarket, the modern data stack was often too heavy from day one. Early stage teams faced a 4 tool setup, a long implementation, and dedicated analytics hiring just to answer routine business questions, which helps explain why the unbundled ideal faded fast outside enthusiast teams.
-
That shift also changed the competitive map. Snowflake and Databricks started bundling more native transformation and workflow features, while dbt expanded into orchestration, catalog, and observability so it could own the layer where teams define metrics, test data, and manage production changes.
The next phase is less about assembling the perfect stack and more about deciding who controls the workflow above the warehouse. If dbt keeps becoming the shared layer for logic, governance, and cross-cloud execution, it stays central even as infrastructure vendors bundle more of the stack underneath it.