First Party Exports Threaten Fivetran
Fivetran: the $200M/yr Zapier of ETL
The real risk is not that Fivetran loses access overnight, but that the best connectors slowly get pulled upstream by the SaaS vendors that own the underlying data and can package warehouse sync as a native product. Once a source app like Stripe or Salesforce can send its own data directly into Snowflake or BigQuery, it can offer fresher exports, fewer breakages, and keep the customer relationship and monetization for itself.
-
Fivetran built its core value by writing and maintaining source specific connectors on top of third party APIs. That creates constant maintenance work because those APIs were usually designed for app workflows, not full table replication at warehouse scale, so schema changes and edge cases turn into ongoing engineering labor.
-
The Plaid analogy is about control of the pipe. Plaid proved a middleware layer can become valuable by normalizing fragmented source systems, but it also showed why source owners push back once they see a third party owning access, distribution, and economics around their data. The same incentive exists for large SaaS vendors with valuable analytical data.
-
Native warehouse exports are attractive because the vendor already has the cleanest access path to its own data and can sell the feature at high margin. Stripe offers Data Pipeline as a first party sync product to warehouses and cloud storage, while Salesforce has built zero copy and connector based sharing into Data Cloud, showing how integration can move from partner feature to core platform surface.
This pushes Fivetran toward a future where the head of the connector market gets claimed by first party exports and the remaining opportunity shifts to cross source standardization, monitoring, and the messy long tail. The winners will be the companies that own not just one connector, but the control plane that helps data teams manage dozens of pipelines across native and third party sources.