Topic
Lakehouse vs warehouse ingestion
Lakehouse ingestion lands data in open table formats - Iceberg, Delta, Hudi - on object storage rather than inside a managed warehouse. Storage is cheaper, query compute is decoupled, and the data is portable across engines. Airbyte and Estuary lead on Iceberg writers in 2026; Fivetran support is partial.source - pulled 2026-06-21
Why lakehouse is winning the storage layer
Iceberg, Delta, and Hudi all give ACID writes on object storage. Compute is decoupled - Spark, Trino, Snowflake, Databricks, DuckDB, and Polars can all query the same tables. The 2026 buying question isn't "warehouse or lakehouse" - it's "does my pipeline tool write to the lakehouse format I've chosen?"
Pipeline tool support, mid-2026
- › Airbyte: Iceberg destination GA; Delta via Databricks connector.
- › Estuary: Iceberg + Delta first-class; native CDC into Iceberg.
- › Fivetran: Partial Iceberg; production-ready Delta via Databricks.
- › Stitch / Hevo: Limited; warehouse-first products.
Cost comparison at 50M MAR
A 50M-MAR workload landing in Iceberg on S3 + Glue + Athena typically costs $300–$700 / month in storage and query compute, compared to roughly $2,000 in Snowflake compute on the same workload. The trade-off is operational maturity - managed warehouses are easier; lakehouse needs more engineering discipline.
Related
Written by Oliver Wakefield-Smith, Founder of Digital Signet. Independent reference, no vendor sponsorship.
Sources logged at /sources. Pricing-change history at /changelog. Last reviewed 2026-06-21.