Databricks Cost Glossary
What is Photon in Databricks?
Photon is Databricks' vectorized, C++-based query engine, available on clusters and SQL warehouses. Enabling it consumes DBUs at a higher rate for the same instance — often around 2×, varying by SKU — in exchange for faster execution. It lowers cost only when wall-clock time drops by more than that premium.
The economics are a race between two numbers: the DBU premium (roughly 2×) and the speedup. A scan-heavy SQL workload that finishes in 40% of the time comes out ahead; a workload dominated by Python UDFs or driver-side logic may speed up barely at all and simply pay double.
That makes “Photon on or off” a per-workload question, not an account setting. The engine shines on SQL and DataFrame operations over columnar data (large scans, joins, aggregations) and adds little where execution time isn’t spent in the query engine.
Try this: run the workload both ways and compare total DBUs consumed, not runtime alone. A job that got faster and cheaper is a win; a job that got faster and more expensive is a choice you should be making deliberately.
Related terms
See these numbers in your own workspaces.
The free tier covers unlimited workspaces with three months of history. No card required.
Start FreePart of the Databricks cost glossary. Last updated July 15, 2026.