Interesting to see how Databricks reacts to this given that they have their own Project Lightspeed (replacing the Spark execution engine).