Neat DuckDB is already fast enough for most single-node workloads, so distributing it opens up some interesting use cases for larger datasets.