Because we use regular CH in production. I'm pretty familiar with it and I can always run large queries on a production server with 1TB of RAM.

Regular CH also support external data sources, so I can read 500GB of JSON from S3 and group by it on production server very fast and in memory.