given snowflake also uses s3. Could aws do something with redshift that moves computation closer to s3 and be even faster than snowflake. Basically take advantage of the fact that they own s3.
basically distributing compute down to the actual storage nodes
The idea that is handles some SQL functionality with additional parallelism seems fine, but then I read that if a query uses any AQUA functionality, it's something like the entire query runs in AQUA, and that seems to me to be potentially a problem.
I'm very curious to start investigating, but there's a list of higher priority matters to investigate.