Definitely not true for my very common use case of Postgres + Hasura + Node API. Especially the DB is way faster based on compute. You're talking ~300+ms gains in perf just in the DB layer, with the graph and node layers also getting significant bumps. I was seeing a single dedicated host in Virginia be way, way faster (like 80% faster reply times) than the top of the line cluster that was in Phoenix (at the time very close to me).
If you are serving a page that needs to do a few joins across a non-trivial dataset that's all it requires to immediately tilt the equation completely back to raw CPU being the dramatic deciding factor in speed.
If you are serving a page that needs to do a few joins across a non-trivial dataset that's all it requires to immediately tilt the equation completely back to raw CPU being the dramatic deciding factor in speed.