There were brief and easily overlooked comments elsewhere indicating that Supabase is non-trivial to port to Fly.io’s “always ≥2 machines” reliability model. Possibly this is an interim measure until the cycle (eventually) turns for PG startups…
(Also, Fly Kubernetes might have changed the in-house calculations, somewhat.)
The preceding paragraph, about the side effects of over-bundling PaaS, is the more insightful and telling of the two—in my opinion:
mrkurt:
I think the most interesting part of this is the PaaS disaggregation. Heroku built an exceptionally good Postgres service. They could not have done that with multiple DBs. Even their redis is pretty meh.
It looks like this overall reasoning lives on, albeit stated in a much more dilute way—and (like you pointed out) with tactical adjustments in one corner…
https://fly.io/developer-cloud
I hope the “distributed filesystems” part in particular pans out, 
…