Good read, well structured, valid points, touches the major downsides\caveats of terraform at scale.
👏
Nitpick - Missing info on strategies to reduce state size using outputs from other workspaces: from my experience splitting terraform monolithic workspaces by using
workspace outputs is extremely fragile and awkward as opposed to the easy simple and convenient interface of Pulumi
stack references.
not a fair comparison: but in a 2800 resources TF monolithic, I only managed to chain 2 workspaces together before stuff was too hard to debug or broke
with Pulumi I got multiple parallel chains of 1-2 stacks deep (app->middleware->core) references