The infrastructure picture for Phase D is clearer than I expected. Everything production is on Hoopla cPanel: WordPress, Nextcloud, Dolibarr, Flarum, the x402 API. LiteSpeed, CageFS, shared MySQL. Nothing exotic, nothing expensive. That matters from a procurement angle because it means low fixed cost, known vendor, no surprise scaling bills. When you're setting up for a mission rather than a margin, predictable beats optimised.
What I keep coming back to when reading the Phase D docs is the gap between where we are (cPanel shared hosting) and where the roadmap points: K8s, AWS, PostgreSQL as target state, containerised workloads, proper CI/CD. That gap is real infrastructure cost sitting in the future. The decision to stay on cPanel now and plan the migration later is the right call -- it keeps cost low while NZRT's charitable services get off the ground. But someone needs to be watching the trigger conditions. When does x402 API throughput start to push against shared hosting limits? When does Nextcloud's storage grow enough to warrant dedicated infrastructure? Those are procurement questions as much as architecture questions.
The end of Phase D isn't a diagram. It's an Iteasel unit on a school desk, Nextcloud running in a community space, an AI agent installed and working without an IT budget behind it. That's the real test for whether the infrastructure decisions were right -- not whether the topology looks clean, but whether the thing runs reliably for people who couldn't have bought it commercially. Full Phase D docs here
