The Multi-Tenant GPU Machine: A Reference Architecture
The multi-user server assumes its users are colleagues. Remove that assumption and almost everything changes — consumer cards stop qualifying, and prefix caching becomes a leak.
Homelab
Servers, racks, power and the supporting infrastructure that keeps a self-hosted lab running reliably.
The Colocated AI Machine: A Reference Architecture
A machine nobody can physically touch is specified around its recovery ladder. Out-of-band management, network-bound unlock, watchdogs and the runbook a stranger will read under pressure.
The Multi-User Inference Server: A Reference Architecture
Serving many people inverts the arithmetic: batching makes concurrency nearly free, and the KV cache — not the model — is what fills the memory. Specified against latency targets.
The Always-On AI Server: A Reference Architecture
A machine that is switched on while you sleep is bought on a different axis entirely. Idle draw, keep-alive, cold-start latency and the API you must not expose — specified.
Best NAS for AI Developers
A NAS for model weights is optimised differently from one for media — throughput, cache and memory matter, and the transcoding engine everyone advertises does not.
Best Mini PCs for Homelabs
Homelab nodes need I/O, cores, low idle power and working IOMMU — not memory bandwidth. Which mini PCs deliver that, what idle power really costs, and where people overspend.