Skip to content
Spec sheet

Validated hardware families.

The rack runs on what you already trust. Intel vPro enterprise PCs with Intel I219-LM, I225, or I226 NICs across enterprise and prosumer lines, plus Apple Silicon Mac mini Pro or Mac Studio for the AI mesh. Apple Silicon hardware is supplied by the buyer unless quoted as a bundle.

Validated hardware families

BrandLineCPUNICvProNote
HPEliteDesk MiniIntel vPro i5 / i7 (recent generations)Intel I219-LMYesValidated for hypervisor host and cluster node roles
LenovoThinkCentre M Tiny / ThinkSystemIntel vPro i5 / i7Intel I219-LMYesValidated for cluster node role
DellOptiPlex MicroIntel vPro i5 / i7Intel I219-LMYesValidated for cluster expansion
AppleMac mini ProApple Silicon (M-series Pro)1GbE, 10GbE configure-to-orderN/AApple Silicon AI mesh node, Tailscale-bridged to the cluster
AppleMac StudioApple Silicon (M-series Max / Ultra)Built-in 10GbEN/AAI inference head on the Apple Silicon mesh
MinisforumEnterprise PC (validated configurations)Intel vPro enterprise PCIntel I225 / I226YesProsumer alternative when enterprise lines are out of stock

Install tier sheet

Scope
Single-node hypervisor (Proxmox by default, yours to swap) + base VM stack + observability
Cluster
1 to 2 nodes
RAM floor
32 GB
Storage floor
1 TB NVMe
Apps
Pick from 80+ you can run, deployed across 1 to N nodes via GitOps.

Beelab Studio

View on pricing →
Scope
Lite + clustering + monitoring + GitOps pipeline (ArgoCD + Forgejo by default, swap for Flux / Gitea)
Cluster
1 to 3 nodes
RAM floor
64 GB per node
Storage floor
2 TB NVMe per node
Apps
Pick from 80+ you can run, deployed across 1 to N nodes via GitOps.
Scope
Studio + production Kubernetes layer (Talos by default, swap on request) + AI layer (gateway, RAG, agent automation) + migration from one provider
Cluster
1 to N hypervisor nodes + 1 to N AI nodes (Proxmox + Apple Silicon mesh by default, yours to swap)
RAM floor
Memory sized to your tier
Storage floor
2 TB NVMe per node + offsite path
Apps
Pick from 80+ you can run, deployed across 1 to N nodes via GitOps.

Beelab Enterprise

View on pricing →
Scope
Pro + PARSO-SAMAD self-assessment scorecard (see /parso-samad) + on-prem inference layer + custom MCP server + multi-rack scaling
Cluster
Scales 1 to N, custom
RAM floor
Sized per workload
Storage floor
Sized per workload
Apps
Custom, anything containerizable

Got hardware in mind that is not listed? Drop a note and I will tell you if it meets the spec.