How to Run OpenClaw on a Raspberry Pi or Mac Mini — The Always-On Home Server
Silent, low-power and always on — a dedicated home server keeps your AI agent available 24/7 without the cloud cost or the privacy trade-off.
Running OpenClaw on your main laptop works for getting started. But closing your laptop stops the agent. Scheduled tasks miss. Customer messages go unanswered. The morning briefing does not arrive. If you want genuine always-on operation, you need a machine that never sleeps.
A dedicated home server is the solution many privacy-conscious OpenClaw users choose — particularly those running Ollama for local models. You own the hardware, you control what runs on it and what data touches it, and the ongoing cost after purchase is minimal electricity.
Raspberry Pi 5 — The Budget Option
The Raspberry Pi 5 with 8GB RAM ($80-100) is sufficient for OpenClaw deployments using cloud models (GPT-4o, Claude). It runs silently, consumes roughly 5 watts of power and fits anywhere. The limitation: not powerful enough to run Llama 3.1 70B or other large local models at useful speeds. For smaller Ollama models (Llama 3.1 8B, Mistral 7B, Phi-3 Mini), performance is acceptable for non-time-sensitive tasks.
|
OpenClaw: The Complete Guide Get the complete home server setup guide for OpenClaw 47 pages covering everything from installation to advanced automations — with step-by-step instructions for every platform, every model and every messaging channel. Get the Complete Guide → |
Mac Mini M2 — The Premium Option
A Mac Mini with Apple Silicon (M1 or M2, 16GB unified memory) is the ideal home server for OpenClaw. The unified memory architecture means Llama 3.1 70B runs comfortably and at reasonable speeds. It is silent, energy efficient (around 10-15 watts under load) and runs macOS which makes setup straightforward. Used M1 Mac Minis are available for $400-600.
Installing OpenClaw on Your Home Server
The installation is identical to a regular machine. SSH into your server (or connect a keyboard and screen for initial setup), follow the standard installation for your OS, configure your channels and skills, then register the daemon:
openclaw gateway --install-daemon
On macOS, this creates a launchd plist that starts OpenClaw automatically at login. On Linux, it creates a systemd service that starts on boot. Either way, OpenClaw starts automatically whenever the machine starts — and since your home server never turns off, it runs continuously.
Maintenance
A home server running OpenClaw requires very little maintenance. Check for updates monthly: npm update -g openclaw && openclaw skills update --all. Monitor disk space if you are storing conversation history or processed files. The openclaw logs command shows recent activity and any errors.
|
Ready to have an AI agent that actually works for you? OpenClaw: The Complete Guide covers every step: installation on Windows, macOS, Linux and Docker; connecting WhatsApp, Telegram, Discord and Slack; the 20 best skills to install; 10 real automations with exact setup instructions; building custom skills; running 100% privately with Ollama; and deploying 24/7 on a VPS or dedicated machine. Get the Complete Guide →Instant PDF download · 47 pages · Works on every platform |