Turn Your Mac Mini Into a 24/7 AI Server
Mac Mini M1 + OpenClaw = a silent, always-on personal AI that costs $15–55/month and runs circles around any SaaS subscription.
Why Mac Mini is Perfect for OpenClaw
Silent & Low Power
Apple Silicon uses ~10W idle. Runs 24/7 for less than $3/month in electricity.
Apple Silicon Performance
M1 chip handles AI inference, code compilation, and multitasking effortlessly.
Small Footprint
Fits in a drawer. Smaller than most books. Ethernet port for stable connectivity.
Your Hardware, Your Data
Nothing leaves your network. Full privacy. No datacenter data exposure.
What You'll Build
By the end of this guide, your Mac Mini will be running:
Hardware Requirements
Mac Mini M1 or newer
M1 is plenty. M2/M3 for heavy local model use.
8GB minimum, 16GB recommended
8GB handles cloud APIs fine. 16GB+ for local Ollama models.
256GB minimum
OpenClaw itself is small. Storage matters for local AI models.
Ethernet cable (preferred)
WiFi works but Ethernet is more stable for always-on use.
$299–599 refurbished
M1 Mac Mini starts at ~$299 on eBay/Facebook Marketplace. New M4 is $599.
Step-by-Step Setup
Follow these 10 steps to go from bare Mac Mini to fully operational AI server.
Initial macOS Setup
Prevent the Mac Mini from sleeping and enable remote access.
# Disable sleep and keep display off (not sleeping) sudo pmset -a sleep 0 sudo pmset -a disksleep 0 # Enable Remote Login (SSH) sudo systemsetup -setremotelogin on # Verify SSH is running sudo launchctl list | grep ssh
💡 Note: In System Settings → Energy, set "Prevent automatic sleeping" to ON. Also enable "Wake for network access" for reliability.
Install Homebrew + OpenClaw
Install the package manager and OpenClaw itself.
# Install Homebrew /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" # Add Homebrew to PATH (Apple Silicon) echo 'eval "$(/opt/homebrew/bin/brew shellenv)"' >> ~/.zprofile eval "$(/opt/homebrew/bin/brew shellenv)" # Install OpenClaw brew install openclaw # Verify installation openclaw --version
💡 Note: On Apple Silicon (M1/M2/M3), Homebrew installs to /opt/homebrew. Make sure to follow the post-install instructions to update your PATH.
Configure API Keys
Add your AI provider credentials to the environment.
# Create environment file cat >> ~/.zprofile << 'EOF' export OPENAI_API_KEY="sk-..." export ANTHROPIC_API_KEY="sk-ant-..." export OPENROUTER_API_KEY="sk-or-..." EOF # Reload profile source ~/.zprofile # Or use openclaw's built-in key manager openclaw config set OPENAI_API_KEY sk-...
💡 Note: Start with OpenRouter — one key gives you access to GPT-5, Claude, Gemini, and more. Add direct provider keys later for cost optimization.
Create Telegram Bot
Get your bot token from BotFather in 2 minutes.
# Open Telegram and search for @BotFather # Send: /newbot # Name your bot (e.g., "My AI Assistant") # Username: my_ai_bot (must end in _bot) # Save the token: 1234567890:ABCdefGHIjklMNOpqrsTUVwxyz # Also get your chat ID by messaging @userinfobot # Or run: curl "https://api.telegram.org/bot<TOKEN>/getUpdates"
💡 Note: Make note of your bot token and your personal chat ID. You'll need both for the config file.
Configure OpenClaw
Set up your config file with bot token and preferences.
# Generate default config
openclaw init
# Edit the config file (~/.openclaw/openclaw.config.json)
{
"model": "openrouter/openrouter/auto",
"plugins": {
"entries": {
"telegram": {
"config": {
"token": "YOUR_BOT_TOKEN_HERE",
"allowedUsers": ["YOUR_TELEGRAM_USER_ID"]
}
}
}
},
"defaults": {
"systemPrompt": "You are my personal AI assistant."
}
}💡 Note: Replace YOUR_BOT_TOKEN_HERE with the token from BotFather. Replace YOUR_TELEGRAM_USER_ID with your numeric user ID.
Install Essential Skills
Add skills to unlock key capabilities.
# Install ClawHub CLI for skill management npm install -g clawhub # Install popular skills clawhub install weather clawhub install summarize clawhub install things-mac clawhub install blogwatcher # List installed skills clawhub list # Browse 5,400+ skills at clawhub.com
💡 Note: Skills are just directories with a SKILL.md that the AI reads before executing tasks. You can also create your own.
Auto-Start with LaunchAgent
Run OpenClaw on boot, no login required.
# Create the LaunchAgent plist
cat > ~/Library/LaunchAgents/com.openclaw.daemon.plist << 'EOF'
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
"http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.openclaw.daemon</string>
<key>ProgramArguments</key>
<array>
<string>/opt/homebrew/bin/openclaw</string>
<string>start</string>
</array>
<key>RunAtLoad</key>
<true/>
<key>KeepAlive</key>
<true/>
<key>StandardOutPath</key>
<string>/tmp/openclaw.log</string>
<key>StandardErrorPath</key>
<string>/tmp/openclaw.err</string>
</dict>
</plist>
EOF
# Load it now
launchctl load ~/Library/LaunchAgents/com.openclaw.daemon.plist
# Check status
launchctl list | grep openclaw💡 Note: KeepAlive: true means macOS will restart OpenClaw automatically if it crashes. Logs go to /tmp/openclaw.log.
Create Your First Automation
Morning briefing sent to Telegram every day at 7am.
# Create the automation script cat > ~/.openclaw/automations/morning-briefing.sh << 'EOF' #!/bin/bash # Morning briefing — runs at 7am daily openclaw run "Send me a morning briefing: 1. Weather for today (San Francisco) 2. Top 3 tech news headlines 3. My calendar events for today 4. Crypto prices: BTC, ETH Keep it concise. Format nicely for Telegram." EOF chmod +x ~/.openclaw/automations/morning-briefing.sh # Add to crontab crontab -e # Add this line: # 0 7 * * * ~/.openclaw/automations/morning-briefing.sh
💡 Note: The OpenClaw run command executes a one-shot task and the result gets sent to your configured Telegram bot automatically.
Remote Access via SSH
Access your Mac Mini AI server from anywhere.
# Find your Mac Mini's local IP ipconfig getifaddr en0 # Ethernet (preferred) ipconfig getifaddr en1 # WiFi fallback # SSH from another machine ssh username@192.168.1.xxx # For remote access outside home network: # Option 1: Tailscale (easiest, free for personal) brew install tailscale tailscale up # Option 2: Port forwarding on your router (advanced) # Forward port 22 to your Mac Mini's local IP # Optional: add SSH key for passwordless login ssh-copy-id username@your-mac-mini
💡 Note: Tailscale is highly recommended for remote access. It creates a secure mesh VPN — no port forwarding, no public IP needed.
Security Hardening
Lock down your always-on server.
# Enable built-in firewall sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setglobalstate on sudo /usr/libexec/ApplicationFirewall/socketfilterfw --setstealthmode on # Enable automatic security updates sudo softwareupdate --schedule on # Enable FileVault encryption # (Do this via System Settings > Privacy & Security > FileVault) # Disable password hints sudo defaults write /Library/Preferences/com.apple.loginwindow RetriesUntilHint -int 0 # Check for listening ports sudo lsof -nP -iTCP -sTCP:LISTEN
💡 Note: Change the default SSH port (optional), use key-based auth only, and consider disabling password auth in /etc/ssh/sshd_config.
Cost Breakdown
Hardware (Mac Mini M1)
one-timeElectricity (~10W × 24/7)
recurringAI API costs (variable)
recurringTotal ongoing: ~$15–55/month
Hardware pays itself off in months vs. ChatGPT Pro ($200/mo)
vs ChatGPT Plus
$20/mo
with far less capability
Pro Tips
Energy Optimization
- System Settings → Energy → "Prevent automatic sleeping"
- Enable Power Nap for background tasks
- Connect via Ethernet for stable wake-on-LAN
Backup Strategy
- Time Machine to external USB drive
- Back up ~/.openclaw/ to cloud (sensitive configs)
- Automate with rsync + cron for offsite
Monitoring
- tail -f /tmp/openclaw.log for real-time logs
- Set up uptime alerts in your morning briefing
- Use htop for resource monitoring via SSH
FAQ
brew install ollama) and run models like Llama 3, Mistral, or Qwen locally. The M1 8GB handles 7B models well; 16GB handles 13B+ models. No API key, no cost, full privacy.Ready to build your AI server?
This guide covers the full Mac Mini setup. Start with the general guide, or go straight to VPS hosting if you'd rather skip the hardware.