OpenClaw on Hetzner Cloud
Best value VPS for self-hosting OpenClaw — EU GDPR, step-by-step setup
Best value VPS for self-hosting OpenClaw — EU GDPR, step-by-step setup
Hetzner Cloud is widely recommended for running OpenClaw—you get 2–3x more compute per dollar than AWS, GCP, or Azure, with full GDPR compliance from German and Finnish data centers. For OpenClaw, the AI work runs on your LLM provider's servers, so your VPS only handles Node.js, channel connections, and the gateway. A modest Hetzner instance is plenty.
We've run OpenClaw on CX11 (2GB RAM) for light use and CX22/CX32 for busier setups. This guide walks through creating a server, installing OpenClaw, and keeping it running with PM2.
Hetzner Cloud offers several tiers. For OpenClaw:
Local LLMs (Ollama) need more RAM; plan for 8GB+ if you want to run models on the same machine. For cloud LLMs (Anthropic, OpenAI), 2–4GB is enough.
ssh root@YOUR_SERVER_IP
apt update && apt upgrade -y
Use the one-liner—it installs Node.js if needed and OpenClaw:
curl -fsSL https://openclaw.ai/install.sh | bash
Then run the onboarding wizard. For a headless server, you'll configure via CLI or copy a config from another machine. See Configuration guide.
PM2 keeps OpenClaw running after you disconnect and across reboots:
npm install -g pm2
openclaw init
pm2 start 'openclaw gateway start' --name openclaw
pm2 save
pm2 startup
Follow the pm2 startup output to enable startup on boot.
Enable UFW and allow SSH, HTTP, HTTPS:
ufw allow 22
ufw allow 80
ufw allow 443
ufw enable
OpenClaw's default port (18789) is loopback-only. For web access, use a reverse proxy (nginx, Caddy) in front. See Deployment tutorial for SSL and reverse proxy setup.