# OpenClaw on GCP: Deploy OpenClaw on Google Cloud Platform > Deploy OpenClaw on GCP (Google Cloud Platform) and run your AI agent in the cloud. This guide shows how to set up a GCP VM and launch OpenClaw step-by-step. **Published:** Mar 28, 2026 | **Author:** Alex Chen | **Read time:** 12 min read Learn how to run OpenClaw on GCP using Google Cloud VM. Step-by-step guide with Docker setup, requirements, and deployment instructions. --- Run [OpenClaw](/blog/what-is-openclaw) on Google Cloud Platform (GCP) to keep your AI agent running 24/7 without relying on your local machine. This setup is ideal if you want a reliable production deployment with low monthly cost and high uptime. Run [OpenClaw](/blog/what-is-openclaw) on Google Cloud Platform (GCP) to keep your AI agent running 24/7 without relying on your local machine. This setup is ideal if you want a reliable production deployment with low monthly cost and high uptime. ## Minimum PC Requirements | Component | Minimum Requirement | Recommended | | --- | --- | --- | | CPU | Dual-Core Processor | Quad-Core (Intel i5 / Ryzen 5) | | RAM | 4GB | 8GB+ | | Storage | 10GB Free | 20GB+ | | Internet | 10 Mbps | 25 Mbps+ | | Operating System | Windows 10 / Linux / macOS | Windows 11 / Linux | | Terminal Access | Command Prompt / PowerShell / Terminal | Windows Terminal | | Browser | Chrome / Edge / Firefox | Latest version | ## Recommended GCP VM Specs | Machine Type | CPU | RAM | Recommendation | | --- | --- | --- | --- | | e2-micro | Shared | 1GB | Not recommended | | e2-small | 2 vCPU | 2GB | Minimum | | e2-medium | 2 vCPU | 4GB | Recommended | ## How to Run OpenClaw on GCP? ### Step 1: Install the gcloud CLI Install the Google Cloud CLI on your local machine: Once installed, initialize and authenticate: Alternatively, every step in this guide can also be done through the Google Cloud Console at `https://console.cloud.google.com/` ### Step 2: Create a GCP Project Create a new project for your OpenClaw deployment: Enable billing for the project at `https://console.cloud.google.com/billing` (required for Compute Engine). Then enable the Compute Engine API: ### Step 3: Create the Compute Engine VM Create a VM with Debian 12 and a 20GB boot disk: ### Step 4: SSH Into the VM If the connection is refused, wait 1–2 minutes after VM creation for SSH key propagation, then retry. ### Step 5: Install Docker Inside the VM, run: Log out for the group change to take effect: SSH back in: Verify Docker is working: ### Step 6: Clone the OpenClaw Repository Then create directories: This ensures your config and workspace survive container rebuilds and VM restarts. ### Step 7: Configure Environment Variables Create a `.env` file in the repository root: Generate strong secrets for `OPENCLAW_GATEWAY_TOKEN` and `GOG_KEYRING_PASSWORD`: Do not commit this file to version control. ### Step 8: Configure Docker Compose Create `docker-compose.yml` in the repository root: ```bash {`services: openclaw-gateway: image: \$ build: . restart: unless-stopped env_file: - .env environment: - HOME=/home/node - NODE_ENV=production - TERM=xterm-256color - OPENCLAW_GATEWAY_BIND=\$ - OPENCLAW_GATEWAY_PORT=\$ - OPENCLAW_GATEWAY_TOKEN=\$ - GOG_KEYRING_PASSWORD=\$ - XDG_CONFIG_HOME=\$ - PATH=/home/linuxbrew/.linuxbrew/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin volumes: - \$:/home/node/.openclaw - \$:/home/node/.openclaw/workspace ports: - "127.0.0.1:\$:18789" command: [ "node", "dist/index.js", "gateway", "--bind", "\$", "--port", "\$", "--allow-unconfigured", ]`} ``` The port binding `127.0.0.1:18789` keeps the gateway loopback-only on the VM. You access it securely through an SSH tunnel. ### Step 9: Build and Launch OpenClaw Build the Docker image and start the container: Check the container is running: ### Step 10: Configure Allowed Origins When binding to LAN, set the trusted browser origin: ### Step 11: Access the Dashboard from Your Local Machine From your local machine (not the VM), open an SSH tunnel: Open in your browser: Get a tokenized dashboard link: If the dashboard shows *unauthorized* or *disconnected (1008): pairing required*, approve your browser: After setup, connect a messaging channel — Telegram, [WhatsApp](/blog/connect-openclaw-whatsapp), Discord, [Slack](https://www.ampere.sh/openclaw-for-slack), Microsoft Teams, or any other supported app — to start chatting with your agent. ## Common Issues and Fix | Issue | Fix | | --- | --- | | SSH connection refused | Wait 1–2 min after VM creation for SSH key propagation | | Docker build fails (exit 137 / OOM) | gcloud compute instances set-machine-type openclaw-gateway --zone=us-central1-a --machine-type=e2-small | | docker: command not found | Log out and back in after sudo usermod -aG docker $USER | | docker compose not found | sudo apt-get install -y docker-compose-plugin | | Gateway container not starting | docker compose logs -f to check error output | | Dashboard shows unauthorized | docker compose run --rm openclaw-cli devices approve | | OS Login permission error | gcloud compute os-login describe-profile — check IAM permissions | | Container stops after reboot | Confirm restart: unless-stopped is set in docker-compose.yml | If you prefer to skip these setup steps entirely, you can deploy OpenClaw on [Ampere.sh](https://www.ampere.sh) in 60 seconds. No Docker, no config files, no infrastructure management required. ## Avoid Cloud Configuration Setting up OpenClaw on GCP requires VM creation, Docker setup, and networking configuration. Run OpenClaw on Ampere.sh and start your AI agent in 60 seconds. [Deploy on Ampere.sh →](https://www.ampere.sh/setup) ## Frequently Asked Questions ### What is OpenClaw on GCP? OpenClaw on GCP means running your OpenClaw AI agent on a Google Cloud virtual machine instead of your local computer. This allows your agent to run 24/7 with better reliability and performance. ### Can I use GCP free tier for OpenClaw? Yes, but it's not recommended. The free-tier e2-micro (1GB RAM) often fails during Docker builds or OpenClaw startup. ### Do I need Docker to run OpenClaw on GCP? Yes. OpenClaw is typically deployed using Docker to ensure stable and persistent setup. ### Does OpenClaw continue running if I close my PC? Yes. Once deployed, OpenClaw runs 24/7 on GCP even if your PC is turned off. ### Can I upgrade my VM later? Yes. You can easily upgrade RAM, CPU, and Storage without reinstalling OpenClaw. ### What happens if my VM stops? OpenClaw stops temporarily. Once you restart the VM, OpenClaw resumes automatically if Docker restart is enabled. ### Do I need technical knowledge? Basic knowledge helps, but most steps are copy-paste friendly. ### Can I connect OpenClaw to Telegram or WhatsApp? Yes. After deployment, you can connect Telegram, Discord, WhatsApp, Slack, and more.