How to Deploy OpenClaw on VPS: Complete Guide (2026)

What You'll Need

• A VPS (any provider—DigitalOcean, Linode, Hetzner, AWS)
• Ubuntu 22.04 LTS (recommended)
• SSH access
• A Telegram account
• An AI provider API key (we'll use free Kimi)
• Basic terminal knowledge (we'll explain each command)

Cost: $5-10/month for VPS + API usage

Time: 1-2 hours (first time), 30 min (if you know what you're doing)

Why Deploy on Your Own VPS?

Benefits:

• Complete control over your infrastructure
• Cheaper long-term ($5-10/mo vs $49/mo managed)
• Learn how OpenClaw works under the hood
• Customize everything

Drawbacks:

• You maintain everything yourself
• Need to handle security updates
• Troubleshooting is on you
• Takes 1-2 hours to set up

Step 1: Create Your VPS (10 minutes)

DigitalOcean (Recommended for Beginners)

1. Go to digitalocean.com and create account
2. Click "Create Droplet"
3. Choose Ubuntu 22.04 LTS
4. Select "Basic" plan
5. Pick $6/month option (1 GB RAM, 1 vCPU—enough for OpenClaw)
6. Choose region closest to you
7. Add SSH key (or use password—we'll secure it later)
8. Click "Create Droplet"

Alternative providers:

• Linode: Similar pricing, great support
• Hetzner: Cheapest ($4/mo), EU-based
• Vultr: More datacenter locations

All providers work the same. Pick whichever you prefer.

Step 2: Connect to Your VPS (5 minutes)

On Mac/Linux:

ssh root@your-vps-ip

On Windows:

Use PuTTY or Windows Terminal with WSL.

You'll see something like:

Welcome to Ubuntu 22.04 LTS
root@openclaw:~#

You're in. Now we install OpenClaw.

Step 3: Install Node.js (5 minutes)

OpenClaw runs on Node.js. Install it:

# Update package list
apt update && apt upgrade -y

# Install Node.js 20.x (LTS)
curl -fsSL https://deb.nodesource.com/setup_20.x | bash -
apt-get install -y nodejs

# Verify installation
node --version   # Should show v20.x.x
npm --version    # Should show 10.x.x

Why Node 20? OpenClaw requires Node 18+ for modern JavaScript features.

Step 4: Install OpenClaw (7 minutes)

# Clone OpenClaw repository
git clone https://github.com/openclaw/openclaw.git
cd openclaw

# Install dependencies
npm install

# Build the project
npm run build

This takes 3-5 minutes. You'll see a bunch of package names scroll by. That's normal.

If you see errors: Check Node version. Needs to be 18+.

Step 5: Configure OpenClaw (10 minutes)

# Copy example config
cp config.example.json config.json

# Edit with nano (or vim if you prefer)
nano config.json

Key settings to change:

{
  "gateway": {
    "host": "localhost",    // IMPORTANT: localhost only for security
    "port": 3000
  },
  "telegram": {
    "botToken": "YOUR_BOT_TOKEN_HERE"  // Get from @BotFather
  },
  "ai": {
    "provider": "nvidia",               // Free Kimi via Nvidia
    "apiKey": "nvapi-...",             // Your Nvidia API key
    "model": "kimi-k2.5"
  }
}

Security note: Always set host: "localhost". Never expose to 0.0.0.0 without authentication.

Save with: Ctrl+X, then Y, then Enter

Step 6: Get Your Telegram Bot Token (10 minutes)

1. Open Telegram, search for @BotFather
2. Send /newbot
3. Choose a name: "My OpenClaw Bot"
4. Choose a username: "myopenclaw_bot" (must end in _bot)
5. Copy the token (looks like: 1234567890:ABCdefGHIjklMNOpqrsTUVwxyz)
6. Paste it in config.json under telegram.botToken

Step 7: Get Free AI Credits (Kimi via Nvidia) (10 minutes)

Why Kimi? 50,000 free tokens/day. No credit card required.

1. Go to build.nvidia.com/moonshot
2. Sign up with email
3. Click "Get API Key"
4. Copy the key (starts with nvapi-...)
5. Paste in config.json under ai.apiKey

Alternative: Use Claude or GPT

• Claude: console.anthropic.com (paid)
• GPT: platform.openai.com (paid)

Step 8: Start OpenClaw (2 minutes)

# Start in foreground (to test)
npm start

You should see:

[gateway] gateway listening on localhost:3000
[telegram] starting provider
[telegram] bot connected: @your_bot_name

Test it: Message your bot on Telegram. It should reply.

If it works, press Ctrl+C to stop. We'll make it run 24/7 next.

Step 9: Make It Run 24/7 with PM2 (5 minutes)

# Install PM2 (process manager)
npm install -g pm2

# Start OpenClaw with PM2
pm2 start npm --name "openclaw" -- start

# Make it start on boot
pm2 startup
pm2 save

Useful PM2 commands:

pm2 status          # Check if running
pm2 logs openclaw   # View logs
pm2 restart openclaw  # Restart
pm2 stop openclaw   # Stop

Step 10: Security Hardening (10 minutes)

CRITICAL: Don't skip this step.

10.1: Set Up Firewall (UFW)

# Install UFW
apt install ufw

# Default rules: deny incoming, allow outgoing
ufw default deny incoming
ufw default allow outgoing

# Allow SSH (otherwise you'll lock yourself out)
ufw allow 22/tcp

# Enable firewall
ufw enable

# Check status
ufw status

10.2: Install fail2ban (Blocks Brute Force)

apt install fail2ban
systemctl enable fail2ban
systemctl start fail2ban

10.3: Create Non-Root User

# Create user
useradd -m openclaw

# Add to sudo group
usermod -aG sudo openclaw

# Switch to this user for running OpenClaw
su - openclaw

Why? Running as root is dangerous. If OpenClaw gets compromised, attacker has full system access.

Step 11: Verify Everything Works (5 minutes)

Checklist:

• [ ] VPS is running
• [ ] Node.js installed (version 20+)
• [ ] OpenClaw cloned and built
• [ ] config.json configured
• [ ] Telegram bot token added
• [ ] AI provider (Kimi) connected
• [ ] PM2 running OpenClaw 24/7
• [ ] Firewall (UFW) active
• [ ] fail2ban protecting SSH

Test: Message your Telegram bot. Should reply instantly.

Common Issues & Fixes

"Bot not responding"

# Check if OpenClaw is running
pm2 status

# Check logs for errors
pm2 logs openclaw

# Verify Telegram token
curl https://api.telegram.org/bot<YOUR_TOKEN>/getMe

"API key invalid"

• Check you copied the full key
• Ensure no extra spaces in config.json
• Verify key hasn't expired (Nvidia keys don't expire, but check dashboard)

"Gateway connection refused"

# Check if port 3000 is listening
netstat -tlnp | grep 3000

# Restart OpenClaw
pm2 restart openclaw

"npm install" fails

• Check Node version: node --version (needs 18+)
• Try: npm install --legacy-peer-deps
• Clear cache: npm cache clean --force

Maintenance Tips

Weekly:

• Check PM2 status: pm2 status
• View logs for errors: pm2 logs openclaw --lines 50

Monthly:

• Update system: apt update && apt upgrade -y
• Update OpenClaw: cd openclaw && git pull && npm install && npm run build && pm2 restart openclaw

Check disk space:

df -h

If / is above 80%, clean logs:

pm2 flush openclaw

Upgrading to Paid AI Models

Free Kimi is great to start, but has limits (50k tokens/day).

When you need more:

Add Claude API

1. Get key from console.anthropic.com
2. Update config.json:

"ai": {
  "provider": "anthropic",
  "apiKey": "sk-ant-...",
  "model": "claude-opus-4.5"
}

3. Restart: pm2 restart openclaw

Switch to GPT

"ai": {
  "provider": "openai",
  "apiKey": "sk-proj-...",
  "model": "gpt-5.2"
}

Next Steps

Your OpenClaw is running. Now what?

1. Add more channels: Discord, Email, WhatsApp
2. Enable browser automation: Let it browse the web
3. Set up web search: Connect to Google/Bing API
4. Create custom skills: Teach it new abilities
5. Backup config: cp config.json config.backup.json

When Manual Setup Isn't Worth It

If this guide feels overwhelming, or you don't want to maintain a VPS yourself, consider the managed option.

5-minute automated deploy: open-claw.space

• No terminal required
• Security handled automatically
• Automatic updates
• $49/mo (includes $15 AI credits)

Most people do manual setup once to learn how it works, then realize maintaining it isn't worth their time.

Resources

• OpenClaw Docs: docs.openclaw.ai
• Community: GitHub Discussions
• Free AI (Nvidia): build.nvidia.com

---

Last updated: February 2026