How to Deploy OpenClaw on a Hostinger VPS in 2026: Beginner Step-by-Step Guide

Transparency note: the Hostinger links in this article are affiliate links. If you sign up through them, we may earn a small commission at no extra cost to you. We only recommend hosts we've actually used for self-hosted AI workloads.

Running OpenClaw on a laptop is fine for an afternoon of tinkering. The moment you want it to answer your Telegram at 3 a.m., keep a WhatsApp pairing alive for weeks, or let a long-running browser task finish while your machine sleeps, you need an always-on box with a public IP. A small VPS is the cheapest way to get one, and a Hostinger KVM plan is a sensible starter choice — low monthly cost, one-click Ubuntu 24.04 images, a browser SSH console, and enough headroom to deploy OpenClaw on a Hostinger VPS without fighting your hosting provider.

This guide assumes zero prior VPS experience. It walks from "I just bought the plan" to "my agent replies to me on Telegram," with every command you need and the realistic security precautions that matter when the agent has shell access.

OpenClaw itself is an open-source, MIT-licensed personal AI assistant built in TypeScript. It runs on Node 24, responds across WhatsApp, Telegram, Discord, Slack, Signal, iMessage and 15+ other messengers, and ships with Docker-based sandboxing for the skills it runs on your behalf. That combination — pluggable model backend, many channels, local-first data — is what makes a VPS install worth doing properly.

Why Deploy OpenClaw on a VPS Instead of a Laptop

Three reasons it's worth moving off your laptop. Uptime first: messaging integrations drop when a laptop sleeps, and some channels (notably WhatsApp and iMessage via BlueBubbles) do not reconnect cleanly after long offline periods. IP reputation next: residential IPs get rate-limited or flagged by commercial APIs, while VPS IPs tend to be cleaner for outbound API calls. And finally isolation — OpenClaw can execute shell commands, so keeping that blast radius off your personal machine is just good hygiene.

For a single user, a VPS also solves the "where do I put secrets" problem. Your Anthropic or OpenAI key, your Telegram bot token, your OpenClaw memory store — all live on one box you can reason about, back up, and rotate cleanly.

Why Hostinger for OpenClaw (Honest Pros and Cons)

Hostinger is not the only reasonable option, and it's fair to be upfront about where it fits.

Where Hostinger works well: low monthly cost on the entry KVM plans, one-click Ubuntu 24.04 LTS images, a built-in firewall in the control panel, browser-based SSH so a broken key doesn't lock you out, and straightforward IPv4 allocation. For a single-user AI assistant that doesn't need multi-region failover, the bottom and middle tiers are usually plenty.

Where it's not ideal: CPU performance on the smallest shared-vCPU plans can be noisy during peak hours, and region coverage is narrower than DigitalOcean or Hetzner. If you need five datacenters or predictable CPU under sustained load, look elsewhere. We compare the alternatives in Best VPS for OpenClaw.

For most people self-hosting a personal AI agent, Hostinger is the cheapest path to always-on that doesn't require you to become a sysadmin. That's the job here.

If you don't have an account yet, you can grab a Hostinger VPS plan here. Using that affiliate link supports these guides at no extra cost to you.

Choose the Right Hostinger VPS Plan

Minimum specs for OpenClaw

OpenClaw's gateway (Node 24 process plus the memory store) runs comfortably in about 1 GB of RAM. Add Docker sandboxes for skills, a headless Chromium for browser control, and enough headroom so you're not paging to swap, and you land in this range:

Which KVM plan to pick

At time of writing, Hostinger's KVM 2 tier is the practical starting point for light use, and KVM 4 is where most readers will be happiest long-term. Skip the smallest KVM 1 plan if you plan to use browser control — Chromium alone eats 1 to 2 GB, and OpenClaw's own process plus Docker overhead will push a 2 GB box into swap within minutes.

If you're undecided, start one tier above what looks "enough." Upgrading is a reboot; debugging OOM kills during a WhatsApp handshake is a weekend.

What You'll Need Before You Start

• A Hostinger account with a purchased KVM plan (sign up here if you don't have one).
• Ubuntu 24.04 LTS selected as the VPS OS (64-bit).
• An SSH client on your local computer. macOS and Linux have one built in; Windows 11 users can use the built-in OpenSSH client or MobaXterm.
• An API key for your model provider — Anthropic Claude or OpenAI.
• The bot credentials for whichever messenger you plan to pair first. A Telegram BotFather token is the simplest starting point; a Discord bot token from the Discord Developer Portal is a close second.
• About 45 to 60 minutes.

Step 1 — Order and Provision Your Hostinger VPS

In the Hostinger panel, pick your KVM plan, select Ubuntu 24.04 LTS (clean) as the operating system, choose a data center close to you, and set a strong root password. Hostinger emails the IPv4 once the VM is provisioned, which usually takes a minute or two.

During initial setup, add your SSH public key if the panel offers the option. If you don't see it, you'll handle keys manually in Step 2. Either way, avoid password-only login past the first boot.

Test the connection from your terminal:

ssh root@YOUR_VPS_IP

If that prompts for the root password you set, you're in. Don't install anything yet — harden the box first.

Step 2 — Harden SSH and Enable the Firewall

OpenClaw is an agent with shell access. Locking the front door before installing anything is not optional.

# Create a non-root sudo user
adduser claw
usermod -aG sudo claw

# Copy your SSH key over to the new user
rsync --archive --chown=claw:claw ~/.ssh /home/claw

# Disable root login and password auth
sudo sed -i 's/^#\?PermitRootLogin.*/PermitRootLogin no/' /etc/ssh/sshd_config
sudo sed -i 's/^#\?PasswordAuthentication.*/PasswordAuthentication no/' /etc/ssh/sshd_config
sudo systemctl restart ssh

# Basic UFW firewall
sudo ufw allow OpenSSH
sudo ufw enable

# Automatic security updates
sudo apt update && sudo apt install -y unattended-upgrades
sudo dpkg-reconfigure -plow unattended-upgrades

Log out, then log back in as claw over SSH key auth. Only move on once that works. If you lock yourself out, Hostinger's browser SSH console is your rescue — it bypasses the firewall and auth settings.

Step 3 — Install Node 24, pnpm, and Docker

OpenClaw requires Node 24 (or Node 22.16+), and Docker is how it sandboxes skill execution. Install both on the same box.

# Node 24 via NodeSource
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs

# pnpm (preferred, but npm works for the onboard CLI)
sudo npm install -g pnpm

# Docker Engine (official repository)
sudo apt install -y ca-certificates curl gnupg
sudo install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg \
  | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] \
  https://download.docker.com/linux/ubuntu $(. /etc/os-release && echo "$VERSION_CODENAME") stable" \
  | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
sudo apt update
sudo apt install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

# Let your user run Docker without sudo
sudo usermod -aG docker $USER
newgrp docker

# Quick sanity check
node --version      # v24.x
docker run --rm hello-world

If either command errors out, fix it before continuing — the OpenClaw onboarder assumes both work without sudo.

Step 4 — Install OpenClaw via the Onboarding CLI

OpenClaw ships a global CLI that handles workspace creation, model selection, and the gateway daemon. From the official install documentation:

npm install -g openclaw@latest
openclaw onboard --install-daemon

The wizard walks you through creating a workspace, naming your assistant, picking a model backend, and pairing a first messenger. It writes config to ~/.openclaw/openclaw.json and secrets to the accompanying .env.

If you prefer to hack on the source directly, the alternative path is:

git clone https://github.com/openclaw/openclaw.git
cd openclaw
pnpm install
pnpm openclaw setup
pnpm gateway:watch

For a VPS install, stick with the global CLI unless you plan to modify the gateway. It's simpler to update and plays nicely with systemd in Step 6.

Step 5 — Configure Your Model Provider and First Messenger

When the onboarder asks for a model provider, paste your Anthropic or OpenAI key. OpenClaw also supports local models if you're running something like Ollama elsewhere on the network, but on a small VPS you'll get better latency and lower cost by using a hosted API.

For Telegram, create a bot in BotFather, copy the token, and paste it when OpenClaw asks. For Discord, go to the Discord Developer Portal, create an application, generate a bot token, and invite the bot into your server with the scopes the OpenClaw wizard lists.

Do not disable DM pairing for unknown senders. This is OpenClaw's default, and it's there because your agent has shell access. A prompt-injection from a random Telegram stranger can become remote code execution if you let an untrusted DM through. If you want to let a friend talk to your assistant, pair them explicitly — that's a 30-second step in the wizard.

Step 6 — Run OpenClaw as a systemd Service

pnpm gateway:watch or openclaw gateway in a terminal works for testing, but you want the gateway to survive reboots and restart on crash. That's a systemd unit.

sudo tee /etc/systemd/system/openclaw.service > /dev/null <<'EOF'
[Unit]
Description=OpenClaw Gateway
After=network-online.target docker.service
Requires=docker.service

[Service]
Type=simple
User=claw
WorkingDirectory=/home/claw
ExecStart=/usr/bin/openclaw gateway
Restart=on-failure
RestartSec=5

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl daemon-reload
sudo systemctl enable --now openclaw
sudo systemctl status openclaw

Adjust WorkingDirectory and ExecStart if your install paths differ. To tail logs: journalctl -u openclaw -f.

Step 7 — Back Up ~/.openclaw and .env

Your memory store, skills, and API keys live in ~/.openclaw. Losing that directory means losing everything the assistant knows about you, so a nightly tarball is worth setting up on day one. Add this line to crontab -e:

0 3 * * * tar czf /home/claw/backups/openclaw-$(date +\%F).tgz /home/claw/.openclaw

Then either enable Hostinger's built-in snapshot schedule from the panel, or push the tarball to an S3-compatible bucket with rclone if you want off-box protection. Treat .env like a password file — never commit it anywhere.

Troubleshooting Common Errors

• EACCES on Docker commands. You forgot usermod -aG docker $USER followed by newgrp docker. Run both, then reopen the shell.
• Telegram bot silently ignores your messages. DM pairing wasn't completed. Open OpenClaw's admin CLI (openclaw pair) and follow the prompt.
• Node complains about an unsupported engine. You installed the distro's default Node, not Node 24. Remove it (sudo apt remove nodejs) and re-run the NodeSource script.
• High memory pressure or OOM kills in journalctl. Browser control plus Docker adds up fast. Upgrade to the next KVM tier — this is the most common reason to move from KVM 2 to KVM 4.
• Gateway dies overnight. Check journalctl -u openclaw -n 200. Nine times out of ten it's the OOM killer or a transient network blip; the Restart=on-failure line brings it back either way.

Is a Hostinger VPS Powerful Enough to Run OpenClaw?

Yes, for a single user. The KVM 2 plan (2 vCPU, 8 GB RAM on current configurations) is enough for Telegram, Discord, Slack, and a few lightweight skills. If you plan to use browser control heavily — scraping sites, filling long forms, keeping Chromium resident — move to KVM 4 for the extra RAM and dedicated CPU headroom. If your assistant is read-only and chat-only, KVM 2 will feel snappy; if you're running five skills that each spin up a Docker container, KVM 4 is the right starting point.

FAQ

Do I need a domain name to run OpenClaw? No. Most messengers (Telegram, Discord, Slack, Signal) pair with OpenClaw via bot tokens and API polling, not inbound webhooks. You can add a domain later if you want a browser-based admin UI, but it's not required for the messaging flow.

Can I use Hostinger's shared hosting plan instead of a VPS? No. OpenClaw needs to run persistent background processes and Docker containers, neither of which shared hosting allows. A KVM VPS (or any other root-access Linux host) is the minimum.

What does running OpenClaw actually cost per month? VPS in the single digits of USD per month, plus your model API usage. For typical personal chatter with Anthropic or OpenAI, expect $5–$15 monthly in API fees. Heavy browser-control workflows can push the API bill higher — cap spending in the provider dashboard until you know your usage pattern.

Is it legal to run a WhatsApp or iMessage bridge? For personal use, it's generally acceptable, but each platform's terms of service vary. WhatsApp and iMessage have stricter rules than Telegram or Discord — bridging either for commercial use or for other people's accounts can get you banned. Keep it to your own account, and read OpenClaw's security documentation before exposing the gateway to any channel.

Can I move from Hostinger to another host later? Yes. Tar ~/.openclaw on the old box, copy it to the new one, reinstall the CLI, and run openclaw onboard. The workspace config and memory store port over cleanly.

What to Do Next

• If you want a broader overview of the whole install path (Docker-only, Podman, Raspberry Pi, home server), read the OpenClaw self-hosting guide for beginners.
• Before you paste your first API key anywhere, read OpenClaw security risks and safe self-hosting. The agent has shell access — the defaults are sensible but the footguns are real.
• If you want to understand what OpenClaw actually is and what it can do before going deeper, start with what is OpenClaw.
• Shopping around? Compare hosting options in the best VPS for OpenClaw.

Ready to deploy? Grab a Hostinger KVM plan here and you can have OpenClaw answering your messages within an hour.