Deploying OpenClaw on Jetson

1. Introduction

OpenClaw is an open-source (MIT) self-hosted gateway that connects popular instant messaging apps (Discord, Telegram, WhatsApp, Signal, etc.) to an AI Agent via channel plugins, with tool-calling, session, and memory capabilities. All data runs on your own hardware, making it ideal for building a local, conversational AI assistant on a Jetson Orin carrier board.

OpenClaw is built on Node.js, and Jetson Orin runs Ubuntu on the ARM64 (aarch64) architecture, so it can be deployed directly on the board.

Scope

This tutorial applies to both Orin Nano and Orin NX series carrier boards (C1901 / C1902 / C2401); the steps are identical.

2. Requirements

Item	Requirement
Hardware	Jetson Orin Nano / NX carrier board (system already flashed)
OS	JetPack 5 / 6 (Ubuntu, ARM64)
Node.js	22.19+ / 23.11+ / 24+ (24 recommended)
Network	Internet access (required for installation and model API calls)
Model API Key	A key from any provider such as Anthropic, OpenAI, or Google

Note

The Node.js version in the apt sources bundled with JetPack is usually too old and cannot be used directly. Follow step 3 below to install a Node.js version that meets the requirements.

3. Install Node.js

JetPack's default Node version is too low. We recommend using nvm to install a newer version — it keeps versions isolated and easy to switch.

# Install nvm
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash
# Make nvm take effect immediately (or reopen the terminal)
source ~/.bashrc

# Install and enable Node.js 24
nvm install 24
nvm use 24
nvm alias default 24

After installation, verify the version (must be ≥ 22.19):

node --version

4. Install OpenClaw

Once Node.js is ready, run the official one-click install script:

curl -fsSL https://openclaw.ai/install.sh | bash

Alternatively, install it globally with npm: npm install -g openclaw@latest

5. Initialize and Install the Daemon

Run the onboarding wizard, which also installs OpenClaw as a background daemon (starts on boot and keeps running after disconnection):

openclaw onboard --install-daemon

The wizard will guide you through:

1. Choosing a model provider (Anthropic / OpenAI / Google, etc.)
2. Entering your API Key
3. Configuring and starting the Gateway

The configuration file is saved at ~/.openclaw/openclaw.json and can be edited manually later (e.g., to change the port):

{
  gateway: {
    port: 18789,
  },
}

6. Verify and Access the Control Panel

Check whether the Gateway is running properly (it listens on port 18789 by default):

openclaw gateway status

Open the Control UI:

openclaw dashboard

The default address is http://127.0.0.1:18789/. Send a message in the panel's chat box to test whether the AI responds normally.

7. Remote Access on a Headless Jetson

Jetson boards are usually used headless (without a display), and the Gateway only listens on the loopback address by default. We recommend using an SSH tunnel to securely forward the port to your computer:

# Run on your own computer (replace user@jetson-ip with the actual address)
ssh -L 18789:localhost:18789 user@jetson-ip

Keep the SSH connection open, then open http://127.0.0.1:18789/ in a browser on your local machine to access the panel on the Jetson.

Security Note

If you need to expose the panel directly on the LAN/internet (by changing the bind address or adding an nginx reverse proxy), make sure to configure authentication first, and treat external input such as webhooks as untrusted data. Whenever possible, prefer the SSH tunnel.

8. Connect Chat Channels

To chat from your phone anytime, the fastest option is Telegram — you only need a Bot Token to connect. For other channels such as Discord, WhatsApp, and Signal, see the Channels section of the official documentation.

9. FAQ

• "Node version not satisfied": Confirm node --version ≥ 22.19; if using nvm, check that nvm use 24 has taken effect and set it as default.
• Install script errors saying npm / node not found: Complete step 3 to install Node.js first, then run the install script.
• Panel won't open: First confirm the service is running with openclaw gateway status; on headless devices, remember to use an SSH tunnel (step 7).
• Resource usage: Model inference goes through the provider's cloud API by default, so local load on the Jetson is minimal; if you switch to a local LLM instead, pay attention to VRAM and compute capacity.

---

Reference: OpenClaw Official Documentation. Installation commands, the daemon, and panel ports are subject to the official documentation; this page provides deployment notes adapted for Jetson (ARM64).