If you’ve been using OpenClaw for a while, you’ve probably noticed one annoying limitation: the moment you close your laptop or put it to sleep, your agent goes offline. No more heartbeat checks. No more proactive emails. No more scheduled tasks running in the background. It just stops.
There’s a simple fix for this, and it doesn’t require expensive hardware or a complicated server setup. You move OpenClaw to a VPS , a small virtual server that stays online 24/7 whether you’re working, sleeping, or on vacation.
Why Your Laptop Isn’t a Good Host for an AI Agent
Laptops are designed to be portable and power-efficient. That means they sleep, they throttle, and they’re not always connected to a stable network. For OpenClaw to do its job , checking your calendar, monitoring your inbox, running scheduled tasks , it needs to be consistently online.
Running it on your laptop also means:
- Your agent goes offline every time you close the lid
- Your laptop runs warm and uses more power around the clock
- Battery life degrades faster from constant operation
- Any network hiccup on your home connection drops the agent
A VPS solves all of this. It sits in a data centre, stays online, and costs a few dollars a month.
What to Look for in a VPS for OpenClaw
Not all VPS hosting is created equal when it comes to running AI agents. You want something that’s:
- Always on , no sleep modes, no downtime windows
- Low cost , OpenClaw doesn’t need massive compute, just reliable uptime
- Easy to manage , you’re not a sysadmin, you just want it to work
- Fast enough for Node.js , OpenClaw runs on Node, so modern ARM chips handle it well
This is exactly why I use Ampere.sh for hosting OpenClaw. It’s a managed VPS platform built specifically for running AI agents , ARM-based servers, fully managed infrastructure, and plans that start under $10/month.
How to Move OpenClaw to a VPS
The process is simpler than it sounds. Here’s the rough overview:
1. Spin up a VPS on Ampere.sh
Sign up, pick a plan, and you’ll have a server ready in minutes. No credit card required to start.
2. Install Node.js on the server
SSH into your server and install Node.js (version 18 or higher works well). Most VPS providers include a one-click Node setup or a simple apt install command.
3. Install OpenClaw
Run npm install -g openclaw on the server, just like you did on your local machine.
4. Copy your config
Transfer your openclaw.json config file and your workspace folder to the server. Your SOUL.md, MEMORY.md, and all your agent files come with it.
5. Run OpenClaw as a background service
Use a process manager like pm2 to keep OpenClaw running after you log out. One command: pm2 start openclaw --name agent. Done.
From that point on, your agent runs continuously , checking, responding, and acting , without your laptop needing to be open.
What Changes When Your Agent Is Always On
The difference is noticeable quickly. Heartbeat checks actually run on schedule. Proactive alerts arrive when they should. Scheduled tasks don’t get skipped because your laptop was closed at 3am.
For a deeper look at how OpenClaw works and what it can do once it’s running 24/7, check out the complete OpenClaw guide.
Ready to keep OpenClaw running around the clock? Ampere.sh has managed ARM VPS plans built for AI agents , no laptop required. Get started free →
This post may contain affiliate links. We may earn a commission if you make a purchase, at no extra cost to you.
Not Ready to Self-Host? Try ClawTrust
Building agents locally is powerful, but managing infrastructure, security, and scaling takes effort. ClawTrust provides managed OpenClaw hosting — your agents run securely in the cloud with automatic scaling, no DevOps overhead, and 24/7 monitoring.
This post contains affiliate links. If you buy through them, I may earn a small commission at no extra cost to you.
About The Author
This is exactly my problem. My agent keeps going offline when I close my laptop at night. Does it matter which VPS provider I use or is Ampere.sh specifically recommended for OpenClaw?
Any VPS with Linux and Node.js support will technically work, but Ampere.sh is worth calling out specifically because the plans are sized well for what OpenClaw needs. Most generic VPS providers either go too small (not enough RAM for the full agent stack) or push you toward plans that are overkill for what is essentially a lightweight Node.js process. Ampere.sh’s ARM-based servers are also efficient at sustained workloads, which is exactly what an always-on agent is. If you already have a VPS somewhere else that is sitting idle, you can absolutely run OpenClaw on it. But if you are starting fresh, Ampere.sh is the most straightforward option for this use case.
What happens to my MEMORY.md and SOUL.md files when I move to a VPS? Do I need to recreate everything from scratch?
Not at all. Those files are just files. You copy your entire workspace folder to the VPS using scp or rsync and everything comes with it: SOUL.md, MEMORY.md, AGENTS.md, your daily memory logs, scripts, everything. The VPS just needs OpenClaw installed (npm install -g openclaw) and your workspace pointed at the right directory. Your agent picks up exactly where it left off, with full memory and identity intact. The only thing that does not transfer automatically is any active session state, but that resets naturally on restart anyway.
How does pm2 keep OpenClaw running? Is that the right tool or are there better options?
pm2 is the right tool for this. It is a process manager specifically designed for Node.js apps, handles restarts on crash, restarts on server reboot (with pm2 startup), and gives you a clean way to view logs with pm2 logs. The commands are simple: pm2 start openclaw –name agent to start, pm2 save to persist the process list, pm2 startup to generate the system service command. Alternatives like systemd work too and some people prefer it for more control, but pm2 is faster to set up and more beginner-friendly. For running OpenClaw on a VPS, pm2 is the standard approach.
Once OpenClaw is on a VPS, can I still access it from Telegram and WhatsApp the same way?
Yes, completely the same. The channel connections (Telegram bot, WhatsApp, Discord, etc.) are all configured in openclaw.json and they connect outbound from wherever OpenClaw is running. Whether that is your laptop or a VPS in a data centre makes no difference to Telegram or WhatsApp. The bot token is the same, the phone number link is the same, everything works identically. The only practical change is that your agent is now reachable 24/7 instead of only when your laptop is on. If anything, the channel connections get more reliable since the VPS has a stable network connection rather than your home Wi-Fi.
Is there a risk of the VPS running up a big bill if the agent goes into some kind of loop?
The VPS cost itself is fixed, that part is predictable. The risk you are thinking of is API costs, and yes, a looping agent can burn through tokens quickly. OpenClaw has built-in loop detection that kicks in if a tool keeps getting called without making progress, and you can configure the threshold. The practical safeguards are: set spending limits on your OpenAI or Anthropic account (both support this), use a cheap model for heartbeat tasks so background polling is inexpensive, and restrict heartbeat to active hours so the agent is not polling at 3am. With those in place, runaway costs become very unlikely.
Didn’t realise Ampere.sh had an always-free tier for ARM instances. Just spun one up and OpenClaw is running smooth. Good find.
It’s a genuinely underrated option , ARM compute is efficient and the free tier is plenty for a lightweight OpenClaw setup. Glad it’s running well! One tip: set up a simple systemd service so it auto-restarts after any instance reboots.