.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
If you want to run OpenClaw on a proper server instead of a personal machine, you can now do that on LumaDock with a ready-to-deploy VPS template. Pick a plan, launch a VPS instance, select the OpenClaw (Moltbot) template, and start from a clean Ubuntu 24.04 base with full root access.
OpenClaw is one of those projects that people try locally first, then quickly realize it belongs on a cloud server. It runs continuously, stores state, and ends up connected to real services. A VPS is simply the right shape for that.
What is OpenClaw (Moltbot/Clawdbot)?
OpenClaw is a self-hosted AI assistant, sometimes described as an AI agent, that you run on your own server. It can keep memory across conversations, run in the background, and interact with systems you connect through integrations. The goal is not “chat for fun”. The goal is an assistant that stays available and can support real tasks over time.
The same project was previously known as Clawdbot/Moltbot. You will still find older setup notes and repositories using the previous names, but OpenClaw is the current name and the template on LumaDock includes both so it’s easy to recognize.
Depending on how you configure it, OpenClaw can plug into chat platforms and external services. Some users run it for simple messaging. Others treat it more like a server-side assistant that triggers workflows, maintains state, and coordinates multiple tools.
Why a VPS is a better home for OpenClaw
OpenClaw is designed as a service. That changes the hosting requirements. It needs stable uptime, persistent storage and a network identity that does not change every time you restart a router.
Hosting OpenClaw on a VPS gives you:
- Always-on runtime so the assistant stays reachable and keeps running tasks
- Persistent disk for memory, logs, config, and data you store alongside it
- Stable networking for integrations, webhooks, callbacks, and API endpoints
- KVM Isolation so it is not tied to your laptop and not mixed with personal apps
And it’s easier to operate. You can monitor one server, back it up, snapshot it before changes, and scale it when you outgrow the initial plan.
How the LumaDock OpenClaw template works
The template is available during VPS deployment. You create a server instance in the LumaDock panel, choose a plan, then select OpenClaw (Moltbot) as the template. It comes with OpenClaw already installed on Ubuntu 24.04. Once your VPS is deployed, the service is ready to run without requiring a manual install step.
This is a native installation, not a Docker container. OpenClaw runs directly on the system as a service, which keeps the setup predictable, easy to inspect, and aligned with how most users operate long-running assistants.
This approach keeps the VPS flexible and avoids the trap of “managed magic” that makes self-hosted projects harder to debug.
Why self-host your AI assistant on LumaDock
A long-running bot is sensitive to small hosting problems. Slow disk writes show up as lag. Network limits show up as broken integrations. Unpredictable performance turns into mystery crashes and wasted time.
LumaDock is built to avoid those headaches. Here are the pieces that matter most for OpenClaw hosting.
- NVMe SSD storage on AMD EPYC CPUs
Our infrastructure runs on AMD EPYC processors with fast NVMe SSD storage. For a OpenClaw VPS, that means responsive disk access for logs, memory files, and anything else your assistant writes continuously. - Unmetered bandwidth included
OpenClaw setups often expand into integrations that move more data than expected. Unmetered bandwidth keeps your VPS costs predictable while you iterate. - IPv4 and IPv6 included
A stable IP address matters for callbacks and webhooks. Both IPv4 and IPv6 are included by default so you can build integrations without extra steps. - Firewall and DDoS protection
If you expose an endpoint or dashboard, you want baseline protection. Firewall management and always-on DDoS protection are included to help keep services reachable and reduce noise. - Backups and snapshots
OpenClaw evolves quickly once you start using it. Snapshots are useful before upgrades and config changes. Backups provide a longer-term safety net when the assistant starts holding data you care about. - Full root access
A OpenClaw VPS should be yours. You get full root access to install dependencies, adjust limits, run containers, and shape the environment around your use case. - Instant deployment, no setup fees
Launch quickly with no setup fees and a 30-day refund period. You can try your setup risk free, and refund later if it's not for you.
Choosing a VPS plan for OpenClaw
If you are using external AI model APIs, OpenClaw does not need extreme resources to start, but it does require a sensible memory baseline. A minimum of 2 GB RAM is recommended for stable operation, especially once memory, logs and integrations are in use.
Two practical guidelines help:
- Start with a smaller server instance if you are testing or running a light setup
- Choose more RAM if you run OpenClaw alongside other services like databases, queues, or automation tools
If you want a more controlled setup, it is also common to separate services across multiple VPS instances and connect them privately.
