Use n8n's cloud version if you want automation running quickly and nobody on your team wants to look after a server; self-host n8n if you need your data to stay on infrastructure you control, have high or unusual volumes, or need to reach systems inside your own network, and you have someone (in-house or a partner) who will own updates, backups and security. For most small businesses starting out, cloud is the safer default; self-hosting becomes worth it when control or scale demands it.
The mistake to avoid is self-hosting because it looks free. The software may cost nothing to run under its licence, but the server, the maintenance and the risk of getting security wrong are real costs. This post compares the two options on control, data residency, cost, maintenance, backups, scaling and security, with a decision table at the end. Plan features, limits and prices change, so check n8n's current pricing and licence terms before deciding.
What's actually different between n8n cloud and self-hosted?
The workflow editor and most nodes are the same. What differs is who runs the infrastructure:
- n8n cloud: n8n operates the servers, applies updates, keeps it available and handles infrastructure security. You sign in and build. Plans set limits on things like executions and features.
- Self-hosted: you install n8n on a server you control, such as a virtual machine, a container platform or your own hardware. You decide where it runs, how big it is and who can reach it. You're also responsible for everything that keeps it running.
Some features may be tied to specific paid plans or licences regardless of where you host. Check the current feature comparison rather than assuming self-hosting unlocks everything. If you're new to the tool, our guide to what n8n is covers the basics.
How do the two options compare?
| Factor | n8n cloud | Self-hosted n8n |
|---|---|---|
| Time to start | Minutes | Hours to days, depending on setup |
| Updates | Handled for you | Your job, including testing before upgrading |
| Data location | n8n's chosen hosting regions | Wherever you put it |
| Access to internal systems | Needs those systems exposed or a secure connection | Can sit inside your network |
| Cost shape | Subscription by plan and usage | Server cost plus people time |
| Scaling | Move up a plan | Add resources and configure queue workers yourself |
| Backups | Platform-level, check what's covered | Entirely yours |
| Security | Shared: n8n runs infrastructure, you manage users and credentials | Yours end to end |
| Custom setup | Limited to what the plan allows | Full control over configuration and environment |
When does data residency push you to self-host?
n8n doesn't just pass data along; execution logs can store the data that went through each step, which may include customer details, invoices or chat messages. If your contracts, sector rules or privacy obligations require that data to stay in a particular country or inside your own infrastructure, self-hosting gives you that control.
Obligations under laws like GDPR, India's DPDP Act or sector rules in healthcare and finance depend on your jurisdiction and situation, so check current rules or ask counsel rather than assuming one option is compliant. Whichever you pick, you can reduce exposure by limiting what execution logs store and how long they're kept. Our guide to AI data privacy and security covers the wider questions, including what happens to data you send to LLM providers, which applies either way.
What does each option really cost?
We won't quote prices, because plans and pricing change. Check n8n's current pricing. Compare the shape of the costs:
Cloud:
- A subscription tied to plan level and usage (typically execution volume and features).
- Little or no infrastructure work.
- Costs rise with volume, which matters if you run many high-frequency workflows.
Self-hosted:
- Server or container hosting, a database, storage for backups, and possibly a domain and monitoring tools.
- People time: setup, updates, monitoring, troubleshooting and security reviews. This is usually the biggest cost and the one most often left out.
- Possibly a paid licence for some features or commercial uses; read the current licence terms.
For illustration only: a business running a handful of workflows a few hundred times a day will often find cloud cheaper once staff time is counted. A business running dozens of workflows processing tens of thousands of events a day may find self-hosting cheaper, if someone competent maintains it. Our post on the total cost of ownership of AI systems has a framework for this comparison.
What maintenance does self-hosting involve?
If you self-host, someone needs to own these jobs on a schedule:
- Updates. n8n releases updates regularly, including security fixes. Read release notes, test in a staging copy, then upgrade production.
- Backups. The database holds your workflows, credentials (encrypted) and execution history. Back it up automatically, keep the encryption key safe and separate, and test a restore at least once.
- Monitoring. Is the instance up? Are workflows failing? Is the disk filling with execution logs?
- Log pruning. Execution data grows fast. Configure retention so old executions are deleted.
- Scaling. For heavier loads, n8n supports a queue-based setup with separate worker processes. Check the current docs; it adds moving parts.
- Security. Covered below, and the most important item on the list.
If nobody will do these, cloud is the better choice, regardless of the cost comparison.
What are your security responsibilities?
n8n holds credentials to your CRM, email, accounting and messaging tools. A compromised instance is a compromised business. On cloud, n8n secures the infrastructure and you manage users, permissions and credentials. On self-hosted, it's all yours:
- Never expose the editor openly. Put it behind authentication, ideally single sign-on or a VPN, and use HTTPS.
- Keep it updated. Unpatched automation servers are an attractive target.
- Protect the encryption key used for stored credentials; losing it means losing access to them, leaking it means exposing them.
- Limit who can edit workflows, because anyone who can edit a workflow can use its credentials.
- Use least-privilege API keys for each connected system.
- Restrict outbound and inbound network access to what workflows need.
- Log admin activity and review it.
How do you decide?
| If this describes you | Lean toward |
|---|---|
| Just starting, few workflows, no one to run servers | Cloud |
| Data must stay in a specific country or inside your network | Self-hosted |
| Need to reach internal databases or on-premise systems | Self-hosted (or cloud plus a secure tunnel) |
| High volumes where plan limits get expensive | Self-hosted, if you can maintain it |
| Want local AI models for sensitive data | Self-hosted on suitable hardware |
| Small team, no IT support, uptime matters | Cloud |
| You have DevOps skills or a partner on retainer | Either; pick on data and cost |
A practical path many businesses take: start on cloud to prove the workflows are worth having, then move to self-hosted if data, volume or integration needs justify it. Workflows can be exported and imported, though credentials need to be recreated and some settings may differ. Plan the migration rather than assuming it's one click.
What precautions apply either way?
- Error alerts on every important workflow, so failures don't go unnoticed. See making n8n workflows production-ready.
- Version control: export workflows regularly so you can roll back a bad change.
- AI cost limits: set spending caps with your LLM provider; hosting choice doesn't protect you from a looping agent.
- Human checkpoints for workflows that send money, messages or change important records. Our guide to human-in-the-loop AI agents covers where they belong.
- Named ownership of every workflow and of the instance itself.
Frequently Asked Questions
Is self-hosted n8n free?
The software can be self-hosted without a subscription under n8n's fair-code licence, with conditions on some commercial uses and some features on paid plans. You still pay for hosting and the time to maintain it. Check the current licence and pricing for your situation.
Is self-hosting n8n more secure than the cloud version?
Not automatically. Self-hosting gives you control, but also full responsibility for updates, access control and backups. A well-run cloud instance is usually more secure than a neglected self-hosted one.
Can I move from n8n cloud to self-hosted later?
Yes, workflows can be exported and imported. Credentials generally need to be recreated, and you should test every workflow after moving. Plan the switch as a small project, not a click.
What server do I need to self-host n8n?
It depends on how many workflows run and how heavy they are. Small setups run on modest virtual machines; high-volume ones need a database, queue mode and worker processes. Check n8n's current hosting documentation for requirements.
Can self-hosted n8n use local AI models?
Yes, if you run a local model server on suitable hardware and connect it. This keeps prompts and data on your infrastructure, at the cost of managing the model and often lower quality than leading hosted models. Test on your own tasks.
Not sure which way to run n8n?
If you're weighing cloud against self-hosting, or your self-hosted instance has become a worry, tell us about your workflows and your data requirements. We'll recommend a setup you can actually maintain. See our AI automation service.