Self-Hosted n8n vs n8n Cloud: What It Really Costs, Beyond the Server Bill

Self-hosting n8n costs a small server bill plus your time, while n8n Cloud costs a fixed monthly fee that includes the servers, updates and backups. For a handful of workflows the server is cheap, often the price of a lunch. The real cost of self-hosting is the maintenance hours and the risk if nobody owns them, and that is what this guide prices up.

Plan prices, server prices and licence wording below were checked on the vendors' own pages and docs on 6 November 2026. They change; follow the links before you commit.

What does n8n Cloud cost compared with a self-hosted server?

On n8n's pricing page, hosted plans start at €20 per month for 2,500 executions (Starter) and €50 per month for 10,000 (Pro), shown as annual-billing prices. Enterprise is quoted individually. The page also lists a Business plan at €667 per month for 40,000 executions, described as self-hosted only, and a free Community Edition for self-hosting. So the choice is not simply "cloud costs money, self-hosting is free": the paid self-hosted route exists, and it is priced as a business product.

A VPS to run the free Community Edition falls into a few cost classes. Provider pages checked on the same date showed:

n8n's own docs do not state minimum CPU or RAM figures on the setup pages I read, and they point to their measure performance and manage execution data pages instead. Server size therefore depends on the drivers: how many executions run at once, how large the payloads are, whether the Code node does heavy work, and how long you keep execution history. Start small, watch memory, and resize.

Add to the server line: a domain, a managed database if you do not run Postgres yourself, off-site backup storage, an uptime monitor and email sending. These are individually cheap, and together they can match the server itself.

What work does self-hosting add?

Updates

n8n's Docker documentation updates by pulling a new image, stopping the old container and starting a fresh one (with Compose, pull, down, then up). That is a ten-minute job when it works. The work is in reading release notes first, testing a copy of your busiest workflows, and having a way back if a node behaves differently. Someone has to do this regularly, not once.

Backups

The CLI can export workflows and credentials with --backup, but n8n's docs say that flag covers workflows and credentials only, and point to a separate backup and restore guide for a whole instance. A sound plan backs up the database (Postgres dumps or volume snapshots), the n8n data folder, and the encryption key, to a place that is not the same server. Test a restore once. A backup you have never restored is a hope.

Security

n8n's security overview for self-hosters lists HTTPS, SSO, two-factor authentication, verifying user emails, blocking nodes people should not use, disabling the public API if it is not needed, SSRF protection, hardening the task runners that execute Code node scripts, redacting execution data and rotating encryption keys. A built-in security audit command scans the instance for risky settings. On Cloud, n8n runs the platform; on a self-hosted box all of this is yours, including operating system patches and firewall rules.

Monitoring and pruning

Execution history grows. n8n documents EXECUTIONS_DATA_MAX_AGE for automatic pruning, and its privacy docs note that on self-hosted instances you are responsible for deleting user data. Left alone, a busy instance fills its disk and falls over at the worst moment.

Why does the encryption key matter so much?

n8n encrypts the credentials it stores (API keys, OAuth tokens, database passwords) with a key. If you do not set N8N_ENCRYPTION_KEY, n8n generates a random one on first launch and stores it in its user folder. The Docker docs say to keep that folder on a persistent volume even when you use Postgres, because it holds the key.

The failure story is simple. You rebuild the server, restore the database, forget the key, and every saved credential becomes unreadable. Fix it by setting the key explicitly, saving it in a password manager or secrets store, and including it in your recovery notes. The CLI also offers a decrypted credentials export for moving between installs with different keys, but n8n's docs warn that all sensitive information is then visible in the files, so treat that file like a password list and delete it after use.

When do you need queue mode?

Default n8n runs executions inside the main process. Queue mode splits the work: the main instance accepts triggers, Redis holds the queue, and separate worker processes run the executions. n8n's queue mode docs require Redis and PostgreSQL (SQLite is not supported for a distributed setup), set the same encryption key on every worker, and note that binary data cannot use filesystem storage in queue mode, so you point it at S3-style external storage. Queue mode is included in the Community Edition, while multi-main setups and a workers view are listed as Enterprise features.

You need it when many workflows fire at once and slow each other down, or when a long batch job blocks webhook responses. You do not need it on day one. It is a reason to plan the database properly early, since moving from SQLite to Postgres later is a migration.

What does the licence let you do?

n8n's source is under the Sustainable Use License, which is source-available, not open source in the usual sense. The licence text limits use to your own internal business purposes or non-commercial or personal use, and allows distributing the software only free of charge for non-commercial purposes. Files with .ee in their name belong to the separate Enterprise licence.

n8n's licence FAQ interprets this for common cases. It says consulting and automation services are explicitly permitted, and that you can build and run workflows for clients on your own instance as long as clients receive outputs and do not access n8n itself. It lists as not allowed: hosting n8n as a service where clients build workflows, white-labelling n8n, and letting outside users configure workflows through a custom interface. If your plan sits anywhere near those lines, write to the contact address in the FAQ (license@n8n.io) and keep the reply. This is general information, not legal advice.

What are the GDPR and data location implications?

On hosted plans, n8n's pricing FAQ says data is stored in the EU on servers in Frankfurt, Germany, and its privacy docs say that for Cloud n8n acts as both controller and processor, with a data processing agreement that includes Standard Contractual Clauses. Its sub-processor page (last updated 22 September 2026 when I read it) lists Microsoft Azure in the EU, plus AI providers and support affiliates in the US and UK. Anything you send to an AI node reaches that provider as well.

When you self-host, n8n's privacy docs say it is neither controller nor processor, because it does not manage your data. That moves responsibility to you and your server provider, so choose a data centre region on purpose, sign your host's data processing terms, and decide whether to turn off n8n's usage telemetry (the docs name the setting N8N_DIAGNOSTICS_ENABLED). Where the data physically sits is a choice you make, not a promise you inherit. For the broader picture of alternatives by region, see our guide to GDPR-friendly Zapier alternatives. None of this is legal advice; confirm with a qualified adviser.

Who should self-host, and who should not?

• Self-host if you have someone comfortable with Docker and Postgres who will own updates; you need data on a specific server or region; your volume is high enough that per-execution tiers would cost more; or you want the workflow code inside your own network.
• Use Cloud if nobody on the team wants to be on call for a server; your volume fits a Starter or Pro tier; you prefer a vendor-managed DPA and Frankfurt hosting; or the workflows are business critical and you cannot tolerate a missed update window.
• Start on Cloud, move later if you are unsure. Workflows export as JSON, though credentials need re-linking or the same encryption key.

The honest tie-breaker is staffing, not price. If you compare the tools first, our n8n vs Make pricing breakdown shows how usage is billed. If you want the setup, hardening and handover done for you, the n8n workflow automation service covers both hosting routes.

Frequently asked questions

Is self-hosted n8n really free?

The Community Edition costs no licence fee, but you pay for the server, storage, backups and the time to maintain it. Some features, such as SSO, Git version control and environments, are listed as paid.

What happens if I lose the n8n encryption key?

Stored credentials become unreadable and you would have to re-enter them. n8n generates a key automatically if you do not set one, so store it separately from the server.

Can an agency host n8n for its clients?

Per n8n's licence FAQ, running workflows for clients on your own instance is allowed if clients do not access n8n, while hosting n8n as a service for clients to build workflows is not. Ask n8n about your exact setup.

Is n8n Cloud GDPR compliant?

n8n states Frankfurt storage for hosted plans and includes a DPA with Standard Contractual Clauses. Whether your use meets GDPR depends on what data you process and your own obligations, so take advice.

Do I need queue mode?

Only when one process cannot keep up, for example many simultaneous webhooks or long batch jobs. It requires Redis and Postgres plus extra workers, so plan for it before volume forces it.