n8n Self-Hosted vs Cloud: The Complete Decision Guide (2026)

The Question Every Serious n8n User Asks

When you move from experimenting with n8n to building production automation systems, the deployment question becomes unavoidable: run n8n yourself on a VPS, or pay for n8n Cloud and let the platform handle infrastructure? Both options work. Both have real trade-offs. The wrong choice costs you either money you did not need to spend, or operational overhead you were not prepared to manage.

This guide is based on PURIST's experience deploying n8n in both configurations across more than 60 client engagements. We will give you the actual numbers, the genuine limitations, and a clear recommendation framework not a vendor comparison that pretends every option is equally valid for every team.

What Self-Hosted n8n Actually Costs

Self-hosted n8n is free to run. The n8n software itself is source-available under the Sustainable Use License, which permits self-hosting for internal business use without a licence fee. What you pay for is the infrastructure to run it.

The minimum viable production setup for self-hosted n8n is a VPS with 2 vCPUs and 4 GB of RAM. At those specs, you can run n8n comfortably for most SMB workloads up to several hundred workflow executions per hour without performance degradation. At Hetzner (our preferred provider for UK and European clients), a CX22 instance with 2 vCPUs, 4 GB RAM, and 40 GB SSD runs at £4.51/month in their Nuremberg data centre. Add a managed PostgreSQL database (required for production n8n the default SQLite is not suitable for concurrent production use) at £8.20/month for a 4 GB Hetzner managed database, and you are at approximately £13/month all-in before backups and monitoring.

For a more comfortable production setup one with headroom for growth and a separate process for the n8n worker we typically spec a CX32 (4 vCPUs, 8 GB RAM) at £9.02/month plus the managed database. Total: approximately £17-18/month. Add daily off-site backups via Hetzner's Snapshot feature (included in the instance cost) and a £5/month uptime monitoring subscription, and you are running a solid production n8n deployment for £22-25/month.

At the top end, a high-throughput deployment handling 5,000+ executions per hour on dedicated infrastructure a CCX23 bare metal cloud instance with 8 vCPUs and 32 GB RAM, a dedicated PostgreSQL cluster, Redis for queue management in worker mode, and a load balancer costs £80-120/month. This is enterprise territory for self-hosted n8n, handling workloads that would cost £500+/month on n8n Cloud's enterprise tier.

The honest cost summary for self-hosted n8n: £15-25/month for standard SMB production use, £40-80/month for high-volume deployments, £80-120/month for large-scale enterprise workloads.

What n8n Cloud Actually Costs

n8n Cloud pricing in 2026 operates on a workflow and execution model. The Starter tier is $20/month (approximately £16/month at current exchange rates) and includes 5 active workflows and 2,500 executions per month. The Pro tier is $50/month for 15 active workflows and 10,000 executions per month. The Business tier at $120/month gives you unlimited workflows and 50,000 executions per month. Enterprise pricing is negotiated and typically starts at $500/month.

The execution limits are the critical number. 10,000 executions per month sounds generous until you run an active CRM automation. A workflow that triggers on every new CRM contact, processes 50 leads per day, seven days a week, consumes 10,500 executions per month on its own before any other workflows run. Businesses with multiple active automations hit the Pro tier ceiling quickly and find themselves on Business at $120/month (£95/month).

At the Business tier, n8n Cloud costs approximately 4-6 times more than a comparable self-hosted setup. At the Pro tier, the gap is smaller roughly 2-3 times the self-hosted cost but the execution limits are tighter.

Infrastructure Responsibility: What You Actually Take On

The cost difference between self-hosted and cloud only tells half the story. The other half is operational responsibility.

With n8n Cloud, you have zero infrastructure responsibility. n8n handles server provisioning, operating system updates, n8n version upgrades, database management, backup procedures, SSL certificates, and uptime monitoring. When something goes wrong at the infrastructure level, n8n's team resolves it. Their SLA at the Pro tier guarantees 99.9% uptime roughly 8.7 hours of downtime per year. You log in, build workflows, and the platform exists.

With self-hosted n8n, you own the infrastructure. That means:

Operating System Maintenance

You are responsible for applying security patches to the host OS. A Debian or Ubuntu VPS needs regular apt updates. Miss a critical kernel patch and you are running vulnerable infrastructure. In practice, most teams set up unattended-upgrades for security patches and handle major version upgrades quarterly. If you are not comfortable managing a Linux server, this is not a theoretical concern it is a concrete skills requirement.

n8n Version Management

n8n releases updates frequently. Self-hosted deployments need manual upgrades pulling the latest Docker image, testing in a staging environment, and deploying. Missing major version upgrades means running outdated software with known security vulnerabilities. Staying current requires a process: a staging environment, a testing protocol, and a deployment procedure. This is not complex, but it takes time typically 2-4 hours per quarter for a diligent team.

Database Management

Production n8n requires PostgreSQL. You need to manage connection pooling, query performance, backup verification, and occasional index maintenance. A managed database service (Hetzner, DigitalOcean, AWS RDS) removes most of this burden at a cost of £8-20/month. The cost is worth it for anyone who is not a DBA.

Incident Response

When your self-hosted n8n goes down at 2am, you are the incident response team. If a disk fills up, a database connection pool exhausts, or a memory leak takes down the process, you need an alert, a runbook, and the availability to respond. n8n Cloud handles this for you. Self-hosted requires you to build it yourself monitoring, alerting, and on-call coverage.

Security and Data Privacy Comparison

This is the dimension where self-hosted wins most clearly, and it is the primary driver behind our recommendation for clients in regulated industries.

Data Residency

With n8n Cloud, your workflow execution data including the payloads that flow through your automations is stored on n8n's infrastructure. As of 2026, n8n Cloud stores data in EU data centres (Frankfurt) for European customers, with US storage for US-based accounts. If you are processing UK patient data under UK GDPR, EU personal data under GDPR Article 44+ transfer rules, or commercially sensitive data that must remain within a specific jurisdiction, you need to verify that n8n's data residency guarantees match your compliance requirements.

With self-hosted n8n on a Hetzner VPS in Nuremberg or a AWS eu-west-2 (London) instance, your data never leaves the infrastructure you control and the jurisdiction you specify. This is a categorical difference, not a marginal one. For PURIST clients in healthcare, legal services, and financial services, self-hosted is the default recommendation specifically because of data residency control.

Credential Storage

Both n8n Cloud and self-hosted n8n encrypt stored credentials at rest. The difference is where the encryption keys live. On n8n Cloud, n8n manages the encryption keys. On self-hosted, you control them. For most businesses, n8n's key management is adequate. For businesses where key custody is a regulatory requirement or a board-level security policy, self-hosted is the only compliant option.

Network Isolation

Self-hosted n8n can be deployed inside a private VPC with no public internet exposure, reachable only via VPN or private network peering. This eliminates an entire category of attack surface. n8n Cloud is inherently a public SaaS service your workflows are accessible via the internet (with authentication), but the infrastructure perimeter is managed by n8n, not by you.

Setup Complexity: Honest Assessment

Setting up self-hosted n8n is not difficult for someone comfortable with Linux and Docker. It is a genuine barrier for someone who is not.

The basic self-hosted deployment n8n running in Docker Compose on a VPS, with PostgreSQL, a reverse proxy (Nginx or Caddy), and SSL takes an experienced DevOps engineer 2-4 hours. It takes a developer who knows Linux but has not done much infrastructure work 4-8 hours. It takes a non-technical business owner significantly longer, with a meaningful chance of misconfiguration.

The production-ready deployment adding worker mode for queue management, Redis for job queuing, automated backups, monitoring, log aggregation, and a staging environment takes an experienced engineer a full day. This is not a weekend project for a non-technical team.

n8n Cloud setup takes 10 minutes. You create an account, choose a plan, and you are running workflows. For teams without dedicated technical staff, this difference is decisive.

Scaling: Where Each Option Hits Its Limits

Scaling Self-Hosted n8n

Self-hosted n8n in queue mode (using Redis and n8n workers) scales horizontally. You can run one main n8n process handling the UI and webhook ingestion, and multiple worker processes executing workflows in parallel. Adding a worker is adding a Docker container it scales linearly with your infrastructure budget. PURIST has deployed self-hosted n8n handling 50,000+ executions per day on a £80/month infrastructure stack. That same volume on n8n Cloud Enterprise would cost £500+/month.

The practical ceiling for self-hosted scaling is your team's infrastructure competence. If you can manage Kubernetes or Docker Swarm, self-hosted n8n scales as far as you need. If you cannot, you will hit complexity limits before you hit performance limits.

Scaling n8n Cloud

n8n Cloud scales transparently up to the execution limits of your tier. Above those limits, you pay more. The Business tier at $120/month gives you 50,000 executions per month enough for most SMB use cases. Beyond that, you are in Enterprise territory. n8n Cloud does not require you to manage infrastructure scaling the platform handles it but you pay a significant premium for that abstraction at high volumes.

Feature Parity: What Cloud Gets That Self-Hosted Does Not (and Vice Versa)

In 2026, n8n Cloud and self-hosted n8n are functionally near-identical in terms of workflow capability. All nodes are available in both. The execution engine is the same. The credential management, workflow templates, and expression syntax are identical.

n8n Cloud has a few exclusive features: managed AI features (n8n's hosted AI assistant for workflow building), automatic version upgrades without downtime, and built-in usage analytics dashboards. These are convenience features, not capability features.

Self-hosted n8n has exclusive capabilities: full filesystem access for workflows that read and write local files, the ability to install custom community nodes without waiting for Cloud platform support, and the ability to run workflows that access resources on your private network (internal APIs, databases, on-premise systems) without exposing them to the internet. That last point is significant for businesses with on-premise ERP systems, legacy databases, or internal tools that have no cloud connectivity.

The PURIST Recommendation Framework

After deploying both options across dozens of client environments, our recommendation is straightforward:

**Choose n8n Cloud if:** You have no dedicated technical staff capable of managing Linux infrastructure. Your workflows do not process regulated personal data requiring specific data residency. Your execution volume falls under 50,000/month. You want zero infrastructure overhead and are willing to pay the premium for that abstraction. You are just starting with n8n and want to learn the tool before committing to infrastructure management.

**Choose self-hosted n8n if:** You process personal data requiring UK or EU data residency controls healthcare, legal, financial services. Your execution volume exceeds 10,000/month (the self-hosted economics become strongly favourable). Your workflows need to access private network resources (internal databases, on-premise APIs) that are not internet-accessible. You have a developer or DevOps resource who can manage the infrastructure. You want to install community nodes not available on Cloud, or need custom node development.

**The hybrid approach:** Several PURIST clients run both. n8n Cloud handles their external-facing workflows lead capture, CRM sync, email automation where the convenience is worth the cost and the data is not regulated. Self-hosted n8n handles internal workflows involving patient data, financial records, or private network resources. This separation of concerns gives them Cloud's convenience where it fits and self-hosted's data control where it matters.

Migration Considerations

If you start on n8n Cloud and later decide to migrate to self-hosted, n8n provides a workflow export function. Your workflow definitions (JSON) export cleanly. Credentials do not export for security reasons you will need to re-enter API keys and authentication details in the self-hosted environment. Execution history does not migrate. For most teams, this migration takes half a day.

Migrating from self-hosted to Cloud follows the same process. The practical risk in either direction is low, which means there is no penalty for starting on Cloud and switching later as your requirements evolve.

Frequently Asked Questions

Is n8n self-hosted truly free?

The n8n software is free for self-hosted internal business use under the Sustainable Use License. You pay only for your infrastructure VPS, database, and any third-party services you connect to. The license restriction is on using self-hosted n8n to build a commercial product that competes with n8n Cloud. For internal business automation, there is no license fee.

What is the minimum server spec for production self-hosted n8n?

Minimum viable production: 2 vCPUs, 4 GB RAM, 40 GB SSD, with a separate managed PostgreSQL database. Do not run n8n in production with the default SQLite backend it is not designed for concurrent production use and will degrade under load. The 4 GB RAM threshold is important: below it, large workflow executions that process significant payloads can cause OOM (out of memory) errors.

Can I run self-hosted n8n on a Raspberry Pi or home server?

Technically yes, but we do not recommend it for production workflows. Home internet connections lack the uptime reliability of data centre infrastructure, power interruptions kill the process, and home IP addresses are frequently blocked by corporate security systems that your integrations need to reach. For learning and development, a home server is fine. For production, use a VPS.

How does n8n Cloud handle GDPR?

n8n offers a Data Processing Agreement (DPA) for Cloud customers on Pro and Business tiers. Their EU-hosted infrastructure processes data in Frankfurt. For UK GDPR purposes post-Brexit, the UK GDPR applies separately from EU GDPR, and n8n's DPA covers both. However, if your compliance requirements specify UK data residency specifically data must not leave UK jurisdiction n8n Cloud's EU-Frankfurt infrastructure does not satisfy that requirement. Self-hosted on UK infrastructure (AWS eu-west-2, Hetzner Helsinki is not UK use AWS London or a UK-based VPS provider) is the compliant option.