“Source Is Open, Competition Is Not”
n8n is a workflow automation platform. It lets you visually design trigger, conditional branching, API calls, and data transformation pipelines. It’s similar to Zapier or Make (formerly Integromat), but with one critical difference: the source code is publicly available.
Yet n8n does not call itself “open source.” It calls itself Fair-code.
| Aspect | MIT / Apache | Fair-code (n8n) | Proprietary |
|---|---|---|---|
| View source code | Yes | Yes | No |
| Modify and self-use | Yes | Yes | No |
| Commercial use (internal) | Yes | Yes | License required |
| Resale / hosted service | Yes | No | No |
| Fork into competing product | Yes | No | No |
| OSI-certified open source | Yes | No | No |
The core principle: “Free for internal use, prohibited to resell.”
Why n8n Didn’t Choose MIT
The AWS Problem
The biggest risk of an MIT license is that a cloud provider takes the code and sells it as a managed service.
graph LR
A["OSS company\ndevelops code"] --> B["Published under MIT"]
B --> C["AWS/Azure/GCP launches\nManaged Service"]
C --> D["Sells same product\nat larger scale than creator"]
D --> E["Creator's revenue\ndeclines"]
style C fill:#ffebee,stroke:#f44336
style D fill:#ffebee,stroke:#f44336
style E fill:#ffebee,stroke:#f44336
This is not hypothetical. It has happened multiple times:
| Company | Product | AWS Response | Outcome |
|---|---|---|---|
| Elastic | Elasticsearch | Amazon OpenSearch Service | Elastic changed license (Apache to SSPL) |
| MongoDB | MongoDB | Amazon DocumentDB | MongoDB changed license (AGPL to SSPL) |
| Redis | Redis | Amazon ElastiCache | Redis changed license (BSD to RSAL) |
When AWS forks an open source project and creates a managed service, the original creator has to compete against their own code. Competing against AWS’s infrastructure, sales force, and existing customer base is virtually impossible.
n8n sidestepped this problem from day one. By starting with Fair-code instead of MIT, they blocked the “fork n8n and sell Managed n8n” scenario at the license level.
The Sustainable Use License
n8n uses the Sustainable Use License. Its key clause:
“You may use the software for your own internal business purposes. You may NOT provide the software to third parties as a hosted or managed service.”
In plain terms: using it inside your own company is OK; selling it as a hosting service to others is not.
For the vast majority of users, this means fully free:
- Internal company automation — free
- Personal projects — free
- Community contributions — welcome
- Building a SaaS on top of n8n and selling it — requires a separate agreement
License Change History: Elastic, MongoDB, HashiCorp
Starting with Fair-code from the beginning, as n8n did, is rare. Most companies begin with MIT, Apache, or AGPL, encounter the AWS problem, and then change their license. These transitions invariably spark serious community conflict.
Elastic: SSPL, Then Back to AGPL
| When | License | Why |
|---|---|---|
| ~2018 | Apache 2.0 | Standard open source |
| 2021-01 | SSPL + Elastic License | Responding to AWS OpenSearch |
| 2024-08 | Added AGPL option | Easing community backlash |
How it unfolded:
- AWS forked Elasticsearch and launched Amazon OpenSearch Service
- Elastic changed from Apache 2.0 to SSPL (Server Side Public License)
- SSPL says: “If you offer this software as a service, you must open-source your entire service stack” — effectively blocking cloud resale
- Community backlash: “It’s no longer open source”
- AWS continued maintaining OpenSearch as an independent project
- In 2024, Elastic offered AGPL as an additional option as a compromise
graph TB
A["Apache 2.0\n(Fully open)"] -->|"AWS fork"| B["SSPL\n(Full stack disclosure\nrequired for service use)"]
B -->|"Community backlash"| C["AGPL added\n(Compromise)"]
D["Community trust"] -->|"SSPL switch"| E["Trust eroded"]
E -->|"AGPL addition"| F["Partially restored"]
style B fill:#ffebee,stroke:#f44336
style E fill:#ffebee,stroke:#f44336
MongoDB: AGPL to SSPL
| When | License | Why |
|---|---|---|
| ~2018 | AGPL v3 | Copyleft (source disclosure when providing as a service) |
| 2018-10 | SSPL | Strengthened defense against AWS DocumentDB |
MongoDB went a step beyond AGPL by creating SSPL. While AGPL required: “If you modify this software, disclose your source,” SSPL demands: “If you offer this software as a service, disclose the source of your entire service stack.”
AWS responded by building DocumentDB, an API-compatible service developed independently rather than using MongoDB directly. The result: SSPL successfully defended against forks, but it couldn’t defend against clones.
HashiCorp: MPL to BSL
| When | License | Why |
|---|---|---|
| ~2023 | MPL 2.0 | Mozilla Public License |
| 2023-08 | BSL 1.1 | Defense against cloud resale |
HashiCorp (Terraform, Vault, Consul) switched to BSL (Business Source License) in 2023. BSL’s key characteristics:
- Source code: Publicly available (readable, modifiable)
- Non-commercial use: Free
- Competing service provision: Prohibited
- Automatic conversion after 4 years: BSL becomes Apache 2.0 (eventually becomes fully open source)
The community response to HashiCorp’s BSL transition gave birth to OpenTofu — an open source fork of Terraform, hosted by the Linux Foundation, maintaining the MPL 2.0 license.
The License Spectrum
graph LR
subgraph OPEN["Open Source (OSI-certified)"]
MIT["MIT\n(Full freedom)"]
APACHE["Apache 2.0\n(Patent protection)"]
AGPL["AGPL\n(Service modifications\nmust be disclosed)"]
end
subgraph SOURCE["Source Available (Not OSI-certified)"]
BSL["BSL\n(Competing services prohibited,\nopens after 4 years)"]
SSPL["SSPL\n(Entire service stack\nmust be disclosed)"]
FAIR["Fair-code\n(Resale prohibited)"]
ELASTIC["Elastic License\n(Competing services\nprohibited)"]
end
subgraph CLOSED["Proprietary"]
PROP["Source closed\n(License purchase required)"]
end
MIT --> APACHE --> AGPL --> BSL --> SSPL --> PROP
style OPEN fill:#e8f5e9,stroke:#4caf50
style SOURCE fill:#fff3e0,stroke:#ff9800
style CLOSED fill:#ffebee,stroke:#f44336
| License | Internal Use | Modification | Resale / Hosting | Fork & Compete | Notable Users |
|---|---|---|---|---|---|
| MIT | Yes | Yes | Yes | Yes | LangChain, MMU |
| Apache 2.0 | Yes | Yes | Yes | Yes | HuggingFace, Qdrant |
| AGPL v3 | Yes | Yes (disclose) | Yes (disclose) | Yes (disclose) | Grafana, MinIO |
| BSL 1.1 | Yes | Yes | No (OK after 4 years) | No (OK after 4 years) | HashiCorp, MariaDB |
| SSPL | Yes | Yes | No | No | MongoDB, Elastic |
| Fair-code | Yes | Yes | No | No | n8n |
| Elastic License | Yes | Yes | No | No | Elastic (current) |
Sentry (BSL license) and PostHog (MIT + paid cloud) follow similar patterns, demonstrating that Fair-code is not an isolated experiment but part of a broader industry trend.
n8n’s Business Results
Let’s verify whether the Fair-code strategy is actually working, with numbers.
| Metric | Figure |
|---|---|
| ARR | $50M+ (as of 2025; estimated — private company, no official disclosure) |
| GitHub Stars | 55K+ |
| Self-hosted instances | Tens of thousands (estimated from Docker pulls) |
| Cloud users | Thousands of organizations (estimated) |
| Funding | Series B, ~$30M+ |
| Team size | ~100+ |
| Integrations | 400+ services |
Pricing Structure
| Plan | Monthly Cost | Executions | Key Feature |
|---|---|---|---|
| Community (self-hosted) | $0 | Unlimited | Fair-code license |
| Starter | $20/mo | 2,500 | Cloud hosted |
| Pro | $50/mo | 10,000 | Advanced features |
| Enterprise | Custom | Unlimited | SSO, audit logs, SLA |
n8n’s $50M+ ARR (estimated) validates the Fair-code model. Under MIT, there’s a high probability AWS would have launched “Amazon Workflow Automation (powered by n8n).” Fair-code prevented this scenario from day one.
Pros and Cons of the n8n Model
Pros
- AWS defense: Resale is blocked at the license level, eliminating the need for a dramatic license change later
- Self-host freedom: Internal use is completely free, minimizing adoption barriers
- Community preservation: Source code is available, so contributions, bug reports, and plugin development remain possible
- Legal clarity: The terms are simple — “use it yourself, OK; sell it to others, no”
Cons
- Not OSI-certified: Cannot be called “open source” — some developers and organizations reject non-OSI licenses on principle
- Smaller community: May attract fewer contributors than MIT projects (no resale means reduced corporate contribution incentive)
- No forking: The community cannot create an alternative like OpenTofu even if dissatisfied
- Legal gray areas: The boundary between “internal use” and “service provision” can be ambiguous. Per n8n’s license FAQ, installing for clients and charging management fees is permitted, but reselling n8n itself as a service (SaaS-style) to third parties is prohibited
Implications for Solo Builders
License Choice Is a Business Model Choice
A license is not a legal appendage. The license determines the boundaries of your viable business models.
| Business Model | MIT | Fair-code |
|---|---|---|
| Free CLI + paid content | Yes | Yes |
| OSS to managed SaaS | Yes (AWS risk) | Yes (defended) |
| Community to enterprise | Yes (active contributions) | Partial (limited contributions) |
| Fork defense | No | Yes |
Why MMU Chose MIT (Revisited)
MMU chose MIT. Even though n8n’s Fair-code offers more protection, here’s why MIT was the right call:
- Scale: MMU is a CLI checklist. The probability of AWS creating a “Managed Checklist Service” is effectively zero
- Adoption first: The first challenge for a solo project is “getting people to use it.” MIT minimizes adoption friction
- Revenue separation: MMU’s revenue comes from content (Playbook Pack), not the CLI code. You can fork the code, but you can’t fork the content
- Community contributions: Maintaining 534 items alone is difficult. MIT maximizes contribution incentives
Fair-code is rational “when your code itself could become your competitor.” Since MMU generates revenue from content rather than code, a code fork isn’t a threat.
Summary
| Key Point | Details |
|---|---|
| What is Fair-code | Source available + resale prohibited. Not OSI-certified, so not “open source” |
| Why n8n chose it | To solve the AWS defense problem from the start |
| License change dramas | Elastic, MongoDB, and HashiCorp all changed after the fact, causing community conflict |
| Validation | n8n’s $50M+ ARR (estimated) proves large-scale business is possible under Fair-code |
| Solo builder takeaway | Use Fair-code when code itself is a competitive threat; use MIT when content is the revenue source |
This post concludes the OSS monetization analysis series (G4a-G4d). The next post covers the practical license selection guide — MIT vs Apache vs Fair-code vs additional clauses, with a decision framework by project scale.
Related Posts
Open Source License Selection Guide — MIT vs Apache vs Fair-code vs Additional Clauses
Comparison of open source licenses (MIT, Apache 2.0, BSL, etc.) and a decision framework based on project scale and revenue model, including MMU's choice of MIT.
AI Search Engine Comparison: ChatGPT Search, Perplexity, AI Overviews
Comparative analysis of ChatGPT Search, Perplexity, and Google AI Overviews on accuracy, speed, and cost, with optimal engine recommendations for each query type.
GEO Definition and Structural Differences from SEO
Definition of GEO and its six structural differences from SEO. Proposes a new framework for brand visibility in an environment where AI search provides answers instead of link clicks.