n8n is not dead. Let me get that out of the way immediately.
But n8n is also not the thing that most people think it is. And the gap between perception and reality is widening fast enough that we need to have an honest conversation about it.
n8n just closed a $180 million Series C at a $2.5 billion valuation. Nvidia participated. The press coverage was glowing. The community celebrated. And I sat there looking at the numbers thinking: this does not add up the way people want it to.
I have been using n8n in production since 2023. We run it at CODERCOPS for internal automation — Slack notifications, CRM syncing, content pipeline triggers. It works. But "it works" and "it is the future of AI-powered automation" are two very different claims, and the second one is what n8n is selling right now.
Let me break down what is actually happening.
The Numbers Behind the Headline
Here is what n8n announced with their Series C:
| Metric | Number |
|---|---|
| Funding raised | $180M Series C |
| Total funding | $240M |
| Valuation | $2.5B |
| ARR | ~$40M+ |
| Users | 200,000+ |
| Self-hosted instances | Hundreds of thousands |
| Revenue growth | 10x year-over-year |
| User growth | 6x year-over-year |
Those growth numbers look incredible. 10x revenue growth. 6x user growth. But here is the number that should make you pause:
$2.5 billion divided by $40 million ARR = a 58x revenue multiple.
For context, Zapier — the category leader with significantly more revenue, more enterprise customers, and a proven business model — trades at a fraction of that multiple. Make (formerly Integromat) is even lower. A 58x multiple means investors are not paying for what n8n is today. They are paying for what they hope it becomes.
And what they hope it becomes is an AI agent orchestration platform. That is the bet. The question is whether that bet makes sense.
What n8n Actually Does Well
Before I get into the problems, let me be fair about what n8n gets right.
Self-hosting is genuinely good. If you need workflow automation that runs on your infrastructure, n8n is the best option available. Period. Zapier cannot do this. Make cannot do this. For companies with strict data residency requirements or those who want to avoid vendor lock-in, n8n self-hosted is compelling.
The workflow builder is powerful. The visual editor for building automation workflows is more flexible than Zapier's. You can write JavaScript or Python in any node, which gives developers escape hatches when the visual interface is not enough. For complex, branching workflows with conditional logic, n8n's approach is superior.
The integration library is extensive. Over 400 integrations. Most of the common ones work well. And the HTTP Request node means you can connect to anything with an API, even if there is no official integration.
Community is strong. 200,000+ users generating templates, sharing workflows, and answering questions. The community forum is active and genuinely helpful. This is n8n's biggest moat.
Open source (sort of). The code is visible on GitHub. You can inspect it, modify it, contribute to it. This matters for trust and transparency.
The Problems Nobody Wants to Discuss
Now let me tell you what I have actually experienced and what the community is talking about when the cameras are off.
Problem 1: The Fair-Code License Is Not Open Source
n8n uses the Sustainable Use License, which they market under the "fair-code" banner. Let me be clear about what this means: you cannot use n8n to compete with n8n.
If you build a product that includes workflow automation features and those features overlap with what n8n offers, you may be in violation of the license. This is not hypothetical — it has caused real confusion among startups trying to embed n8n into their products.
The license also means:
- You cannot offer n8n as a service to third parties
- Commercial redistribution has restrictions
- The "open source" label is misleading — this is source-available, not open source by the OSI definition
For personal projects and internal tooling, this does not matter. For companies building products, it is a real constraint that does not get discussed enough.
Problem 2: Security Vulnerabilities Are a Pattern
In 2024 and 2025, multiple security vulnerabilities were disclosed in n8n, including:
- CVE-2024-28890: A critical vulnerability allowing unauthorized access to credentials
- Multiple SSRF (Server-Side Request Forgery) vulnerabilities
- Authentication bypass issues in certain configurations
- Path traversal vulnerabilities in file handling nodes
For a self-hosted automation tool that often has access to API keys, database credentials, and internal systems, security is not optional. Every vulnerability in n8n is a potential pathway to your entire infrastructure.
The n8n team has been responsive about patching, but the pattern of vulnerabilities suggests that security was not a design priority from the beginning. When you self-host n8n, you are taking on the responsibility of keeping it patched and monitored. Many teams I have spoken with do not realize this.
Problem 3: The AI Agent Story Is Oversold
This is the big one. n8n is positioning itself as the orchestration layer for AI agents. Their Series C pitch was heavily focused on "AI-powered workflow automation." But here is what AI agents in n8n actually look like today:
- You connect an LLM node (usually OpenAI or Anthropic)
- You chain it with other nodes for data retrieval, transformation, and output
- You hope the LLM output is structured enough that downstream nodes can parse it
That is not agent orchestration. That is an API call with extra steps.
Real AI agent orchestration requires:
- Memory management across workflow runs
- Dynamic tool selection based on context
- Error recovery when the LLM produces unexpected output
- Multi-agent coordination with handoffs
- Streaming for long-running agent tasks
n8n has some of these partially implemented, but none of them are production-grade. The LangChain and LangGraph integrations help, but they add complexity without adding reliability.
I have tried building a customer support agent in n8n. It took three times as long as building it directly with the Claude API and a simple Python script. The visual builder actually made the agent logic harder to reason about, not easier.
Problem 4: Performance at Scale Is Concerning
n8n works great for workflows that run a few hundred times a day. When you start scaling to thousands of executions per hour, things get interesting:
- Memory consumption grows linearly with execution history
- The UI becomes slow when viewing workflow execution logs at scale
- Webhook response times degrade under load
- Queue-based execution (using Redis/BullMQ) helps but adds infrastructure complexity
For enterprise-grade automation, you need to think carefully about n8n's infrastructure requirements. It is not a "deploy and forget" solution at scale.
Problem 5: The Billing Model Change
n8n recently shifted from flat-rate pricing to execution-based billing for their cloud product. This means your costs scale with usage, which sounds fair until you realize:
- Retry logic means failed-and-retried workflows count as multiple executions
- Polling triggers consume executions even when nothing happens
- Complex workflows with many nodes consume more executions per run
- Cost becomes unpredictable for bursty workloads
Several community members have reported unexpected cost increases after the billing change. For teams that chose n8n specifically to avoid Zapier's per-task pricing, this feels like a bait and switch.
How n8n Compares in 2026
Let me give you the honest comparison across the automation landscape:
| Feature | n8n | Zapier | Make | Temporal | Custom Code |
|---|---|---|---|---|---|
| Self-hosting | Yes | No | No | Yes | Yes |
| Visual builder | Excellent | Good | Excellent | No | No |
| AI agent support | Basic | Basic | Basic | Excellent | Full control |
| Enterprise security | Improving | Strong | Good | Excellent | Depends |
| Cost at scale | Medium | Expensive | Cheap | Cheap | Cheap |
| Learning curve | Medium | Low | Medium | High | High |
| Reliability at scale | Medium | High | Medium | Excellent | Depends |
| License | Fair-code | Proprietary | Proprietary | MIT | Varies |
If you need visual workflow automation for internal tools — n8n is still a solid choice.
If you need reliable AI agent orchestration — use Temporal, LangGraph, or custom code with the Claude/OpenAI API directly.
If you need maximum simplicity — Zapier is still easier for non-technical teams.
If you need cost-effective high-volume automation — Make (Integromat) often wins on price.
What the $2.5 Billion Valuation Really Means
The valuation tells you one thing: investors believe n8n can become the default platform for AI agent orchestration. Not workflow automation — that market is mature and n8n is not winning it. Agent orchestration is the growth story.
But here is the problem with that thesis:
AI agent frameworks are evolving faster than n8n can integrate them. By the time n8n adds proper support for a framework, two new ones have emerged.
The companies building serious AI agents are not using visual workflow tools. They are using code. Claude Code, Cursor, custom frameworks — the serious builders want control, not drag-and-drop.
The 58x revenue multiple means n8n has almost no margin for error. They need to grow ARR from $40M to $400M+ within a few years to justify that valuation. That requires either enterprise sales (where Temporal and custom solutions dominate) or massive self-serve growth (where the billing changes are creating friction).
Nvidia's investment is a signal, but not the one people think. Nvidia invests in AI infrastructure broadly. They invest in dozens of companies per year. Their participation does not validate n8n's specific approach to AI agents.
Is n8n Dead?
No. n8n is very much alive and growing.
But n8n is at a critical inflection point. The automation market it started in is commoditizing. The AI agent market it wants to enter is dominated by code-first solutions. And the $2.5 billion valuation means the pressure to deliver transformative growth is immense.
Here is what I think will actually happen:
Best case: n8n successfully builds an AI agent orchestration layer that is genuinely better than coding agents from scratch. They capture the market of teams that want agent capabilities without a dedicated ML engineering team. The visual builder becomes the interface for managing agents, not building them. ARR reaches $200M+ by 2028.
Most likely case: n8n remains a strong workflow automation tool. The AI agent features are used by a niche audience. Growth continues but slows. The company either raises another round at a lower multiple or gets acquired by a larger platform (Salesforce, ServiceNow, Atlassian).
Worst case: The AI agent pivot distracts from the core automation product. Enterprise customers choose Temporal. SMB customers stick with Zapier and Make. Self-hosted users never convert to paid. The 58x multiple corrects painfully.
What We Use Instead
At CODERCOPS, here is our current automation stack:
- Simple automations (Slack notifications, CRM sync): n8n self-hosted. It works, it is free, and we control the data.
- AI agent workflows: Custom Python with Claude API. Full control over memory, tool use, and error handling.
- Client-facing automation: Depends on the client's stack. Often Zapier for non-technical teams, custom solutions for technical ones.
- Complex orchestration: Temporal for anything that needs reliability guarantees, retry logic, and long-running workflows.
We stopped trying to force AI agent logic into n8n six months ago. The time we saved by switching to direct API calls has been significant.
The Bottom Line
n8n is a good product in a market that is rapidly changing. The $180 million in funding gives them runway to evolve. The 200,000+ user community gives them distribution. The self-hosting capability gives them a moat that Zapier and Make cannot replicate.
But the AI agent narrative is overblown. The 58x revenue multiple is aggressive. And the real competition is not other automation platforms — it is developers who are discovering that building agents with code is often simpler, more reliable, and more maintainable than building them with visual tools.
n8n is not dead. But n8n is not the future of AI agents either. It is a workflow automation tool that does workflow automation well. Whether that is worth $2.5 billion is a question the market will answer over the next two years.
Need help choosing the right automation stack for your team? At CODERCOPS, we have built automation systems with n8n, Temporal, custom code, and everything in between. We can help you pick the right tool for your specific use case — not the one with the best marketing. Get in touch.
Comments