Key Intelligence Insight
n8n quadrupled its revenue in eight months. The prior six years produced the same amount.
That compression is not a product story. It is a market-timing story with a structural thesis underneath it: when enterprises needed to deploy AI agents into production, they discovered they needed an orchestration layer first -- and n8n had spent six years quietly becoming that layer. "When this whole AI wave started, we honestly were worried a little bit," Oberhauser told Sequoia Capital partners in 2025. "We knew it's probably going to mean one of two things. Either it's going to be a huge opportunity or actually the demise of the company." It became the former. The mechanism: while most competitors added an OpenAI API node and called it AI integration, n8n rebuilt its architecture to let users construct full agentic systems -- memory, vector stores, tool-calling, output parsers -- without writing a line of Python.
The numbers frame the scale of the inflection. Annual recurring revenue reached $40 million in July 2025, up from $7.2 million in 2024 -- a 5.5x increase in twelve months. At roughly 67 employees at the time of that milestone, that represents approximately $600,000 in ARR per employee, a capital efficiency profile almost never seen at this growth rate. Headcount has since expanded aggressively, reaching 365 employees by early 2026, a 718% year-over-year increase that signals a company in active deployment of capital rather than cautious accumulation of it.
The funding sequence confirms the market's read. A €55 million Series B in March 2025 valued n8n at $270 million. Seven months later, a $180 million Series C led by Accel -- with participation from NVIDIA's NVentures, Meritech, Redpoint, T.Capital (Deutsche Telekom), Visionaries Club, and Evantic, plus follow-on investments from Sequoia, Felicis, Highland Europe, and HV Capital -- valued the company at $2.5 billion. A 9x valuation step-up in seven months is not a negotiation outcome. It is a signal that multiple top-tier funds concluded they were looking at a category-defining business.
The thesis at the centre of that bet: n8n is not a workflow tool that added AI features. It is an AI orchestration layer that will absorb workflow automation as a subset of its surface area.
Founding Story
Jan Oberhauser did not come from enterprise software. He came from visual effects -- building post-production pipelines for films including Maleficent and Happy Feet 2. The automation instinct came first; the software company came second.
He built n8n while holding two jobs. During the day, he worked at a startup. On evenings and weekends, with a wife and children at home, he wrote the first version of the product. The constraint was not capital. It was time. That constraint shaped the architecture of what he built: a tool designed so that the person who understood the problem could solve it themselves, without requiring an engineer as an intermediary. "The people that have the problems are the best equipped to solve them themselves," Oberhauser said in a 2025 interview with Sequoia Capital. "And I think that's the world where we're going."
n8n launched in April 2019 on Product Hunt. Within the first week, contributors appeared. One of the earliest -- Ricardo, a developer in Florida who discovered the product independently -- went on to build 50 to 60 integrations and eventually became one of n8n's first employees. The community preceded the company.
The funding origin story is instructive about European venture capital. When early traction became clear, 80% of inbound investor interest came from US funds, 20% from UK, and 0% from Germany. German investors engaged only after Sequoia submitted a term sheet. Firstminute Capital and Sequoia led the seed round. Felicis joined the Series A. The pattern held: US and UK funds moved in days; European funds required external validation before acting.
Oberhauser has been explicit about his exit thesis from the start. The goal is an IPO, not an acquisition. n8n is structured as a German GmbH, with US and UK entities as subsidiaries. The legal architecture matches the ambition: a European technology company built to stand independently, not to be absorbed by one of the platforms it competes with or connects to.
Product
n8n sits at the intersection of three things that have historically required separate tools and separate teams: deterministic code logic, AI agent execution, and human-in-the-loop oversight.
The core interface is a visual workflow builder where discrete operations -- called nodes -- connect inputs to outputs across any application or API. That description covers the same basic functionality as Zapier or Make. What separates n8n is the depth beneath that surface: users can drop into raw JavaScript or Python at any node, construct multi-step AI agents with memory and vector store integrations, set conditional branching logic of arbitrary complexity, and deploy the entire workflow on their own infrastructure.
The AI layer is not a feature added to a workflow tool. It is a first-class architectural component. n8n supports any large language model -- OpenAI, Gemini, Anthropic, Ollama for local deployment, and models the company has no relationship with -- via standards rather than proprietary integrations. The model-agnostic stance is structural: n8n positions itself as the orchestration layer between AI capabilities and business systems, not as a distribution channel for any single AI provider. The company describes MCP -- Anthropic's Model Context Protocol -- as "the HTTP of AI workflows," and n8n as the orchestration layer across diverse MCP agents and tools.
The licensing model is deliberately non-standard. n8n calls it "faircode": source code is fully available, free to use internally at any scale, from individual developers to government ministries. What users cannot do is commercialise the code -- resell it, white-label it, or build a competing hosted product on it. Oberhauser designed it this way from day one, and has been direct about the reasoning: "I'm not building n8n and giving it away for free because I'm a good person. I actually want to build a business around it. I want to make sure I can get paid. I want to make sure all the other people can get paid as well -- because I actually think it's in the interest of everybody." This eliminates the risk that AWS, Microsoft, or any large cloud provider extracts the value n8n creates. Every dollar that would otherwise flow to infrastructure aggregators stays within n8n's economic ecosystem.
Self-hosting is a genuine product differentiator, not a checkbox. Workflows are stored as portable JSON files, transferable between environments without proprietary lock-in. For cybersecurity customers -- a segment n8n has penetrated specifically -- this matters at the infrastructure level: air-gapped deployments, local model execution via Ollama, and complete data residency are achievable within the same product used by SMB customers on n8n's managed cloud. The product does not compromise to serve both segments. It serves both by design.
The community integration library has crossed 1,000 nodes built outside n8n's core team. The company has begun enabling those community integrations on its cloud product -- a flywheel mechanic that compounds the surface area of the platform without scaling n8n's own integration maintenance burden proportionally.
Market, Competition & Business Performance
Market
The workflow automation market has always been large. The AI agent deployment market is larger, and it has no incumbent.
n8n operates at the intersection of both. The entry vector was workflow automation -- a market historically dominated by Zapier and Make, where the buyers were operations teams and the use cases were relatively shallow integrations between SaaS tools. That market is being displaced from below by AI agents capable of replacing entire manual process chains and from above by enterprises demanding production-grade deployment infrastructure for those agents.
n8n's position is that both displacement vectors run through it. Simpler automation needs are being captured by AI-native no-code tools. Complex, production-grade AI deployments require an orchestration layer with code-level flexibility, human-in-the-loop controls, and the ability to connect any model to any system. The SMB segment is adopting fast; the enterprise segment is moving more slowly but represents the majority of available revenue. The Middle East, according to Oberhauser, is moving faster than Europe on enterprise AI deployment.
The strategic logic of the current growth phase is explicit: capture the smallest builders first, because they are the path into the largest organisations. n8n is already inside enterprises not through top-down sales but through individual practitioners who self-hosted the product, built workflows, and then introduced it to their organisations' infrastructure teams. Zapier built the reverse: it sold to buyers who then handed it to practitioners. n8n's motion runs upward from practitioners to procurement, not downward from procurement to practitioners.
That population is growing, not shrinking.
Competition
Zapier and Make represent the prior generation of the market. They price per operation, restrict self-hosting, and built their AI integrations as standard API nodes -- HTTP request wrappers to OpenAI, not native agentic frameworks. For straightforward, high-volume SaaS integrations, they remain viable. For anything requiring AI agents, custom code, data privacy, or production-grade reliability, their architecture creates friction n8n does not.
The switch pattern is documented. Practitioners arrive at Zapier or Make, build until they hit structural limitations -- pricing at scale, inability to self-host, complexity of AI integration -- and migrate to n8n. The migration is facilitated by n8n's JSON-portable workflow format: copying a workflow is a file transfer. The switching cost structure runs asymmetrically against Zapier and Make; migrating away from n8n requires building the equivalent infrastructure elsewhere.
The more serious competitive question is vertical AI agent platforms. A growing number of companies are building purpose-built AI agents for specific functions: sales, customer support, legal review. Each solves one problem precisely. The mechanism by which horizontal platforms like n8n absorb this pressure is the same mechanism that made SaaS automation necessary in the first place: as vertical tools proliferate, the complexity of connecting them compounds, and the orchestration layer becomes more valuable, not less. n8n has been explicit about this dynamic.
OpenAI is a potential long-term disintermediation risk. If model providers build production-grade orchestration natively into their platforms, the case for a separate orchestration layer weakens. n8n's defence against this is the model-agnostic architecture: no enterprise building critical workflows on n8n is locked to any model provider, which means model provider consolidation does not threaten the workflows themselves. The product is designed to survive any single model losing the AI race.
Business Model
n8n operates a three-tier model: a free self-hosted version, a managed cloud product for SMBs, and an enterprise offering for large organisations requiring dedicated infrastructure, compliance controls, and SLAs.
The free tier is not a loss leader in the conventional sense. It is the distribution mechanism. Every self-hosted deployment is a potential enterprise account, reached without a sales motion. The community of 230,000+ active users generates integrations, tutorials, templates, and YouTube content that n8n's marketing budget did not pay for. Revenue-generating customers are discovered by the platform; the platform is extended by users the company has never contracted with.
The enterprise tier is the long-term revenue concentration point. Oberhauser has observed the pattern across open-source companies: community builds surface area, enterprise captures margin. n8n has government ministries, Vodafone, SoftBank, and Seat as enterprise references -- the kind of names that compress procurement cycles for the next enterprise sale. The company does not disclose enterprise ARR as a proportion of total, but the direction of investment is clear: the headcount expansion to 365 employees includes heavy hiring in enterprise sales, customer success, and go-to-market roles.
The economic architecture of faircode reinforces the business model. n8n captures commercial value from organisations that want to build on top of its code -- a category that would otherwise become free infrastructure for cloud providers to monetise. The constraint targets the 0.1% of potential users it is designed to target; for the remaining 99.9%, the experience is functionally identical to open source.
Traction
The headline metric -- $40 million ARR in July 2025, against $7.2 million twelve months prior -- does not fully capture the velocity.
n8n's AI feature adoption rate stands at 75% of active customers, in a user base that crossed 230,000 active users with 70,000+ GitHub stars. Those are not vanity metrics in the developer tooling context: GitHub stars are a leading indicator of community density, and community density in this market precedes enterprise penetration by 18 to 36 months.
The product's real-world impact is documented at scale. A major online travel agency automated 70% of customer support queries using an n8n-built AI agent, with customers reporting 15% higher satisfaction rates than with human responses -- not despite the automation, but because of it. A G7 government ministry deployed n8n as its SOAR platform for threat detection and automated response. An early n8n workflow built at ThreatConnect, a cybersecurity firm, saved 50 man-hours per customer success manager per month across a 15-20 person team.
The company's internal operation is itself a proof of concept. n8n runs hundreds of internal workflows across sales, marketing, support, and operations -- automation at the rate of $600K ARR per employee, before the current hiring cycle, is partly a function of the company eating its own product at scale.
The momentum created a hiring problem that is itself a traction signal. Revenue goals ran several months ahead of plan before the Series C closed. The bottleneck was not capital or customers: it was the rate at which a company growing this fast could interview, evaluate, and onboard talent without lowering the hiring bar. That is not a complaint. That is a definition of product-market fit.
The question now is whether n8n compounds this position into the enterprise stack before a better-capitalised competitor replicates the model. Oberhauser's own framing of the destination is unambiguous: "If people in a few years think about 'I have to build AI, I have to do anything with AI' -- we are starting to be the default building tool already." With $240 million in total funding, a $2.5 billion valuation, NVIDIA's endorsement, and a community flywheel that took six years to build, the answer requires more than capital to dispute.