Vibe Coding Revenue Race: Cursor at $4B, Lovable Past $600M — the 2026 Numbers

Introduction

Eighteen months ago, "vibe coding" did not exist as a phrase. Andrej Karpathy coined it in February 2025 to describe coding where you "fully give in to the vibes" and let an AI write the code. Collins Dictionary named it Word of the Year for 2025. In that same stretch, the tools that sell this way of building went from side projects to some of the fastest revenue ramps in software history.

The numbers below are the whole leaderboard as it stands in November 2026: Cursor past $4 billion in annualized revenue, Claude Code past $2.5 billion, Cognition at $900 million, Lovable past $600 million. Every figure carries its date and its source type — company-disclosed, press-reported, or third-party estimate — because in this market the date matters as much as the number. A run-rate from June is already five months old, and in this race five months is a lifetime.

There is also a twist at the top of the table that arrived yesterday. OpenAI told SpaceX in August that it intends to wind down the contract supplying its models to Cursor, with a proposed shutoff date of November 12, 2026. On November 12 it said it had temporarily extended access beyond that date. The biggest revenue machine in AI coding is, for now, running on borrowed model supply. Keep that in mind while reading the numbers: they measure demand, not durability.

The leaderboard

The table ranks the field by revenue run-rate — one month's revenue multiplied by twelve. That is not the same as annual revenue, and the comparison is not perfectly clean: the figures were disclosed on different dates, some are estimates, and some companies do not disclose at all. What the table does show is relative commercial scale.

Add up the standalone companies in the table and their latest valuations or deal prices come to about $136.8 billion, according to Value Add VC's market tracker, checked against sources in October 2026. The revenue behind that valuation is concentrated at the top: Cursor alone reports more than the next three combined.

Two details in the table deserve a second look. GitHub Copilot, the incumbent, does not disclose revenue at all — but Microsoft said on its July 2026 earnings call that it had reached 50 million users. And Bolt.new, the breakout story of late 2024, has published no updated revenue figure in 2026. Its $40 million ARR is real — it went from $4 million in four weeks to $20 million in three months to $40 million in five months after launching on October 3, 2024 — but it is a 2025 number, and in this market a 2025 number is a history lesson, not a ranking.

Cursor: the fastest ramp, and the strangest week

Cursor's curve is the one every other company is measured against. It crossed $100 million in annual recurring revenue in January 2025, passed $500 million by June 2025, $1 billion in November 2025, $2 billion by March 2026, $3 billion in May, and a reported $4 billion in annualized revenue in early June 2026. Fortune called it, by some measures, the fastest-scaling B2B software company ever. It did this in about four years from founding.

The demand side is as striking as the revenue. Fortune reported in June 2026 that 67% of the Fortune 500 use Cursor's tools. Then, in June 2026, SpaceX agreed to buy Cursor's parent company, Anysphere, in an all-stock deal valued at $60 billion. The deal closed on August 14, 2026, making Cursor the largest AI-coding deal to date — and, two weeks later, the reason OpenAI moved to cut off its models.

OpenAI's stated reason was not technical and not commercial. In its August 28 post, it wrote: "We are making this choice because we cannot be confident that SpaceX will use our technology within our terms of service, based on our experience with Elon Musk's companies violating contracts." Cursor's chief executive, Michael Truell, said publicly that OpenAI models serve about 5% of Cursor user traffic. The practical damage is therefore small — Claude, Gemini, Grok, and Cursor's own Composer models are unaffected — but the precedent is large. The company with the biggest revenue in this market can lose a supplier because of who owns it. That is a sentence worth remembering before building a business on any single tool in this table.

Lovable: $2.7 million of revenue per employee

If Cursor is the biggest number, Lovable is the most instructive one for a solo founder. The Swedish app builder launched publicly in November 2024. Its reported run-rate went from roughly $17 million in February 2025 to $100 million in July 2025, $200 million by November, more than $300 million in January 2026, and $400 million in February 2026 — adding $100 million of ARR in a single month. It told TechCrunch it passed $500 million on June 9, 2026, and co-founder Fabian Hedin said in September that it had crossed $600 million.

The headcount behind that curve is the statistic solo operators quote most, and the one they most often misread. Lovable had 146 full-time employees when it crossed $400 million in February 2026, according to Business Insider — roughly $2.7 million of ARR per employee, among the highest ratios ever reported in software. The company has said it plans to grow to about 350 employees by the end of the year. The valuation tracked the revenue: $1.8 billion in July 2025, $6.6 billion in December 2025, and $13.3 billion in August 2026, when it raised a $400 million Series C.

Two cautions belong next to those numbers. First, the 146 people were not running a $500 million business alone: the company had raised more than $530 million in venture funding behind them, paying for acquisition, enterprise sales, infrastructure, and support. Second, none of these figures is audited. Lovable is private and does not publish financials; the numbers are company-disclosed and reported by the press, not independently verified. The growth rate itself is not in dispute. The exact ratio is.

What is not in dispute is the scale of usage: Lovable says more than 60 million projects have been created on the platform since launch, and apps built on it receive more than 900 million monthly visits. Its co-founder says two-thirds of the Fortune 500 use the product. The company that sells "anyone can build software" is, increasingly, selling to the same enterprises as Cursor.

Two markets, one word

The leaderboard hides a split that matters more than the ranking. Cursor, Cognition, Factory, and Claude Code sell to professional developers. Lovable, Replit, Emergent, and Vercel's v0 sell app building to people who may never read the code. That second group is what most people mean by "vibe coding" — and it is the smaller revenue pool today, but the larger audience.

The demand data makes the split visible. In Stack Overflow's 2025 Developer Survey, 84% of developers said they use or plan to use AI tools in their development process, up from 76% a year earlier, and 51% of professional developers use AI tools daily. Yet 72% said vibe coding is not part of their professional work. Developers use AI coding tools every day; most do not yet trust prompt-only building for professional work. The gap between those two numbers is the market the app builders are selling into: prototypes, internal tools, and MVPs built by people who are not professional developers.

Search demand tells the same story from the other side. The keyword "vibe coding" draws about 110,000 monthly searches in the US, with low competition and a $12.73 cost per click, according to SignalDig data updated October 15, 2026. Search volume peaked at about 165,000 in March 2026 and has settled near 90,000 a month since — a real audience, past its hype peak, still searching. Related queries are practical, not curious: "vibe coding ai", "vibe coding tool", "is vibe coding bad", "vibe coding cursor". People are not asking what it is anymore. They are asking which tool to use and whether to trust it.

What the numbers do not say

A leaderboard invites false precision, so here is what is not settled. Run-rate revenue is one month multiplied by twelve; it is not annual revenue, and a single good month can flatter it. Company-disclosed figures are not audited, and the fastest-growing companies have the most reason to disclose. Replit's $525 million is a third-party estimate, not a company figure, and its $1 billion target for the end of 2026 is a hope, stated to Forbes, not a result. Vercel's estimated $600 million covers hosting and cloud as well as v0, so it is not a like-for-like comparison with Lovable. Bolt.new's current standing is genuinely unknown rather than small. And valuations are not revenue: Cognition went from a $26 billion valuation in May 2026 to $48 billion in September, while its run-rate went from $492 million to $900 million — the valuation nearly doubled while revenue did less than that, because investors are pricing the next two years, not the last two quarters.

The honest summary is narrower than the table looks: a handful of AI coding companies are generating real, large, fast-growing revenue; the biggest one just changed owners and nearly lost a model supplier in the same quarter; and the second market — non-developers building apps — is real but still smaller than the professional-developer market. Everything else is a forecast wearing a number's clothes.

Lessons: three mechanisms for a solo creator

The leaderboard matters less than what a one-person business does with it. Three mechanisms are worth setting up now.

1. Treat the leaderboard as a vendor-risk map, not a shopping list: The Cursor-OpenAI episode is the mechanism made visible: a tool you depend on can lose a model because of who bought its parent company, and the notice period was about ten weeks. The executable version is to build on two tools, not one — keep your code in your own git repository, export every project out of the builder, and make sure the app runs without the tool that generated it. The guardrail is a monthly test: can you open your project, change one feature, and deploy it without opening the app builder at all? If the answer is no, the tool owns your product, not you. The payoff is boring and decisive: when a supplier changes terms, you switch tools in an afternoon instead of rebuilding a business.

2. Copy the revenue-per-employee math, not the headcount: Lovable's $2.7 million of ARR per employee did not come from working harder. It came from a $20-a-month subscription sold to millions of people, with no custom work per customer. The executable version for a solo builder is to price one product, one way, and refuse the custom-work trap: a single subscription price, a self-serve onboarding flow, and a written rule that no customer gets a feature built just for them. The guardrail is the calendar: if more than a day a week goes to individual customers, the model is consulting with a subscription label on it. The payoff is the only leverage a solo operator has — revenue that grows while headcount stays at one.

3. Sell to the 72%, not the 28%: Most professional developers say vibe coding is not part of their professional work, yet 51% use AI tools daily and search demand for "vibe coding" runs near 90,000 searches a month. The buyers are not developers shopping for a better editor. They are business owners who want an internal tool and do not care what it is built with. The executable version is to pick one narrow business job — a quote generator for a trades business, a booking page for a tutor, a stock-check tool for a small shop — build it with a vibe-coding tool in a week, and charge monthly for it. The guardrail is scope: one job, one customer type, one price. The moment the tool does three jobs for three kinds of customer, it is a product nobody can describe. The payoff is that the customer never asks which AI wrote the code. They ask whether the job gets done.

Closing

A year ago, "vibe coding" was a joke in a tweet. Today the companies selling it report billions in run-rate revenue, carry a combined $136.8 billion in valuations, and count two-thirds of the Fortune 500 as users — while the biggest of them nearly lost a model supplier over an acquisition, and the most efficient of them still leans on half a billion dollars of venture funding. The numbers are real, the growth is real, and so is the fragility underneath them. For a solo creator, the lesson is not to envy the leaderboard. It is to read it the way a builder reads a weather report: use the tailwind, build for the storm, and own the code either way.