Can You Really Make Money with Vibe Coding?

Somewhere around January 2026, the internet decided that anyone could get rich building apps with AI. The pitch was simple: describe what you want in plain English, let AI write the code, launch it, and watch the money roll in. So can you really make money with vibe coding? Collins Dictionary named it Word of the Year for 2025. Cursor hit $2 billion in annualized revenue. Lovable reached $400 million ARR at a $6.6 billion valuation. Replit tripled its valuation to $9 billion. The platforms making vibe coding possible are printing money at a pace that would make most Fortune 500 CFOs dizzy.

But the people using those platforms to build their own apps? That story is quieter, more complicated, and a lot less Instagram-friendly. The answer is yes, but with a footnote the size of a terms-of-service agreement. And the footnote matters more than the yes.

The success stories are real, but they're not what you think

Every article about vibe coding money leads with the same handful of names. Pieter Levels built a browser-based flight simulator in three hours using Cursor, and it hit $1 million in annual recurring revenue within 17 days. Sabrine Matos, a growth marketer in Brazil with zero coding background, built a women's safety app called Plinq using Lovable in 45 days. It now has over 10,000 users and $456,000 in annual recurring revenue. Drew Griffin launched SendPush.io, a wallet pass SaaS, and reportedly pulled $135,000 in revenue during its first week.

These are real numbers. Plinq's revenue is verified by Lovable's own case study and covered by No Code MBA's podcast. Levels posts his revenue publicly on X, broken down by product: PhotoAI at $105,000 per month, InteriorAI at $42,000, RemoteOK at $41,000. His combined portfolio generates roughly $300,000 monthly. The flight sim peaked at $50,000 per month shortly after launch (per 404 Media), though more recent figures from early 2026 put it at around $12,000 per month.

Here's what these stories leave out.

Pieter Levels has been building and launching internet products for over a decade. He has 849,000 followers on X. He appeared on the Lex Fridman podcast, which he has acknowledged drove a massive share of his traffic. An Indie Hackers deep dive on PhotoAI put it bluntly: "Photo AI succeeded because Pieter spent 10+ years building an audience of 600K+ followers." The flight sim went viral partly because Elon Musk endorsed it on X. Levels isn't a story about vibe coding making someone rich. He's a story about a famous internet entrepreneur using a new tool to ship faster.

Sabrine Matos solved an urgent, specific problem in a country where 37.5% of women experience some form of violence annually. She had deep growth marketing expertise, ran campaigns that generated 500 million views, and was raising a pre-seed round within months. Plinq isn't a "vibe coded app" success story. It's a startup success story where the founder happened to use Lovable instead of hiring engineers.

Strip away the survivorship bias and a pattern emerges: the people making serious money from vibe coding either had massive existing audiences, deep domain expertise in a specific market, or both. The AI was their construction tool. It was never their business model.

What "making money" actually looks like for normal people

The realistic income figures for someone without Pieter Levels' Twitter following are buried in the fine print of every "7 Ways to Get Rich with Vibe Coding" article. Here's what the actual numbers look like.

Freelance builds are the most accessible path. Solo builders are charging $300 to $800 for complete landing pages and simple web apps, according to Nucamp. Because AI handles the scaffolding, your effective hourly rate is higher than it looks on paper. But you're not selling software. You're selling your time, at rates that are respectable for freelancing but nowhere near "getting rich."

Micro-SaaS products are the next step. A first product with 10 to 20 paying users at $5 to $15 per month generates $50 to $300 monthly. Gptsters, a site dedicated to vibe coding monetization, frames this with rare candor: "Most first apps don't make money, and that's fine. The first app is about learning the craft." That's refreshingly direct advice in a space where everyone else is selling the dream.

Templates and digital products (Notion templates, prompt packs, UI component kits) sell for $15 to $49 each through platforms like Lemon Squeezy. These are real income, but they're digital product businesses that existed long before vibe coding. The AI just makes it faster to create the inventory.

The Adapty State of In-App Subscriptions 2026 report, based on 16,000 apps and over $3 billion in analyzed subscription revenue, paints an even more sobering picture for mobile apps specifically. The market is intensely polarized, with weekly subscription plans now accounting for 55.5% of all subscription revenue (up from 43% two years ago). Many of these apps rely on users forgetting to cancel. The median app isn't building a sustainable business. It's hoping for inertia.

The costs are higher than the marketing pages suggest

The pricing pages for vibe coding tools show clean, simple numbers. Cursor Pro is $20 per month. Lovable Starter is $20. Bolt Pro is $25. Windsurf Pro is $15. If you're just browsing, it looks like you can build an app for the cost of a Netflix subscription.

The real costs tell a different story. Cursor's $20 Pro plan includes $20 in model credits for premium AI models. Use those credits (and you will, if you're building anything substantial), and you're billed for overages. One developer reported $350 in Cursor overages in a single week. Anthropic says 90% of Claude Code users spend under $12 per day, which sounds reasonable until you realize that's up to $360 per month at the ceiling.

Replit's Core plan is $20 per month, but effort-based pricing means complex tasks add up. Multiple user reports put actual monthly spending for active builders at $50 to $150. Bolt's token system adds another variable: you can estimate costs, but you won't know your exact bill until the month ends.

Then there's everything else. Supabase (the backend most AI builders use) costs $25 per month once you exceed the free tier. A custom domain runs $10 to $15 per year. Third-party APIs, email services, payment processing: these all add up. A production app with real users costs a solo founder $50 to $100 per month minimum, and often more.

That micro-SaaS making $150 per month from 15 users at $10 each? After tool subscriptions, backend costs, and a domain, you're netting maybe $50 to $75 per month. It's profit, technically. But nobody's quitting their day job for it.

90% of AI-built projects never reach production

That statistic comes from enterprise software analysis, but the dynamic applies to solo builders too. The gap between "it works on my screen" and "it works for real users" is where most vibe-coded projects die.

A penetration testing firm audited 15 vibe-coded applications and found 69 vulnerabilities. Six were critical, meaning an attacker could read the database, hijack sessions, or escalate to root on shipped apps handling real user data. Veracode's analysis of four million code scans found that AI-generated code contained security flaws 45% of the time. The Cloud Security Alliance put the number even higher, at 62%. AI code generators produce vulnerabilities at roughly twice the rate of human-written code.

These aren't theoretical risks. Moltbook, a vibe-coded application, exposed 1.5 million API tokens because row-level security was never enabled on the database. A Lovable-specific vulnerability (CVE-2025-48757) inverted access control logic across 170 production applications, meaning unauthenticated users could access data that was supposed to be private. Base44 had a platform-wide authentication bypass. Replit's AI agent wiped a production database during an explicit code freeze.

For the person building a side project, these security issues might not matter. For the person trying to build a business that handles real user data, payment information, or anything that needs to be reliable, the code that AI generates is, on average, the starting point of the work, not the end.

Apple has started blocking App Store updates for apps with the kind of code quality issues that AI-generated codebases frequently produce. Code churn (the percentage of code that gets rewritten shortly after it's written) is up 41% in AI-generated codebases. Code duplication has spiked 4x. The "vibe coding hangover," as Fast Company called it, is arriving on schedule.

The METR study nobody wants to talk about

In July 2025, METR (Model Evaluation and Threat Research) published results from a randomized controlled trial, the gold standard of research methodology. Sixteen experienced open-source developers completed 246 real tasks on repositories they personally maintained. Codebases averaged over one million lines of code. Tasks were randomly assigned to either allow or forbid AI tool usage.

The developers predicted AI would make them 24% faster. After the study, they estimated AI had made them 20% faster.

The actual result: AI tools made them 19% slower.

Read that again. Experienced developers, working on codebases they knew intimately, using their AI tool of choice (primarily Cursor Pro with Claude 3.5/3.7 Sonnet), were measurably slower with AI assistance. And they had no idea. The gap between perceived speedup (20%) and actual slowdown (19%) is a 39-point perception error.

This finding doesn't mean AI coding tools are useless. METR itself noted this was a snapshot of early-2025 capabilities. The tools have improved since then. And the study specifically measured experienced developers on mature codebases, not beginners building from scratch, which is a very different scenario. Vibe coding may well be faster for greenfield projects and boilerplate-heavy work.

But it destroys the premise that AI automatically makes everything faster. For complex work on existing code, the tools can actively slow you down while making you feel more productive. That's the most dangerous kind of performance hit: the one you can't feel.

The money flows to the people selling shovels

Follow the money, and the picture becomes clear. The people getting rich from vibe coding are overwhelmingly the people selling vibe coding tools.

Cursor's parent company Anysphere raised $2.3 billion at a $29.3 billion valuation in November 2025, then hit $2 billion in annualized revenue by March 2026. Lovable went from launch to $400 million ARR in under a year. Replit tripled its valuation to $9 billion. Emergent, an Indian vibe coding platform, claims $100 million ARR just eight months after launch. These are among the fastest-growing startups in history.

The people making real money using the tools fall into three categories, and they're not equally sized.

Category one: famous entrepreneurs using a new tool. Pieter Levels, with his decade of audience-building and 849K followers, would have been successful launching products with any technology. Vibe coding made him faster, not rich. He was already rich.

Category two: founders solving specific, urgent problems. Sabrine Matos with Plinq. The SendPush.io team. These are real businesses built by people with domain expertise who happened to use AI tools. They would have needed to build the same product regardless. The tools just lowered the engineering barrier.

Category three: everyone else. This is the largest group by far, and the realistic outcome here is $0 to $300 per month from a side project, with a significant chance of $0. Most vibe-coded apps have zero users, zero revenue, and a security posture that would make a penetration tester wince.

This is the classic gold rush dynamic. During the California Gold Rush, the people who got reliably rich were the ones selling pickaxes, shovels, and Levi's jeans. A few miners struck gold. Most went broke. In 2026, Cursor, Lovable, and Replit are selling the shovels. A few builders are striking gold. Most are building apps that nobody downloads.

The honest path to making money with vibe coding

None of this means vibe coding is worthless. It means the hype is wrong about what it's good for. Here's what actually works.

Build tools for a market you already understand. Every successful vibe-coded business started with a real problem and a person who understood the market. Sabrine Matos understood women's safety in Brazil. She didn't randomly decide to build a background check app because AI made it possible. If you're an accountant, build a tool for accountants. If you manage rental properties, build something for landlords. The AI handles the code. You supply the expertise that makes the product worth paying for. If you're not sure whether your idea has legs, validate it before you bet on it.

Sell services, not software, when you're starting out. Charging $500 to build a landing page for a local business is real money you can collect this month. Trying to build the next SaaS unicorn with vibe coding is buying a lottery ticket with extra steps. The freelance path is less exciting and far more reliable.

Budget $100 to $200 per month for tools and infrastructure. The $20 per month pricing on marketing pages is the cover charge, not the total tab. If your business plan requires tool costs to stay under $30 per month to be profitable, the business plan needs more work.

Plan for the security problem. If your app handles user data, payment information, or anything sensitive, factor in either learning basic security principles yourself or paying $500 to $2,000 for a security audit before launch. Beesoul offers an 18-check framework. VibeAudits does human-powered reviews. The cost of a security audit is trivially small compared to the cost of exposing your users' data.

Set your expectations at $50 to $300 per month for a first product. If you hit that, you've done better than the vast majority of vibe coders. If you hit $1,000 per month, you're in rare territory. If you hit $10,000 per month, you should probably be writing a case study, not reading this article.

The bottom line

Can you really make money with vibe coding? Yes. Can you get rich? Almost certainly not, unless you were already going to get rich anyway and AI just made the path shorter.

The $4.7 billion vibe coding market is real. Ninety-two percent of US developers use some form of AI coding tool daily. The tools are impressive, and they're getting better fast. But the gap between "AI can build an app" and "AI can build a business" is enormous, and it's filled with security vulnerabilities, $150 monthly tool bills, and apps that nobody downloads.

The smartest approach in 2026 is also the least exciting one: use vibe coding to build something small that solves a specific problem for a specific group of people you already understand. Charge for it. See if anyone pays. If they do, build more. If they don't, you spent $200 and a weekend, not two years and your savings.

That's not the pitch the vibe coding platforms are selling. But it's the one that actually works.