"Vibe coding" went from a half-joking tweet to how a huge share of software now gets written, all in about a year. Whether that's exciting or terrifying usually depends on whether you're the one shipping it or the one buying it. So let's skip the vibes and look at the numbers.
01 · AdoptionIt's not niche anymore
Start with the least controversial number: almost everyone is doing this. In Stack Overflow's Developer Survey, roughly three in four developers said they were using or planning to use AI tools in their workflow, with a clear majority already using them regularly.
That's the important context for everything below: AI-assisted — and increasingly fully AI-generated — code isn't a fringe experiment. It's the mainstream way software gets built now, which means a growing share of the projects that get bought and sold were vibe-coded at least in part.
02 · QualityFaster to write, harder to maintain
Here's where it gets interesting. AI tools unambiguously make code faster to produce. What the data questions is whether that code is as durable.
The most-cited analysis here is GitClear's, which looked at hundreds of millions of changed lines of code as AI assistants rolled out. It found the fingerprints of speed-over-craft: rising code churn (code rewritten or deleted soon after it's added), and more duplicated, copy-pasted code — while the share of "moved" code, the signal of thoughtful refactoring, fell.
More code is written and then quickly thrown away — a sign of iterating past the target rather than toward it.
Copy-pasted blocks climbed as assistants made it easy to generate the same thing again instead of reusing it.
Less code is reorganized and consolidated — the maintenance work that keeps a codebase healthy over time.
The honest read isn't "AI code is bad." It's that AI is very good at producing a lot of plausible code and indifferent about whether it's clean, deduplicated, and maintainable. Left unreviewed, a vibe-coded project accrues the kind of debt a buyer will find later.
03 · SecurityConfident, and sometimes wrong
The most cautionary finding is about security. In a controlled Stanford study, developers with an AI assistant wrote less secure code than those without one — and, more worryingly, were more confident that their code was safe. AI is fluent, and fluency reads as correctness even when it isn't.
In practice that shows up as the boring, dangerous stuff: missing input validation, weak auth, and — the classic — secrets committed straight into the repo. None of it is exotic. All of it is exactly what a careful buyer checks for before they wire money.
AI didn't make code quality matter less. It made it matter more — because now there's far more code, written far faster, by people far more sure it's fine.
04 · The marketSo — do vibe-coded apps sell?
Yes. A working app with real users and revenue is a real asset, and no buyer worth selling to cares whether a human or a model typed the lines. What the data above tells you is which vibe-coded apps sell — and for how much.
This is precisely why serious marketplaces put the code under a lens before a deal. On Vertos, every listing includes a free, automated read of the actual source — languages, dependencies, a readiness checklist (tests, CI, license), and the frameworks and integrations it really uses — so a buyer isn't taking "trust me" at face value, and a seller who did the work gets credit for it. The market rewards proof, and the fastest way to prove a vibe-coded build is to let the code speak.
Built it with AI?
Let the code prove it.
List your project free on Vertos — a built-in analysis shows buyers exactly what's under the hood, so a clean build earns its price. Keep 95%; flat 5% only when it sells.
List your project →Ship fast — but review before you sell.
— The Vertos team