Apple’s App Store added roughly 560,000 new apps in the first half of 2026 alone — within striking distance of the ~600,000 apps released across all of 2025, itself a 30% jump on the year before. The driver is no mystery: AI “vibe coding” tools that turn a natural-language description into a working app have collapsed the cost of shipping something to the store from weeks of engineering to an afternoon of prompting.
Downloads didn’t follow. App Store downloads grew just 2% in the first half of 2026, down from 3% growth across 2025. Supply is compounding while demand is essentially flat — and the apps absorbing that gap are, by most accounts, not thriving. Coverage of the trend points to a rising share of copycat apps, near-identical clones, and listings that simply don’t work as described, alongside the quality-control pushback already showing up elsewhere this year: Apple’s slower AI-app review process and Google Play’s own quality crackdown are both direct responses to the same flood.
Why this is the real vibe-coding story, not the hype version
The pitch behind every AI app builder — Lovable, Bolt, v0, and the rest — has been speed: anyone can now build an app. That part is true and no longer interesting on its own; the App Store numbers prove it. What the download data shows is the part that speed doesn’t solve: getting found, getting downloaded, and getting used. A working prototype generated in an afternoon and a product that earns a place on someone’s phone are different problems, and the gap between them is exactly where most vibe-coded apps are currently failing.
So what
If you’re weighing “get an MVP out fast with an AI builder” against “invest in a properly built app,” the App Store data answers a narrower question than it looks like it does — it tells you shipping is cheap now, not that shipping is enough. A vibe-coded prototype is a genuinely good way to validate an idea fast. But turning that into something that actually finds an audience — decent onboarding, real performance, a store listing and positioning that doesn’t read as one of 560,000 near-identical entries — is still a product and engineering problem, not a prompting problem. That’s the gap between a weekend build and a launch we help clients close in custom software development work — get in touch if your AI-built prototype needs to become something people actually keep.