Vibecoding is doubling app releases year over year. Downloads grew 2%. Apple is now paying to review a marketplace its revenue engine can’t cash in.
App submissions to Apple’s App Store roughly doubled to 560,000 in the first half of 2026 as AI “vibecoding” tools let non-programmers ship apps. But downloads rose just 2%, and most of the new apps run ads instead of the in-app purchases Apple taxes. More supply, flat demand, thinner economics.
Marco Pérez had never built a mobile app. Over five winter weeks he described one to an AI tool and let it write the code. His scrapbooking app, Stampa, took about a week and has pulled 20,000 downloads and roughly $3,000 in profit. Stories like his are why Apple’s App Store is filling up faster than it has in a decade. They are also why that flood might not be worth much to Apple.
What Happened
The App Store is in the middle of a supply shock. New app releases grew 30% to about 600,000 in 2025, and in the first half of 2026 they doubled again to roughly 560,000, according to estimates from the analytics firm Sensor Tower. On pace, 2026 will add more than a million new apps, a volume the store has never seen.
The driver is vibecoding: AI tools that turn a plain-English description into a working app, no coding required. The barrier that kept most people out of app development has effectively dropped to zero.
Demand did not get the memo. App Store downloads grew just 3% in 2025 to 35.4 billion, then 2% in the first half of 2026 to 17.6 billion. Supply is roaring. Demand is flat. And every one of those submissions still has to clear Apple’s manual review, where developers have started complaining about month-long waits.
The Backstory
This has happened before, at a smaller scale. When Apple opened the App Store in 2008, developers rushed in. Flappy Bird, the deceptively simple 2013 game, was reportedly earning $50,000 a day at its peak. New releases climbed until 2016, when they hit about 890,000, then stalled. By 2022 the store bottomed out at roughly 420,000 new apps a year as the market matured and consumers settled into a handful of apps they actually opened.
Vibecoding broke that plateau. As AI got good at writing code over the past year, a second gold rush began, and the release curve turned sharply back up.
The Plan
Apple is publicly delighted. A spokesman said the company was “thrilled to see a new generation of developers embracing the latest tools,” noted that 90% of submissions are reviewed within 48 hours, and pointed out that downloads are only one measure of the store’s value.
The optimistic math is simple. Apple takes 30% of in-app purchases, or 15% from smaller developers. More apps should mean more purchases, and more purchases mean more revenue. Developers are betting on the same curve: Pérez is planning to leave a job at JPMorgan Chase to build apps full time, chasing the stories he hears of builders whose sixth app finally blows up.
The Business Model Angle
Here is the part the boom narrative skips. Apple’s App Store is a funnel, and Apple only monetizes the bottom of it: apps lead to downloads, downloads lead to in-app purchases, and Apple takes its cut of those purchases. Creation sits at the top of the funnel, and Apple has never charged for creation. It charges for transactions.
Vibecoding inflates the top of the funnel at almost no cost and does nothing for the bottom. The download data proves it. Consumers already consolidated around the apps they use years ago, and a million new options does not change how many apps a person opens in a day. You cannot tax a transaction that never happens.
It gets worse for Apple’s cut. Vibecoded micro-apps tend to be simple, single-purpose tools, and simple tools monetize with ads, not the in-app purchases that are fiddly to set up. Ad revenue is revenue Apple does not touch. As mobile analyst Eric Seufert put it, “Apple doesn’t benefit from that.” So the surge routes around the exact mechanism Apple built the store to capture.
Meanwhile the cost side is very real. Every submission consumes Apple’s scarce manual-review capacity, the human bottleneck it has always used to justify its curation moat. Costs rise, taxable revenue does not. For context on the stakes, Apple’s Services segment hit a record $109.16 billion in fiscal 2025, and the App Store is a load-bearing pillar of it. In-app purchases generated an estimated $114.02 billion across the store last year, of which Apple collected $34.21 billion, per the analytics firm Appfigures. That is the engine. Vibecoding is not feeding it.
The deeper lesson is one every marketplace eventually learns. The scarce resource was never the ability to build an app. It was consumer attention. AI collapsed build cost to zero, which does not create attention, it just crowds the fight for it. Value migrates to whoever controls discovery, not whoever owns the shelf. Apple owns the shelf.
The Risk
The counterargument, from someone who ran the store, is that this may not hurt Apple much. Phillip Shoemaker, who led the App Store from 2009 to 2016, notes it is “not like a typical storefront, where you have shelf space.” Apps nobody uses do not physically crowd out the ones people want, so junk is not automatically a problem.
The risks he does flag are the ones that bite the model indirectly. More junk degrades discovery, which makes the few apps that actually monetize harder to find, quietly shrinking the taxable base. And a flood of low-review-scrutiny apps makes it easier for a bad actor to slip in inappropriate or unsafe content, which is a direct threat to the quality and trust Apple sells as the reason its store is worth a 30% toll. Longer review times chip at that promise too. There is also a regulatory shadow: Europe has already pried open the App Store, and a store stuffed with junk plus emerging alternative storefronts weakens Apple’s position to keep charging the toll at all.
Quick Questions
What is vibecoding? Using AI tools to build software by describing what you want in plain language, without writing or knowing how to write code. The AI generates the app.
Does Apple actually make money from the app surge? Barely, so far. Apple earns its cut from in-app purchases, but downloads are flat and many vibecoded apps run ads rather than purchases, which Apple does not tax.
How many new apps are being added? Roughly 560,000 in the first half of 2026, about double the prior-year pace and on track to exceed one million for the year, per Sensor Tower.
Are people downloading them? Not really. Downloads grew just 2% in the first half of 2026, to 17.6 billion, even as the number of apps exploded.
Which AI tools do the builders use? Popular options include OpenAI’s ChatGPT and Anthropic’s Claude, along with dedicated code editors built for the workflow.
The Business Model Analyst Take
Frame this as “boon versus clutter” and you miss the actual mechanics. Apple’s model taxes transactions, not creation. AI made creation free and did nothing for transactions. So this is a cost event wearing the costume of a growth event.
The number that matters is not 560,000 new apps. It is 2% download growth. Supply doubled straight into a demand wall, and the marginal app is an ad-supported tool Apple cannot bill. The $99 annual developer fee scales nicely with the crowd of aspiring builders, but that is rounding error against a Services business measured in the hundreds of billions.
If Apple wants to convert this flood into money, it has to start monetizing the top of the funnel it now involuntarily owns: priority review as a paid tier, AI-submission tooling, or a real position in the on-device ad stack these apps actually use. Until it does, vibecoding is a gift to developers and a bill Apple keeps paying without a way to collect. The store has never looked busier. It has rarely mattered less to the bottom line per app.
Based on reporting by Kalley Huang for The New York Times, with market data from Sensor Tower and Appfigures.
